diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,31048 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.995119305856833,
+  "eval_steps": 500,
+  "global_step": 4422,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0013557483731019523,
+      "grad_norm": 0.8601107597351074,
+      "learning_rate": 0.0,
+      "loss": 0.4122,
+      "step": 1
+    },
+    {
+      "epoch": 0.0027114967462039045,
+      "grad_norm": 0.8024901747703552,
+      "learning_rate": 1.8208790046993825e-05,
+      "loss": 0.3969,
+      "step": 2
+    },
+    {
+      "epoch": 0.004067245119305857,
+      "grad_norm": 0.8655965328216553,
+      "learning_rate": 2.8860249407989837e-05,
+      "loss": 0.388,
+      "step": 3
+    },
+    {
+      "epoch": 0.005422993492407809,
+      "grad_norm": 0.5744454860687256,
+      "learning_rate": 3.641758009398765e-05,
+      "loss": 0.3964,
+      "step": 4
+    },
+    {
+      "epoch": 0.0067787418655097615,
+      "grad_norm": 0.329508900642395,
+      "learning_rate": 4.227950118402034e-05,
+      "loss": 0.3463,
+      "step": 5
+    },
+    {
+      "epoch": 0.008134490238611713,
+      "grad_norm": 0.3552570641040802,
+      "learning_rate": 4.706903945498366e-05,
+      "loss": 0.3536,
+      "step": 6
+    },
+    {
+      "epoch": 0.009490238611713666,
+      "grad_norm": 0.25449544191360474,
+      "learning_rate": 5.111853636314163e-05,
+      "loss": 0.3467,
+      "step": 7
+    },
+    {
+      "epoch": 0.010845986984815618,
+      "grad_norm": 0.18209263682365417,
+      "learning_rate": 5.462637014098147e-05,
+      "loss": 0.3327,
+      "step": 8
+    },
+    {
+      "epoch": 0.01220173535791757,
+      "grad_norm": 0.16959260404109955,
+      "learning_rate": 5.7720498815979674e-05,
+      "loss": 0.3187,
+      "step": 9
+    },
+    {
+      "epoch": 0.013557483731019523,
+      "grad_norm": 0.17062894999980927,
+      "learning_rate": 6.048829123101417e-05,
+      "loss": 0.3255,
+      "step": 10
+    },
+    {
+      "epoch": 0.014913232104121476,
+      "grad_norm": 0.1766664683818817,
+      "learning_rate": 6.299206402569856e-05,
+      "loss": 0.3251,
+      "step": 11
+    },
+    {
+      "epoch": 0.016268980477223426,
+      "grad_norm": 0.16922810673713684,
+      "learning_rate": 6.527782950197749e-05,
+      "loss": 0.3268,
+      "step": 12
+    },
+    {
+      "epoch": 0.01762472885032538,
+      "grad_norm": 0.14840057492256165,
+      "learning_rate": 6.738052990918817e-05,
+      "loss": 0.3106,
+      "step": 13
+    },
+    {
+      "epoch": 0.01898047722342733,
+      "grad_norm": 0.1292823851108551,
+      "learning_rate": 6.932732641013545e-05,
+      "loss": 0.2938,
+      "step": 14
+    },
+    {
+      "epoch": 0.020336225596529284,
+      "grad_norm": 0.1699257791042328,
+      "learning_rate": 7.113975059201017e-05,
+      "loss": 0.3131,
+      "step": 15
+    },
+    {
+      "epoch": 0.021691973969631236,
+      "grad_norm": 0.1286689192056656,
+      "learning_rate": 7.28351601879753e-05,
+      "loss": 0.2532,
+      "step": 16
+    },
+    {
+      "epoch": 0.02304772234273319,
+      "grad_norm": 0.15176823735237122,
+      "learning_rate": 7.44277527012163e-05,
+      "loss": 0.3065,
+      "step": 17
+    },
+    {
+      "epoch": 0.02440347071583514,
+      "grad_norm": 0.12532933056354523,
+      "learning_rate": 7.59292888629735e-05,
+      "loss": 0.2943,
+      "step": 18
+    },
+    {
+      "epoch": 0.025759219088937094,
+      "grad_norm": 0.12740236520767212,
+      "learning_rate": 7.734962022714278e-05,
+      "loss": 0.2963,
+      "step": 19
+    },
+    {
+      "epoch": 0.027114967462039046,
+      "grad_norm": 0.12903445959091187,
+      "learning_rate": 7.869708127800799e-05,
+      "loss": 0.3112,
+      "step": 20
+    },
+    {
+      "epoch": 0.028470715835141,
+      "grad_norm": 0.13024842739105225,
+      "learning_rate": 7.997878577113146e-05,
+      "loss": 0.2982,
+      "step": 21
+    },
+    {
+      "epoch": 0.02982646420824295,
+      "grad_norm": 0.13486573100090027,
+      "learning_rate": 8.120085407269239e-05,
+      "loss": 0.2935,
+      "step": 22
+    },
+    {
+      "epoch": 0.031182212581344904,
+      "grad_norm": 0.12043455243110657,
+      "learning_rate": 8.236858992241086e-05,
+      "loss": 0.2716,
+      "step": 23
+    },
+    {
+      "epoch": 0.03253796095444685,
+      "grad_norm": 0.1397937536239624,
+      "learning_rate": 8.348661954897131e-05,
+      "loss": 0.3014,
+      "step": 24
+    },
+    {
+      "epoch": 0.03389370932754881,
+      "grad_norm": 0.1222696602344513,
+      "learning_rate": 8.455900236804067e-05,
+      "loss": 0.2732,
+      "step": 25
+    },
+    {
+      "epoch": 0.03524945770065076,
+      "grad_norm": 0.12033706158399582,
+      "learning_rate": 8.558931995618199e-05,
+      "loss": 0.2808,
+      "step": 26
+    },
+    {
+      "epoch": 0.036605206073752714,
+      "grad_norm": 0.13982202112674713,
+      "learning_rate": 8.65807482239695e-05,
+      "loss": 0.2726,
+      "step": 27
+    },
+    {
+      "epoch": 0.03796095444685466,
+      "grad_norm": 0.12233714759349823,
+      "learning_rate": 8.753611645712928e-05,
+      "loss": 0.2839,
+      "step": 28
+    },
+    {
+      "epoch": 0.03931670281995662,
+      "grad_norm": 0.11522915959358215,
+      "learning_rate": 8.84579559925641e-05,
+      "loss": 0.2776,
+      "step": 29
+    },
+    {
+      "epoch": 0.04067245119305857,
+      "grad_norm": 0.14247016608715057,
+      "learning_rate": 8.9348540639004e-05,
+      "loss": 0.2805,
+      "step": 30
+    },
+    {
+      "epoch": 0.04202819956616052,
+      "grad_norm": 0.12002000212669373,
+      "learning_rate": 9.020992046742607e-05,
+      "loss": 0.2587,
+      "step": 31
+    },
+    {
+      "epoch": 0.04338394793926247,
+      "grad_norm": 0.13062691688537598,
+      "learning_rate": 9.104395023496914e-05,
+      "loss": 0.2687,
+      "step": 32
+    },
+    {
+      "epoch": 0.04473969631236443,
+      "grad_norm": 0.14863581955432892,
+      "learning_rate": 9.18523134336884e-05,
+      "loss": 0.2648,
+      "step": 33
+    },
+    {
+      "epoch": 0.04609544468546638,
+      "grad_norm": 0.1561175286769867,
+      "learning_rate": 9.263654274821012e-05,
+      "loss": 0.2676,
+      "step": 34
+    },
+    {
+      "epoch": 0.047451193058568326,
+      "grad_norm": 0.1534643918275833,
+      "learning_rate": 9.339803754716197e-05,
+      "loss": 0.2771,
+      "step": 35
+    },
+    {
+      "epoch": 0.04880694143167028,
+      "grad_norm": 0.13867072761058807,
+      "learning_rate": 9.413807890996732e-05,
+      "loss": 0.254,
+      "step": 36
+    },
+    {
+      "epoch": 0.05016268980477223,
+      "grad_norm": 0.16411322355270386,
+      "learning_rate": 9.485784259434291e-05,
+      "loss": 0.2603,
+      "step": 37
+    },
+    {
+      "epoch": 0.05151843817787419,
+      "grad_norm": 0.1712709665298462,
+      "learning_rate": 9.555841027413661e-05,
+      "loss": 0.25,
+      "step": 38
+    },
+    {
+      "epoch": 0.052874186550976136,
+      "grad_norm": 0.17689694464206696,
+      "learning_rate": 9.624077931717798e-05,
+      "loss": 0.2669,
+      "step": 39
+    },
+    {
+      "epoch": 0.05422993492407809,
+      "grad_norm": 0.1266891360282898,
+      "learning_rate": 9.690587132500182e-05,
+      "loss": 0.2459,
+      "step": 40
+    },
+    {
+      "epoch": 0.05558568329718004,
+      "grad_norm": 0.12961214780807495,
+      "learning_rate": 9.755453961794159e-05,
+      "loss": 0.2405,
+      "step": 41
+    },
+    {
+      "epoch": 0.056941431670282,
+      "grad_norm": 0.13253234326839447,
+      "learning_rate": 9.818757581812529e-05,
+      "loss": 0.2657,
+      "step": 42
+    },
+    {
+      "epoch": 0.058297180043383946,
+      "grad_norm": 0.14288417994976044,
+      "learning_rate": 9.880571565777295e-05,
+      "loss": 0.2658,
+      "step": 43
+    },
+    {
+      "epoch": 0.0596529284164859,
+      "grad_norm": 0.16480830311775208,
+      "learning_rate": 9.94096441196862e-05,
+      "loss": 0.2502,
+      "step": 44
+    },
+    {
+      "epoch": 0.06100867678958785,
+      "grad_norm": 0.15202082693576813,
+      "learning_rate": 0.0001,
+      "loss": 0.2684,
+      "step": 45
+    },
+    {
+      "epoch": 0.06236442516268981,
+      "grad_norm": 0.15980659425258636,
+      "learning_rate": 0.0001,
+      "loss": 0.2674,
+      "step": 46
+    },
+    {
+      "epoch": 0.06372017353579176,
+      "grad_norm": 0.2368554025888443,
+      "learning_rate": 0.0001,
+      "loss": 0.2575,
+      "step": 47
+    },
+    {
+      "epoch": 0.0650759219088937,
+      "grad_norm": 0.1307699829339981,
+      "learning_rate": 0.0001,
+      "loss": 0.2596,
+      "step": 48
+    },
+    {
+      "epoch": 0.06643167028199566,
+      "grad_norm": 0.13891352713108063,
+      "learning_rate": 0.0001,
+      "loss": 0.2661,
+      "step": 49
+    },
+    {
+      "epoch": 0.06778741865509762,
+      "grad_norm": 0.11806360632181168,
+      "learning_rate": 0.0001,
+      "loss": 0.2469,
+      "step": 50
+    },
+    {
+      "epoch": 0.06914316702819957,
+      "grad_norm": 0.12294347584247589,
+      "learning_rate": 0.0001,
+      "loss": 0.2488,
+      "step": 51
+    },
+    {
+      "epoch": 0.07049891540130152,
+      "grad_norm": 0.1333683878183365,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 52
+    },
+    {
+      "epoch": 0.07185466377440347,
+      "grad_norm": 0.13483688235282898,
+      "learning_rate": 0.0001,
+      "loss": 0.2502,
+      "step": 53
+    },
+    {
+      "epoch": 0.07321041214750543,
+      "grad_norm": 0.13390597701072693,
+      "learning_rate": 0.0001,
+      "loss": 0.2478,
+      "step": 54
+    },
+    {
+      "epoch": 0.07456616052060737,
+      "grad_norm": 0.12571237981319427,
+      "learning_rate": 0.0001,
+      "loss": 0.2395,
+      "step": 55
+    },
+    {
+      "epoch": 0.07592190889370933,
+      "grad_norm": 0.14393945038318634,
+      "learning_rate": 0.0001,
+      "loss": 0.2426,
+      "step": 56
+    },
+    {
+      "epoch": 0.07727765726681128,
+      "grad_norm": 0.17252618074417114,
+      "learning_rate": 0.0001,
+      "loss": 0.2527,
+      "step": 57
+    },
+    {
+      "epoch": 0.07863340563991324,
+      "grad_norm": 0.11947691440582275,
+      "learning_rate": 0.0001,
+      "loss": 0.2396,
+      "step": 58
+    },
+    {
+      "epoch": 0.07998915401301518,
+      "grad_norm": 0.15561673045158386,
+      "learning_rate": 0.0001,
+      "loss": 0.2536,
+      "step": 59
+    },
+    {
+      "epoch": 0.08134490238611713,
+      "grad_norm": 0.18086113035678864,
+      "learning_rate": 0.0001,
+      "loss": 0.2408,
+      "step": 60
+    },
+    {
+      "epoch": 0.08270065075921909,
+      "grad_norm": 0.15263773500919342,
+      "learning_rate": 0.0001,
+      "loss": 0.2496,
+      "step": 61
+    },
+    {
+      "epoch": 0.08405639913232105,
+      "grad_norm": 0.13235825300216675,
+      "learning_rate": 0.0001,
+      "loss": 0.2509,
+      "step": 62
+    },
+    {
+      "epoch": 0.08541214750542299,
+      "grad_norm": 0.15344364941120148,
+      "learning_rate": 0.0001,
+      "loss": 0.2455,
+      "step": 63
+    },
+    {
+      "epoch": 0.08676789587852494,
+      "grad_norm": 0.12660299241542816,
+      "learning_rate": 0.0001,
+      "loss": 0.2438,
+      "step": 64
+    },
+    {
+      "epoch": 0.0881236442516269,
+      "grad_norm": 0.19659064710140228,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 65
+    },
+    {
+      "epoch": 0.08947939262472886,
+      "grad_norm": 0.15083153545856476,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 66
+    },
+    {
+      "epoch": 0.0908351409978308,
+      "grad_norm": 0.17154455184936523,
+      "learning_rate": 0.0001,
+      "loss": 0.2494,
+      "step": 67
+    },
+    {
+      "epoch": 0.09219088937093275,
+      "grad_norm": 0.1493559330701828,
+      "learning_rate": 0.0001,
+      "loss": 0.2587,
+      "step": 68
+    },
+    {
+      "epoch": 0.09354663774403471,
+      "grad_norm": 0.13032770156860352,
+      "learning_rate": 0.0001,
+      "loss": 0.2467,
+      "step": 69
+    },
+    {
+      "epoch": 0.09490238611713665,
+      "grad_norm": 0.12377802282571793,
+      "learning_rate": 0.0001,
+      "loss": 0.2497,
+      "step": 70
+    },
+    {
+      "epoch": 0.09625813449023861,
+      "grad_norm": 0.15083643794059753,
+      "learning_rate": 0.0001,
+      "loss": 0.2561,
+      "step": 71
+    },
+    {
+      "epoch": 0.09761388286334056,
+      "grad_norm": 0.14834199845790863,
+      "learning_rate": 0.0001,
+      "loss": 0.2592,
+      "step": 72
+    },
+    {
+      "epoch": 0.09896963123644252,
+      "grad_norm": 0.12019278109073639,
+      "learning_rate": 0.0001,
+      "loss": 0.2493,
+      "step": 73
+    },
+    {
+      "epoch": 0.10032537960954446,
+      "grad_norm": 0.12064848095178604,
+      "learning_rate": 0.0001,
+      "loss": 0.2429,
+      "step": 74
+    },
+    {
+      "epoch": 0.10168112798264642,
+      "grad_norm": 0.12743614614009857,
+      "learning_rate": 0.0001,
+      "loss": 0.2316,
+      "step": 75
+    },
+    {
+      "epoch": 0.10303687635574837,
+      "grad_norm": 0.12821342051029205,
+      "learning_rate": 0.0001,
+      "loss": 0.2552,
+      "step": 76
+    },
+    {
+      "epoch": 0.10439262472885033,
+      "grad_norm": 0.11239595711231232,
+      "learning_rate": 0.0001,
+      "loss": 0.2237,
+      "step": 77
+    },
+    {
+      "epoch": 0.10574837310195227,
+      "grad_norm": 0.11386699974536896,
+      "learning_rate": 0.0001,
+      "loss": 0.2379,
+      "step": 78
+    },
+    {
+      "epoch": 0.10710412147505423,
+      "grad_norm": 0.12098106741905212,
+      "learning_rate": 0.0001,
+      "loss": 0.2397,
+      "step": 79
+    },
+    {
+      "epoch": 0.10845986984815618,
+      "grad_norm": 0.12121766060590744,
+      "learning_rate": 0.0001,
+      "loss": 0.2392,
+      "step": 80
+    },
+    {
+      "epoch": 0.10981561822125814,
+      "grad_norm": 0.12529510259628296,
+      "learning_rate": 0.0001,
+      "loss": 0.2391,
+      "step": 81
+    },
+    {
+      "epoch": 0.11117136659436008,
+      "grad_norm": 0.13585695624351501,
+      "learning_rate": 0.0001,
+      "loss": 0.2279,
+      "step": 82
+    },
+    {
+      "epoch": 0.11252711496746204,
+      "grad_norm": 0.1295752078294754,
+      "learning_rate": 0.0001,
+      "loss": 0.2491,
+      "step": 83
+    },
+    {
+      "epoch": 0.113882863340564,
+      "grad_norm": 0.14579786360263824,
+      "learning_rate": 0.0001,
+      "loss": 0.2529,
+      "step": 84
+    },
+    {
+      "epoch": 0.11523861171366595,
+      "grad_norm": 0.15249188244342804,
+      "learning_rate": 0.0001,
+      "loss": 0.254,
+      "step": 85
+    },
+    {
+      "epoch": 0.11659436008676789,
+      "grad_norm": 0.12780269980430603,
+      "learning_rate": 0.0001,
+      "loss": 0.2391,
+      "step": 86
+    },
+    {
+      "epoch": 0.11795010845986985,
+      "grad_norm": 0.13740313053131104,
+      "learning_rate": 0.0001,
+      "loss": 0.2384,
+      "step": 87
+    },
+    {
+      "epoch": 0.1193058568329718,
+      "grad_norm": 0.13653717935085297,
+      "learning_rate": 0.0001,
+      "loss": 0.2342,
+      "step": 88
+    },
+    {
+      "epoch": 0.12066160520607375,
+      "grad_norm": 0.15235239267349243,
+      "learning_rate": 0.0001,
+      "loss": 0.2186,
+      "step": 89
+    },
+    {
+      "epoch": 0.1220173535791757,
+      "grad_norm": 0.1675601750612259,
+      "learning_rate": 0.0001,
+      "loss": 0.2438,
+      "step": 90
+    },
+    {
+      "epoch": 0.12337310195227766,
+      "grad_norm": 0.14582861959934235,
+      "learning_rate": 0.0001,
+      "loss": 0.2532,
+      "step": 91
+    },
+    {
+      "epoch": 0.12472885032537961,
+      "grad_norm": 0.1964140683412552,
+      "learning_rate": 0.0001,
+      "loss": 0.2553,
+      "step": 92
+    },
+    {
+      "epoch": 0.12608459869848157,
+      "grad_norm": 0.1260741651058197,
+      "learning_rate": 0.0001,
+      "loss": 0.2433,
+      "step": 93
+    },
+    {
+      "epoch": 0.12744034707158353,
+      "grad_norm": 0.20553891360759735,
+      "learning_rate": 0.0001,
+      "loss": 0.2317,
+      "step": 94
+    },
+    {
+      "epoch": 0.12879609544468545,
+      "grad_norm": 0.17866551876068115,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 95
+    },
+    {
+      "epoch": 0.1301518438177874,
+      "grad_norm": 0.15488256514072418,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 96
+    },
+    {
+      "epoch": 0.13150759219088937,
+      "grad_norm": 0.10783703625202179,
+      "learning_rate": 0.0001,
+      "loss": 0.2437,
+      "step": 97
+    },
+    {
+      "epoch": 0.13286334056399132,
+      "grad_norm": 0.10871797055006027,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 98
+    },
+    {
+      "epoch": 0.13421908893709328,
+      "grad_norm": 0.11559201776981354,
+      "learning_rate": 0.0001,
+      "loss": 0.2254,
+      "step": 99
+    },
+    {
+      "epoch": 0.13557483731019523,
+      "grad_norm": 0.13876475393772125,
+      "learning_rate": 0.0001,
+      "loss": 0.2334,
+      "step": 100
+    },
+    {
+      "epoch": 0.1369305856832972,
+      "grad_norm": 0.1475064605474472,
+      "learning_rate": 0.0001,
+      "loss": 0.2464,
+      "step": 101
+    },
+    {
+      "epoch": 0.13828633405639915,
+      "grad_norm": 0.11311455816030502,
+      "learning_rate": 0.0001,
+      "loss": 0.2243,
+      "step": 102
+    },
+    {
+      "epoch": 0.13964208242950107,
+      "grad_norm": 0.12905581295490265,
+      "learning_rate": 0.0001,
+      "loss": 0.2318,
+      "step": 103
+    },
+    {
+      "epoch": 0.14099783080260303,
+      "grad_norm": 0.1492420881986618,
+      "learning_rate": 0.0001,
+      "loss": 0.2486,
+      "step": 104
+    },
+    {
+      "epoch": 0.142353579175705,
+      "grad_norm": 0.12183994799852371,
+      "learning_rate": 0.0001,
+      "loss": 0.2124,
+      "step": 105
+    },
+    {
+      "epoch": 0.14370932754880694,
+      "grad_norm": 0.1277727633714676,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 106
+    },
+    {
+      "epoch": 0.1450650759219089,
+      "grad_norm": 0.1179315373301506,
+      "learning_rate": 0.0001,
+      "loss": 0.2261,
+      "step": 107
+    },
+    {
+      "epoch": 0.14642082429501085,
+      "grad_norm": 0.1135738417506218,
+      "learning_rate": 0.0001,
+      "loss": 0.2335,
+      "step": 108
+    },
+    {
+      "epoch": 0.1477765726681128,
+      "grad_norm": 0.1145508885383606,
+      "learning_rate": 0.0001,
+      "loss": 0.2198,
+      "step": 109
+    },
+    {
+      "epoch": 0.14913232104121474,
+      "grad_norm": 0.12893252074718475,
+      "learning_rate": 0.0001,
+      "loss": 0.2327,
+      "step": 110
+    },
+    {
+      "epoch": 0.1504880694143167,
+      "grad_norm": 0.11888323724269867,
+      "learning_rate": 0.0001,
+      "loss": 0.2294,
+      "step": 111
+    },
+    {
+      "epoch": 0.15184381778741865,
+      "grad_norm": 0.1548166424036026,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 112
+    },
+    {
+      "epoch": 0.1531995661605206,
+      "grad_norm": 0.1422528773546219,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 113
+    },
+    {
+      "epoch": 0.15455531453362256,
+      "grad_norm": 0.12698225677013397,
+      "learning_rate": 0.0001,
+      "loss": 0.2297,
+      "step": 114
+    },
+    {
+      "epoch": 0.15591106290672452,
+      "grad_norm": 0.11766814440488815,
+      "learning_rate": 0.0001,
+      "loss": 0.2312,
+      "step": 115
+    },
+    {
+      "epoch": 0.15726681127982647,
+      "grad_norm": 0.13471224904060364,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 116
+    },
+    {
+      "epoch": 0.15862255965292843,
+      "grad_norm": 0.15393678843975067,
+      "learning_rate": 0.0001,
+      "loss": 0.234,
+      "step": 117
+    },
+    {
+      "epoch": 0.15997830802603036,
+      "grad_norm": 0.126940056681633,
+      "learning_rate": 0.0001,
+      "loss": 0.2426,
+      "step": 118
+    },
+    {
+      "epoch": 0.16133405639913231,
+      "grad_norm": 0.11630528420209885,
+      "learning_rate": 0.0001,
+      "loss": 0.2332,
+      "step": 119
+    },
+    {
+      "epoch": 0.16268980477223427,
+      "grad_norm": 0.12796275317668915,
+      "learning_rate": 0.0001,
+      "loss": 0.225,
+      "step": 120
+    },
+    {
+      "epoch": 0.16404555314533623,
+      "grad_norm": 0.143882155418396,
+      "learning_rate": 0.0001,
+      "loss": 0.2276,
+      "step": 121
+    },
+    {
+      "epoch": 0.16540130151843818,
+      "grad_norm": 0.17849501967430115,
+      "learning_rate": 0.0001,
+      "loss": 0.2406,
+      "step": 122
+    },
+    {
+      "epoch": 0.16675704989154014,
+      "grad_norm": 0.1186647042632103,
+      "learning_rate": 0.0001,
+      "loss": 0.2283,
+      "step": 123
+    },
+    {
+      "epoch": 0.1681127982646421,
+      "grad_norm": 0.13291320204734802,
+      "learning_rate": 0.0001,
+      "loss": 0.2453,
+      "step": 124
+    },
+    {
+      "epoch": 0.16946854663774402,
+      "grad_norm": 0.12601213157176971,
+      "learning_rate": 0.0001,
+      "loss": 0.2476,
+      "step": 125
+    },
+    {
+      "epoch": 0.17082429501084598,
+      "grad_norm": 0.11712577939033508,
+      "learning_rate": 0.0001,
+      "loss": 0.2422,
+      "step": 126
+    },
+    {
+      "epoch": 0.17218004338394793,
+      "grad_norm": 0.11569856852293015,
+      "learning_rate": 0.0001,
+      "loss": 0.2285,
+      "step": 127
+    },
+    {
+      "epoch": 0.1735357917570499,
+      "grad_norm": 0.10918959230184555,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 128
+    },
+    {
+      "epoch": 0.17489154013015185,
+      "grad_norm": 0.13561537861824036,
+      "learning_rate": 0.0001,
+      "loss": 0.2373,
+      "step": 129
+    },
+    {
+      "epoch": 0.1762472885032538,
+      "grad_norm": 0.14456422626972198,
+      "learning_rate": 0.0001,
+      "loss": 0.2501,
+      "step": 130
+    },
+    {
+      "epoch": 0.17760303687635576,
+      "grad_norm": 0.13148345053195953,
+      "learning_rate": 0.0001,
+      "loss": 0.2288,
+      "step": 131
+    },
+    {
+      "epoch": 0.1789587852494577,
+      "grad_norm": 0.15666794776916504,
+      "learning_rate": 0.0001,
+      "loss": 0.2421,
+      "step": 132
+    },
+    {
+      "epoch": 0.18031453362255964,
+      "grad_norm": 0.17364318668842316,
+      "learning_rate": 0.0001,
+      "loss": 0.2527,
+      "step": 133
+    },
+    {
+      "epoch": 0.1816702819956616,
+      "grad_norm": 0.1292874664068222,
+      "learning_rate": 0.0001,
+      "loss": 0.2292,
+      "step": 134
+    },
+    {
+      "epoch": 0.18302603036876355,
+      "grad_norm": 0.1437186449766159,
+      "learning_rate": 0.0001,
+      "loss": 0.2042,
+      "step": 135
+    },
+    {
+      "epoch": 0.1843817787418655,
+      "grad_norm": 0.1345144361257553,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 136
+    },
+    {
+      "epoch": 0.18573752711496747,
+      "grad_norm": 0.1124841719865799,
+      "learning_rate": 0.0001,
+      "loss": 0.2321,
+      "step": 137
+    },
+    {
+      "epoch": 0.18709327548806942,
+      "grad_norm": 0.12669357657432556,
+      "learning_rate": 0.0001,
+      "loss": 0.2211,
+      "step": 138
+    },
+    {
+      "epoch": 0.18844902386117138,
+      "grad_norm": 0.14159433543682098,
+      "learning_rate": 0.0001,
+      "loss": 0.232,
+      "step": 139
+    },
+    {
+      "epoch": 0.1898047722342733,
+      "grad_norm": 0.12368479371070862,
+      "learning_rate": 0.0001,
+      "loss": 0.2332,
+      "step": 140
+    },
+    {
+      "epoch": 0.19116052060737526,
+      "grad_norm": 0.118979811668396,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 141
+    },
+    {
+      "epoch": 0.19251626898047722,
+      "grad_norm": 0.10225681960582733,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 142
+    },
+    {
+      "epoch": 0.19387201735357917,
+      "grad_norm": 0.11519869416952133,
+      "learning_rate": 0.0001,
+      "loss": 0.2223,
+      "step": 143
+    },
+    {
+      "epoch": 0.19522776572668113,
+      "grad_norm": 0.12480252981185913,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 144
+    },
+    {
+      "epoch": 0.19658351409978309,
+      "grad_norm": 0.1287064403295517,
+      "learning_rate": 0.0001,
+      "loss": 0.2349,
+      "step": 145
+    },
+    {
+      "epoch": 0.19793926247288504,
+      "grad_norm": 0.152523010969162,
+      "learning_rate": 0.0001,
+      "loss": 0.2377,
+      "step": 146
+    },
+    {
+      "epoch": 0.199295010845987,
+      "grad_norm": 0.13616617023944855,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 147
+    },
+    {
+      "epoch": 0.20065075921908893,
+      "grad_norm": 0.145973339676857,
+      "learning_rate": 0.0001,
+      "loss": 0.2315,
+      "step": 148
+    },
+    {
+      "epoch": 0.20200650759219088,
+      "grad_norm": 0.14834924042224884,
+      "learning_rate": 0.0001,
+      "loss": 0.2321,
+      "step": 149
+    },
+    {
+      "epoch": 0.20336225596529284,
+      "grad_norm": 0.1307169795036316,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 150
+    },
+    {
+      "epoch": 0.2047180043383948,
+      "grad_norm": 0.12045511603355408,
+      "learning_rate": 0.0001,
+      "loss": 0.2325,
+      "step": 151
+    },
+    {
+      "epoch": 0.20607375271149675,
+      "grad_norm": 0.12030395120382309,
+      "learning_rate": 0.0001,
+      "loss": 0.2313,
+      "step": 152
+    },
+    {
+      "epoch": 0.2074295010845987,
+      "grad_norm": 0.10707315057516098,
+      "learning_rate": 0.0001,
+      "loss": 0.2488,
+      "step": 153
+    },
+    {
+      "epoch": 0.20878524945770066,
+      "grad_norm": 0.09037518501281738,
+      "learning_rate": 0.0001,
+      "loss": 0.2139,
+      "step": 154
+    },
+    {
+      "epoch": 0.2101409978308026,
+      "grad_norm": 0.11992931365966797,
+      "learning_rate": 0.0001,
+      "loss": 0.2458,
+      "step": 155
+    },
+    {
+      "epoch": 0.21149674620390455,
+      "grad_norm": 0.12397589534521103,
+      "learning_rate": 0.0001,
+      "loss": 0.2521,
+      "step": 156
+    },
+    {
+      "epoch": 0.2128524945770065,
+      "grad_norm": 0.09755264222621918,
+      "learning_rate": 0.0001,
+      "loss": 0.2279,
+      "step": 157
+    },
+    {
+      "epoch": 0.21420824295010846,
+      "grad_norm": 0.11418113857507706,
+      "learning_rate": 0.0001,
+      "loss": 0.2341,
+      "step": 158
+    },
+    {
+      "epoch": 0.2155639913232104,
+      "grad_norm": 0.10694784671068192,
+      "learning_rate": 0.0001,
+      "loss": 0.2388,
+      "step": 159
+    },
+    {
+      "epoch": 0.21691973969631237,
+      "grad_norm": 0.11156194657087326,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 160
+    },
+    {
+      "epoch": 0.21827548806941433,
+      "grad_norm": 0.13871735334396362,
+      "learning_rate": 0.0001,
+      "loss": 0.241,
+      "step": 161
+    },
+    {
+      "epoch": 0.21963123644251628,
+      "grad_norm": 0.14062196016311646,
+      "learning_rate": 0.0001,
+      "loss": 0.2318,
+      "step": 162
+    },
+    {
+      "epoch": 0.2209869848156182,
+      "grad_norm": 0.10900415480136871,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 163
+    },
+    {
+      "epoch": 0.22234273318872017,
+      "grad_norm": 0.1620100885629654,
+      "learning_rate": 0.0001,
+      "loss": 0.2189,
+      "step": 164
+    },
+    {
+      "epoch": 0.22369848156182212,
+      "grad_norm": 0.11354006826877594,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 165
+    },
+    {
+      "epoch": 0.22505422993492408,
+      "grad_norm": 0.11705230176448822,
+      "learning_rate": 0.0001,
+      "loss": 0.2327,
+      "step": 166
+    },
+    {
+      "epoch": 0.22640997830802603,
+      "grad_norm": 0.13396400213241577,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 167
+    },
+    {
+      "epoch": 0.227765726681128,
+      "grad_norm": 0.11610525846481323,
+      "learning_rate": 0.0001,
+      "loss": 0.2395,
+      "step": 168
+    },
+    {
+      "epoch": 0.22912147505422994,
+      "grad_norm": 0.18780137598514557,
+      "learning_rate": 0.0001,
+      "loss": 0.2288,
+      "step": 169
+    },
+    {
+      "epoch": 0.2304772234273319,
+      "grad_norm": 0.12105163931846619,
+      "learning_rate": 0.0001,
+      "loss": 0.2512,
+      "step": 170
+    },
+    {
+      "epoch": 0.23183297180043383,
+      "grad_norm": 0.1066543236374855,
+      "learning_rate": 0.0001,
+      "loss": 0.2242,
+      "step": 171
+    },
+    {
+      "epoch": 0.23318872017353579,
+      "grad_norm": 0.12877358496189117,
+      "learning_rate": 0.0001,
+      "loss": 0.2213,
+      "step": 172
+    },
+    {
+      "epoch": 0.23454446854663774,
+      "grad_norm": 0.10897350311279297,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 173
+    },
+    {
+      "epoch": 0.2359002169197397,
+      "grad_norm": 0.12142904102802277,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 174
+    },
+    {
+      "epoch": 0.23725596529284165,
+      "grad_norm": 0.12763909995555878,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 175
+    },
+    {
+      "epoch": 0.2386117136659436,
+      "grad_norm": 0.1041029840707779,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 176
+    },
+    {
+      "epoch": 0.23996746203904556,
+      "grad_norm": 0.12840965390205383,
+      "learning_rate": 0.0001,
+      "loss": 0.2409,
+      "step": 177
+    },
+    {
+      "epoch": 0.2413232104121475,
+      "grad_norm": 0.1092042475938797,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "step": 178
+    },
+    {
+      "epoch": 0.24267895878524945,
+      "grad_norm": 0.10624870657920837,
+      "learning_rate": 0.0001,
+      "loss": 0.2279,
+      "step": 179
+    },
+    {
+      "epoch": 0.2440347071583514,
+      "grad_norm": 0.10692845284938812,
+      "learning_rate": 0.0001,
+      "loss": 0.2188,
+      "step": 180
+    },
+    {
+      "epoch": 0.24539045553145336,
+      "grad_norm": 0.121277816593647,
+      "learning_rate": 0.0001,
+      "loss": 0.2301,
+      "step": 181
+    },
+    {
+      "epoch": 0.24674620390455532,
+      "grad_norm": 0.13551722466945648,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 182
+    },
+    {
+      "epoch": 0.24810195227765727,
+      "grad_norm": 0.1198442280292511,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 183
+    },
+    {
+      "epoch": 0.24945770065075923,
+      "grad_norm": 0.1625175178050995,
+      "learning_rate": 0.0001,
+      "loss": 0.2336,
+      "step": 184
+    },
+    {
+      "epoch": 0.25081344902386116,
+      "grad_norm": 0.11888563632965088,
+      "learning_rate": 0.0001,
+      "loss": 0.2207,
+      "step": 185
+    },
+    {
+      "epoch": 0.25216919739696314,
+      "grad_norm": 0.1410103291273117,
+      "learning_rate": 0.0001,
+      "loss": 0.2323,
+      "step": 186
+    },
+    {
+      "epoch": 0.25352494577006507,
+      "grad_norm": 0.11480198055505753,
+      "learning_rate": 0.0001,
+      "loss": 0.2442,
+      "step": 187
+    },
+    {
+      "epoch": 0.25488069414316705,
+      "grad_norm": 0.12025165557861328,
+      "learning_rate": 0.0001,
+      "loss": 0.2257,
+      "step": 188
+    },
+    {
+      "epoch": 0.256236442516269,
+      "grad_norm": 0.09422005712985992,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 189
+    },
+    {
+      "epoch": 0.2575921908893709,
+      "grad_norm": 0.11959129571914673,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 190
+    },
+    {
+      "epoch": 0.2589479392624729,
+      "grad_norm": 0.10061118006706238,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 191
+    },
+    {
+      "epoch": 0.2603036876355748,
+      "grad_norm": 0.09783045202493668,
+      "learning_rate": 0.0001,
+      "loss": 0.2009,
+      "step": 192
+    },
+    {
+      "epoch": 0.2616594360086768,
+      "grad_norm": 0.08924517035484314,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 193
+    },
+    {
+      "epoch": 0.26301518438177873,
+      "grad_norm": 0.1062329113483429,
+      "learning_rate": 0.0001,
+      "loss": 0.2158,
+      "step": 194
+    },
+    {
+      "epoch": 0.2643709327548807,
+      "grad_norm": 0.11631181836128235,
+      "learning_rate": 0.0001,
+      "loss": 0.2183,
+      "step": 195
+    },
+    {
+      "epoch": 0.26572668112798264,
+      "grad_norm": 0.1399419903755188,
+      "learning_rate": 0.0001,
+      "loss": 0.2187,
+      "step": 196
+    },
+    {
+      "epoch": 0.2670824295010846,
+      "grad_norm": 0.1161571741104126,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 197
+    },
+    {
+      "epoch": 0.26843817787418656,
+      "grad_norm": 0.11201664060354233,
+      "learning_rate": 0.0001,
+      "loss": 0.2105,
+      "step": 198
+    },
+    {
+      "epoch": 0.2697939262472885,
+      "grad_norm": 0.12170354276895523,
+      "learning_rate": 0.0001,
+      "loss": 0.232,
+      "step": 199
+    },
+    {
+      "epoch": 0.27114967462039047,
+      "grad_norm": 0.13186457753181458,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 200
+    },
+    {
+      "epoch": 0.2725054229934924,
+      "grad_norm": 0.13673870265483856,
+      "learning_rate": 0.0001,
+      "loss": 0.2203,
+      "step": 201
+    },
+    {
+      "epoch": 0.2738611713665944,
+      "grad_norm": 0.10910363495349884,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 202
+    },
+    {
+      "epoch": 0.2752169197396963,
+      "grad_norm": 0.11227238178253174,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 203
+    },
+    {
+      "epoch": 0.2765726681127983,
+      "grad_norm": 0.13266952335834503,
+      "learning_rate": 0.0001,
+      "loss": 0.246,
+      "step": 204
+    },
+    {
+      "epoch": 0.2779284164859002,
+      "grad_norm": 0.12319870293140411,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 205
+    },
+    {
+      "epoch": 0.27928416485900215,
+      "grad_norm": 0.11409920454025269,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 206
+    },
+    {
+      "epoch": 0.28063991323210413,
+      "grad_norm": 0.1181999072432518,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 207
+    },
+    {
+      "epoch": 0.28199566160520606,
+      "grad_norm": 0.10515511780977249,
+      "learning_rate": 0.0001,
+      "loss": 0.2078,
+      "step": 208
+    },
+    {
+      "epoch": 0.28335140997830804,
+      "grad_norm": 0.11817390471696854,
+      "learning_rate": 0.0001,
+      "loss": 0.209,
+      "step": 209
+    },
+    {
+      "epoch": 0.28470715835141,
+      "grad_norm": 0.11866628378629684,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 210
+    },
+    {
+      "epoch": 0.28606290672451196,
+      "grad_norm": 0.13757768273353577,
+      "learning_rate": 0.0001,
+      "loss": 0.2254,
+      "step": 211
+    },
+    {
+      "epoch": 0.2874186550976139,
+      "grad_norm": 0.14699029922485352,
+      "learning_rate": 0.0001,
+      "loss": 0.2186,
+      "step": 212
+    },
+    {
+      "epoch": 0.2887744034707158,
+      "grad_norm": 0.11950292438268661,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 213
+    },
+    {
+      "epoch": 0.2901301518438178,
+      "grad_norm": 0.1287502497434616,
+      "learning_rate": 0.0001,
+      "loss": 0.2232,
+      "step": 214
+    },
+    {
+      "epoch": 0.2914859002169197,
+      "grad_norm": 0.1426144242286682,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 215
+    },
+    {
+      "epoch": 0.2928416485900217,
+      "grad_norm": 0.18343636393547058,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 216
+    },
+    {
+      "epoch": 0.29419739696312364,
+      "grad_norm": 0.15232282876968384,
+      "learning_rate": 0.0001,
+      "loss": 0.2216,
+      "step": 217
+    },
+    {
+      "epoch": 0.2955531453362256,
+      "grad_norm": 0.09034806489944458,
+      "learning_rate": 0.0001,
+      "loss": 0.2166,
+      "step": 218
+    },
+    {
+      "epoch": 0.29690889370932755,
+      "grad_norm": 0.10274829715490341,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 219
+    },
+    {
+      "epoch": 0.2982646420824295,
+      "grad_norm": 0.09992995113134384,
+      "learning_rate": 0.0001,
+      "loss": 0.2153,
+      "step": 220
+    },
+    {
+      "epoch": 0.29962039045553146,
+      "grad_norm": 0.13918644189834595,
+      "learning_rate": 0.0001,
+      "loss": 0.2174,
+      "step": 221
+    },
+    {
+      "epoch": 0.3009761388286334,
+      "grad_norm": 0.12290824949741364,
+      "learning_rate": 0.0001,
+      "loss": 0.2104,
+      "step": 222
+    },
+    {
+      "epoch": 0.30233188720173537,
+      "grad_norm": 0.08313800394535065,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 223
+    },
+    {
+      "epoch": 0.3036876355748373,
+      "grad_norm": 0.10721699148416519,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 224
+    },
+    {
+      "epoch": 0.3050433839479393,
+      "grad_norm": 0.11350128799676895,
+      "learning_rate": 0.0001,
+      "loss": 0.2304,
+      "step": 225
+    },
+    {
+      "epoch": 0.3063991323210412,
+      "grad_norm": 0.11244690418243408,
+      "learning_rate": 0.0001,
+      "loss": 0.2059,
+      "step": 226
+    },
+    {
+      "epoch": 0.30775488069414314,
+      "grad_norm": 0.11260519921779633,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 227
+    },
+    {
+      "epoch": 0.3091106290672451,
+      "grad_norm": 0.11820735782384872,
+      "learning_rate": 0.0001,
+      "loss": 0.225,
+      "step": 228
+    },
+    {
+      "epoch": 0.31046637744034705,
+      "grad_norm": 0.11709001660346985,
+      "learning_rate": 0.0001,
+      "loss": 0.2154,
+      "step": 229
+    },
+    {
+      "epoch": 0.31182212581344904,
+      "grad_norm": 0.11795724928379059,
+      "learning_rate": 0.0001,
+      "loss": 0.2396,
+      "step": 230
+    },
+    {
+      "epoch": 0.31317787418655096,
+      "grad_norm": 0.11690838634967804,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 231
+    },
+    {
+      "epoch": 0.31453362255965295,
+      "grad_norm": 0.1207461804151535,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 232
+    },
+    {
+      "epoch": 0.3158893709327549,
+      "grad_norm": 0.12474185973405838,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 233
+    },
+    {
+      "epoch": 0.31724511930585686,
+      "grad_norm": 0.10955285280942917,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 234
+    },
+    {
+      "epoch": 0.3186008676789588,
+      "grad_norm": 0.12927089631557465,
+      "learning_rate": 0.0001,
+      "loss": 0.2273,
+      "step": 235
+    },
+    {
+      "epoch": 0.3199566160520607,
+      "grad_norm": 0.11585133522748947,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 236
+    },
+    {
+      "epoch": 0.3213123644251627,
+      "grad_norm": 0.1234295442700386,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 237
+    },
+    {
+      "epoch": 0.32266811279826463,
+      "grad_norm": 0.13424554467201233,
+      "learning_rate": 0.0001,
+      "loss": 0.2282,
+      "step": 238
+    },
+    {
+      "epoch": 0.3240238611713666,
+      "grad_norm": 0.1243305578827858,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 239
+    },
+    {
+      "epoch": 0.32537960954446854,
+      "grad_norm": 0.1105075404047966,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 240
+    },
+    {
+      "epoch": 0.3267353579175705,
+      "grad_norm": 0.12168488651514053,
+      "learning_rate": 0.0001,
+      "loss": 0.2162,
+      "step": 241
+    },
+    {
+      "epoch": 0.32809110629067245,
+      "grad_norm": 0.1125117763876915,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 242
+    },
+    {
+      "epoch": 0.3294468546637744,
+      "grad_norm": 0.11101528257131577,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 243
+    },
+    {
+      "epoch": 0.33080260303687636,
+      "grad_norm": 0.13331718742847443,
+      "learning_rate": 0.0001,
+      "loss": 0.2273,
+      "step": 244
+    },
+    {
+      "epoch": 0.3321583514099783,
+      "grad_norm": 0.13222917914390564,
+      "learning_rate": 0.0001,
+      "loss": 0.24,
+      "step": 245
+    },
+    {
+      "epoch": 0.3335140997830803,
+      "grad_norm": 0.12964850664138794,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "step": 246
+    },
+    {
+      "epoch": 0.3348698481561822,
+      "grad_norm": 0.12921074032783508,
+      "learning_rate": 0.0001,
+      "loss": 0.2256,
+      "step": 247
+    },
+    {
+      "epoch": 0.3362255965292842,
+      "grad_norm": 0.10905551165342331,
+      "learning_rate": 0.0001,
+      "loss": 0.2212,
+      "step": 248
+    },
+    {
+      "epoch": 0.3375813449023861,
+      "grad_norm": 0.11016571521759033,
+      "learning_rate": 0.0001,
+      "loss": 0.218,
+      "step": 249
+    },
+    {
+      "epoch": 0.33893709327548804,
+      "grad_norm": 0.13196800649166107,
+      "learning_rate": 0.0001,
+      "loss": 0.2266,
+      "step": 250
+    },
+    {
+      "epoch": 0.34029284164859,
+      "grad_norm": 0.13022322952747345,
+      "learning_rate": 0.0001,
+      "loss": 0.2203,
+      "step": 251
+    },
+    {
+      "epoch": 0.34164859002169196,
+      "grad_norm": 0.12467219680547714,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 252
+    },
+    {
+      "epoch": 0.34300433839479394,
+      "grad_norm": 0.11198827624320984,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 253
+    },
+    {
+      "epoch": 0.34436008676789587,
+      "grad_norm": 0.12606126070022583,
+      "learning_rate": 0.0001,
+      "loss": 0.2229,
+      "step": 254
+    },
+    {
+      "epoch": 0.34571583514099785,
+      "grad_norm": 0.11265147477388382,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 255
+    },
+    {
+      "epoch": 0.3470715835140998,
+      "grad_norm": 0.11789364367723465,
+      "learning_rate": 0.0001,
+      "loss": 0.2213,
+      "step": 256
+    },
+    {
+      "epoch": 0.3484273318872017,
+      "grad_norm": 0.12589780986309052,
+      "learning_rate": 0.0001,
+      "loss": 0.2171,
+      "step": 257
+    },
+    {
+      "epoch": 0.3497830802603037,
+      "grad_norm": 0.1173444613814354,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 258
+    },
+    {
+      "epoch": 0.3511388286334056,
+      "grad_norm": 0.0853557363152504,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 259
+    },
+    {
+      "epoch": 0.3524945770065076,
+      "grad_norm": 0.12345444411039352,
+      "learning_rate": 0.0001,
+      "loss": 0.2139,
+      "step": 260
+    },
+    {
+      "epoch": 0.35385032537960953,
+      "grad_norm": 0.12503233551979065,
+      "learning_rate": 0.0001,
+      "loss": 0.2146,
+      "step": 261
+    },
+    {
+      "epoch": 0.3552060737527115,
+      "grad_norm": 0.11263001710176468,
+      "learning_rate": 0.0001,
+      "loss": 0.2284,
+      "step": 262
+    },
+    {
+      "epoch": 0.35656182212581344,
+      "grad_norm": 0.13283030688762665,
+      "learning_rate": 0.0001,
+      "loss": 0.2181,
+      "step": 263
+    },
+    {
+      "epoch": 0.3579175704989154,
+      "grad_norm": 0.134423166513443,
+      "learning_rate": 0.0001,
+      "loss": 0.2068,
+      "step": 264
+    },
+    {
+      "epoch": 0.35927331887201736,
+      "grad_norm": 0.12410351634025574,
+      "learning_rate": 0.0001,
+      "loss": 0.226,
+      "step": 265
+    },
+    {
+      "epoch": 0.3606290672451193,
+      "grad_norm": 0.12743990123271942,
+      "learning_rate": 0.0001,
+      "loss": 0.2264,
+      "step": 266
+    },
+    {
+      "epoch": 0.36198481561822127,
+      "grad_norm": 0.1160358265042305,
+      "learning_rate": 0.0001,
+      "loss": 0.2415,
+      "step": 267
+    },
+    {
+      "epoch": 0.3633405639913232,
+      "grad_norm": 0.10311786830425262,
+      "learning_rate": 0.0001,
+      "loss": 0.2321,
+      "step": 268
+    },
+    {
+      "epoch": 0.3646963123644252,
+      "grad_norm": 0.11841302365064621,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 269
+    },
+    {
+      "epoch": 0.3660520607375271,
+      "grad_norm": 0.10332412272691727,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 270
+    },
+    {
+      "epoch": 0.3674078091106291,
+      "grad_norm": 0.10372753441333771,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 271
+    },
+    {
+      "epoch": 0.368763557483731,
+      "grad_norm": 0.12738436460494995,
+      "learning_rate": 0.0001,
+      "loss": 0.2274,
+      "step": 272
+    },
+    {
+      "epoch": 0.37011930585683295,
+      "grad_norm": 0.1005726009607315,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 273
+    },
+    {
+      "epoch": 0.37147505422993493,
+      "grad_norm": 0.10664396733045578,
+      "learning_rate": 0.0001,
+      "loss": 0.2281,
+      "step": 274
+    },
+    {
+      "epoch": 0.37283080260303686,
+      "grad_norm": 0.13087576627731323,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 275
+    },
+    {
+      "epoch": 0.37418655097613884,
+      "grad_norm": 0.11902423948049545,
+      "learning_rate": 0.0001,
+      "loss": 0.2417,
+      "step": 276
+    },
+    {
+      "epoch": 0.37554229934924077,
+      "grad_norm": 0.09176304191350937,
+      "learning_rate": 0.0001,
+      "loss": 0.2176,
+      "step": 277
+    },
+    {
+      "epoch": 0.37689804772234275,
+      "grad_norm": 0.10103939473628998,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 278
+    },
+    {
+      "epoch": 0.3782537960954447,
+      "grad_norm": 0.10040771216154099,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 279
+    },
+    {
+      "epoch": 0.3796095444685466,
+      "grad_norm": 0.09590920060873032,
+      "learning_rate": 0.0001,
+      "loss": 0.2139,
+      "step": 280
+    },
+    {
+      "epoch": 0.3809652928416486,
+      "grad_norm": 0.11730661243200302,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 281
+    },
+    {
+      "epoch": 0.3823210412147505,
+      "grad_norm": 0.10751647502183914,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 282
+    },
+    {
+      "epoch": 0.3836767895878525,
+      "grad_norm": 0.12843811511993408,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 283
+    },
+    {
+      "epoch": 0.38503253796095444,
+      "grad_norm": 0.13020364940166473,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 284
+    },
+    {
+      "epoch": 0.3863882863340564,
+      "grad_norm": 0.1126871183514595,
+      "learning_rate": 0.0001,
+      "loss": 0.2143,
+      "step": 285
+    },
+    {
+      "epoch": 0.38774403470715835,
+      "grad_norm": 0.11909125000238419,
+      "learning_rate": 0.0001,
+      "loss": 0.2312,
+      "step": 286
+    },
+    {
+      "epoch": 0.38909978308026033,
+      "grad_norm": 0.1167333722114563,
+      "learning_rate": 0.0001,
+      "loss": 0.2199,
+      "step": 287
+    },
+    {
+      "epoch": 0.39045553145336226,
+      "grad_norm": 0.11328355967998505,
+      "learning_rate": 0.0001,
+      "loss": 0.2046,
+      "step": 288
+    },
+    {
+      "epoch": 0.3918112798264642,
+      "grad_norm": 0.09974413365125656,
+      "learning_rate": 0.0001,
+      "loss": 0.1996,
+      "step": 289
+    },
+    {
+      "epoch": 0.39316702819956617,
+      "grad_norm": 0.11363832652568817,
+      "learning_rate": 0.0001,
+      "loss": 0.2183,
+      "step": 290
+    },
+    {
+      "epoch": 0.3945227765726681,
+      "grad_norm": 0.12176577746868134,
+      "learning_rate": 0.0001,
+      "loss": 0.225,
+      "step": 291
+    },
+    {
+      "epoch": 0.3958785249457701,
+      "grad_norm": 0.11985570192337036,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 292
+    },
+    {
+      "epoch": 0.397234273318872,
+      "grad_norm": 0.10016416013240814,
+      "learning_rate": 0.0001,
+      "loss": 0.2194,
+      "step": 293
+    },
+    {
+      "epoch": 0.398590021691974,
+      "grad_norm": 0.10597855597734451,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 294
+    },
+    {
+      "epoch": 0.3999457700650759,
+      "grad_norm": 0.11846265941858292,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 295
+    },
+    {
+      "epoch": 0.40130151843817785,
+      "grad_norm": 0.11736635863780975,
+      "learning_rate": 0.0001,
+      "loss": 0.2311,
+      "step": 296
+    },
+    {
+      "epoch": 0.40265726681127983,
+      "grad_norm": 0.12369747459888458,
+      "learning_rate": 0.0001,
+      "loss": 0.2198,
+      "step": 297
+    },
+    {
+      "epoch": 0.40401301518438176,
+      "grad_norm": 0.1182849109172821,
+      "learning_rate": 0.0001,
+      "loss": 0.2325,
+      "step": 298
+    },
+    {
+      "epoch": 0.40536876355748375,
+      "grad_norm": 0.12920567393302917,
+      "learning_rate": 0.0001,
+      "loss": 0.2108,
+      "step": 299
+    },
+    {
+      "epoch": 0.4067245119305857,
+      "grad_norm": 0.13568568229675293,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 300
+    },
+    {
+      "epoch": 0.40808026030368766,
+      "grad_norm": 0.10643009841442108,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 301
+    },
+    {
+      "epoch": 0.4094360086767896,
+      "grad_norm": 0.14213237166404724,
+      "learning_rate": 0.0001,
+      "loss": 0.2165,
+      "step": 302
+    },
+    {
+      "epoch": 0.4107917570498915,
+      "grad_norm": 0.12988242506980896,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 303
+    },
+    {
+      "epoch": 0.4121475054229935,
+      "grad_norm": 0.12043695151805878,
+      "learning_rate": 0.0001,
+      "loss": 0.2224,
+      "step": 304
+    },
+    {
+      "epoch": 0.4135032537960954,
+      "grad_norm": 0.12304119020700455,
+      "learning_rate": 0.0001,
+      "loss": 0.2187,
+      "step": 305
+    },
+    {
+      "epoch": 0.4148590021691974,
+      "grad_norm": 0.17552980780601501,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 306
+    },
+    {
+      "epoch": 0.41621475054229934,
+      "grad_norm": 0.13057579100131989,
+      "learning_rate": 0.0001,
+      "loss": 0.2058,
+      "step": 307
+    },
+    {
+      "epoch": 0.4175704989154013,
+      "grad_norm": 0.10953882336616516,
+      "learning_rate": 0.0001,
+      "loss": 0.2158,
+      "step": 308
+    },
+    {
+      "epoch": 0.41892624728850325,
+      "grad_norm": 0.10787948966026306,
+      "learning_rate": 0.0001,
+      "loss": 0.2359,
+      "step": 309
+    },
+    {
+      "epoch": 0.4202819956616052,
+      "grad_norm": 0.11360488086938858,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 310
+    },
+    {
+      "epoch": 0.42163774403470716,
+      "grad_norm": 0.12294072657823563,
+      "learning_rate": 0.0001,
+      "loss": 0.2249,
+      "step": 311
+    },
+    {
+      "epoch": 0.4229934924078091,
+      "grad_norm": 0.1099947914481163,
+      "learning_rate": 0.0001,
+      "loss": 0.218,
+      "step": 312
+    },
+    {
+      "epoch": 0.4243492407809111,
+      "grad_norm": 0.10253001004457474,
+      "learning_rate": 0.0001,
+      "loss": 0.2146,
+      "step": 313
+    },
+    {
+      "epoch": 0.425704989154013,
+      "grad_norm": 0.1293550580739975,
+      "learning_rate": 0.0001,
+      "loss": 0.2162,
+      "step": 314
+    },
+    {
+      "epoch": 0.427060737527115,
+      "grad_norm": 0.1126367375254631,
+      "learning_rate": 0.0001,
+      "loss": 0.2277,
+      "step": 315
+    },
+    {
+      "epoch": 0.4284164859002169,
+      "grad_norm": 0.12729854881763458,
+      "learning_rate": 0.0001,
+      "loss": 0.22,
+      "step": 316
+    },
+    {
+      "epoch": 0.4297722342733189,
+      "grad_norm": 0.12628163397312164,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 317
+    },
+    {
+      "epoch": 0.4311279826464208,
+      "grad_norm": 0.11287037283182144,
+      "learning_rate": 0.0001,
+      "loss": 0.2269,
+      "step": 318
+    },
+    {
+      "epoch": 0.43248373101952275,
+      "grad_norm": 0.18584799766540527,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 319
+    },
+    {
+      "epoch": 0.43383947939262474,
+      "grad_norm": 0.11298429220914841,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 320
+    },
+    {
+      "epoch": 0.43519522776572667,
+      "grad_norm": 0.10693620890378952,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 321
+    },
+    {
+      "epoch": 0.43655097613882865,
+      "grad_norm": 0.0995495617389679,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 322
+    },
+    {
+      "epoch": 0.4379067245119306,
+      "grad_norm": 0.09863682091236115,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 323
+    },
+    {
+      "epoch": 0.43926247288503256,
+      "grad_norm": 0.09735404700040817,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 324
+    },
+    {
+      "epoch": 0.4406182212581345,
+      "grad_norm": 0.11333636194467545,
+      "learning_rate": 0.0001,
+      "loss": 0.2271,
+      "step": 325
+    },
+    {
+      "epoch": 0.4419739696312364,
+      "grad_norm": 0.0983201265335083,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 326
+    },
+    {
+      "epoch": 0.4433297180043384,
+      "grad_norm": 0.10815490037202835,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 327
+    },
+    {
+      "epoch": 0.44468546637744033,
+      "grad_norm": 0.11426856368780136,
+      "learning_rate": 0.0001,
+      "loss": 0.2218,
+      "step": 328
+    },
+    {
+      "epoch": 0.4460412147505423,
+      "grad_norm": 0.12337986379861832,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 329
+    },
+    {
+      "epoch": 0.44739696312364424,
+      "grad_norm": 0.1043037623167038,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 330
+    },
+    {
+      "epoch": 0.4487527114967462,
+      "grad_norm": 0.1006832867860794,
+      "learning_rate": 0.0001,
+      "loss": 0.2135,
+      "step": 331
+    },
+    {
+      "epoch": 0.45010845986984815,
+      "grad_norm": 0.11578216403722763,
+      "learning_rate": 0.0001,
+      "loss": 0.2105,
+      "step": 332
+    },
+    {
+      "epoch": 0.4514642082429501,
+      "grad_norm": 0.11663608998060226,
+      "learning_rate": 0.0001,
+      "loss": 0.214,
+      "step": 333
+    },
+    {
+      "epoch": 0.45281995661605207,
+      "grad_norm": 0.10902926325798035,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 334
+    },
+    {
+      "epoch": 0.454175704989154,
+      "grad_norm": 0.1364833116531372,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 335
+    },
+    {
+      "epoch": 0.455531453362256,
+      "grad_norm": 0.10804006457328796,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 336
+    },
+    {
+      "epoch": 0.4568872017353579,
+      "grad_norm": 0.10479926317930222,
+      "learning_rate": 0.0001,
+      "loss": 0.2167,
+      "step": 337
+    },
+    {
+      "epoch": 0.4582429501084599,
+      "grad_norm": 0.12120767682790756,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 338
+    },
+    {
+      "epoch": 0.4595986984815618,
+      "grad_norm": 0.11616289615631104,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 339
+    },
+    {
+      "epoch": 0.4609544468546638,
+      "grad_norm": 0.12456631660461426,
+      "learning_rate": 0.0001,
+      "loss": 0.2143,
+      "step": 340
+    },
+    {
+      "epoch": 0.46231019522776573,
+      "grad_norm": 0.11844991147518158,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 341
+    },
+    {
+      "epoch": 0.46366594360086766,
+      "grad_norm": 0.10664281249046326,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 342
+    },
+    {
+      "epoch": 0.46502169197396964,
+      "grad_norm": 0.1413905918598175,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 343
+    },
+    {
+      "epoch": 0.46637744034707157,
+      "grad_norm": 0.11242861300706863,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 344
+    },
+    {
+      "epoch": 0.46773318872017355,
+      "grad_norm": 0.1394144594669342,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 345
+    },
+    {
+      "epoch": 0.4690889370932755,
+      "grad_norm": 0.1527680903673172,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 346
+    },
+    {
+      "epoch": 0.47044468546637747,
+      "grad_norm": 0.11193721741437912,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 347
+    },
+    {
+      "epoch": 0.4718004338394794,
+      "grad_norm": 0.11452329903841019,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 348
+    },
+    {
+      "epoch": 0.4731561822125813,
+      "grad_norm": 0.14150890707969666,
+      "learning_rate": 0.0001,
+      "loss": 0.218,
+      "step": 349
+    },
+    {
+      "epoch": 0.4745119305856833,
+      "grad_norm": 0.12915198504924774,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 350
+    },
+    {
+      "epoch": 0.47586767895878523,
+      "grad_norm": 0.1374678611755371,
+      "learning_rate": 0.0001,
+      "loss": 0.2401,
+      "step": 351
+    },
+    {
+      "epoch": 0.4772234273318872,
+      "grad_norm": 0.10585629940032959,
+      "learning_rate": 0.0001,
+      "loss": 0.1876,
+      "step": 352
+    },
+    {
+      "epoch": 0.47857917570498915,
+      "grad_norm": 0.10894684493541718,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 353
+    },
+    {
+      "epoch": 0.47993492407809113,
+      "grad_norm": 0.12032096087932587,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 354
+    },
+    {
+      "epoch": 0.48129067245119306,
+      "grad_norm": 0.09291663020849228,
+      "learning_rate": 0.0001,
+      "loss": 0.1955,
+      "step": 355
+    },
+    {
+      "epoch": 0.482646420824295,
+      "grad_norm": 0.10697571188211441,
+      "learning_rate": 0.0001,
+      "loss": 0.2111,
+      "step": 356
+    },
+    {
+      "epoch": 0.48400216919739697,
+      "grad_norm": 0.14519456028938293,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 357
+    },
+    {
+      "epoch": 0.4853579175704989,
+      "grad_norm": 0.13694286346435547,
+      "learning_rate": 0.0001,
+      "loss": 0.2242,
+      "step": 358
+    },
+    {
+      "epoch": 0.4867136659436009,
+      "grad_norm": 0.11500401794910431,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 359
+    },
+    {
+      "epoch": 0.4880694143167028,
+      "grad_norm": 0.10939672589302063,
+      "learning_rate": 0.0001,
+      "loss": 0.1963,
+      "step": 360
+    },
+    {
+      "epoch": 0.4894251626898048,
+      "grad_norm": 0.12584058940410614,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 361
+    },
+    {
+      "epoch": 0.4907809110629067,
+      "grad_norm": 0.10783881694078445,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 362
+    },
+    {
+      "epoch": 0.49213665943600865,
+      "grad_norm": 0.11246507614850998,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 363
+    },
+    {
+      "epoch": 0.49349240780911063,
+      "grad_norm": 0.11833275854587555,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 364
+    },
+    {
+      "epoch": 0.49484815618221256,
+      "grad_norm": 0.11138807237148285,
+      "learning_rate": 0.0001,
+      "loss": 0.2264,
+      "step": 365
+    },
+    {
+      "epoch": 0.49620390455531455,
+      "grad_norm": 0.09807030856609344,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 366
+    },
+    {
+      "epoch": 0.4975596529284165,
+      "grad_norm": 0.13736803829669952,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 367
+    },
+    {
+      "epoch": 0.49891540130151846,
+      "grad_norm": 0.12985488772392273,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 368
+    },
+    {
+      "epoch": 0.5002711496746204,
+      "grad_norm": 0.10536885261535645,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 369
+    },
+    {
+      "epoch": 0.5016268980477223,
+      "grad_norm": 0.12244919687509537,
+      "learning_rate": 0.0001,
+      "loss": 0.2114,
+      "step": 370
+    },
+    {
+      "epoch": 0.5029826464208242,
+      "grad_norm": 0.10150042921304703,
+      "learning_rate": 0.0001,
+      "loss": 0.2114,
+      "step": 371
+    },
+    {
+      "epoch": 0.5043383947939263,
+      "grad_norm": 0.12140253186225891,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 372
+    },
+    {
+      "epoch": 0.5056941431670282,
+      "grad_norm": 0.12637700140476227,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 373
+    },
+    {
+      "epoch": 0.5070498915401301,
+      "grad_norm": 0.13199326395988464,
+      "learning_rate": 0.0001,
+      "loss": 0.2281,
+      "step": 374
+    },
+    {
+      "epoch": 0.5084056399132321,
+      "grad_norm": 0.11179718375205994,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 375
+    },
+    {
+      "epoch": 0.5097613882863341,
+      "grad_norm": 0.10628020018339157,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 376
+    },
+    {
+      "epoch": 0.511117136659436,
+      "grad_norm": 0.10334998369216919,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 377
+    },
+    {
+      "epoch": 0.512472885032538,
+      "grad_norm": 0.10926699638366699,
+      "learning_rate": 0.0001,
+      "loss": 0.2105,
+      "step": 378
+    },
+    {
+      "epoch": 0.5138286334056399,
+      "grad_norm": 0.11310454457998276,
+      "learning_rate": 0.0001,
+      "loss": 0.2067,
+      "step": 379
+    },
+    {
+      "epoch": 0.5151843817787418,
+      "grad_norm": 0.13486787676811218,
+      "learning_rate": 0.0001,
+      "loss": 0.2288,
+      "step": 380
+    },
+    {
+      "epoch": 0.5165401301518439,
+      "grad_norm": 0.11885882169008255,
+      "learning_rate": 0.0001,
+      "loss": 0.2225,
+      "step": 381
+    },
+    {
+      "epoch": 0.5178958785249458,
+      "grad_norm": 0.09247157722711563,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 382
+    },
+    {
+      "epoch": 0.5192516268980477,
+      "grad_norm": 0.14512018859386444,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 383
+    },
+    {
+      "epoch": 0.5206073752711496,
+      "grad_norm": 0.15175794064998627,
+      "learning_rate": 0.0001,
+      "loss": 0.2193,
+      "step": 384
+    },
+    {
+      "epoch": 0.5219631236442517,
+      "grad_norm": 0.10738217830657959,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 385
+    },
+    {
+      "epoch": 0.5233188720173536,
+      "grad_norm": 0.11284061521291733,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 386
+    },
+    {
+      "epoch": 0.5246746203904555,
+      "grad_norm": 0.13472841680049896,
+      "learning_rate": 0.0001,
+      "loss": 0.2194,
+      "step": 387
+    },
+    {
+      "epoch": 0.5260303687635575,
+      "grad_norm": 0.10424435883760452,
+      "learning_rate": 0.0001,
+      "loss": 0.2214,
+      "step": 388
+    },
+    {
+      "epoch": 0.5273861171366594,
+      "grad_norm": 0.1054137796163559,
+      "learning_rate": 0.0001,
+      "loss": 0.2065,
+      "step": 389
+    },
+    {
+      "epoch": 0.5287418655097614,
+      "grad_norm": 0.10224529355764389,
+      "learning_rate": 0.0001,
+      "loss": 0.2208,
+      "step": 390
+    },
+    {
+      "epoch": 0.5300976138828634,
+      "grad_norm": 0.10680197179317474,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 391
+    },
+    {
+      "epoch": 0.5314533622559653,
+      "grad_norm": 0.10615722090005875,
+      "learning_rate": 0.0001,
+      "loss": 0.2196,
+      "step": 392
+    },
+    {
+      "epoch": 0.5328091106290672,
+      "grad_norm": 0.11012542247772217,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 393
+    },
+    {
+      "epoch": 0.5341648590021691,
+      "grad_norm": 0.1112055554986,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 394
+    },
+    {
+      "epoch": 0.5355206073752712,
+      "grad_norm": 0.11492089927196503,
+      "learning_rate": 0.0001,
+      "loss": 0.2086,
+      "step": 395
+    },
+    {
+      "epoch": 0.5368763557483731,
+      "grad_norm": 0.13781945407390594,
+      "learning_rate": 0.0001,
+      "loss": 0.2156,
+      "step": 396
+    },
+    {
+      "epoch": 0.538232104121475,
+      "grad_norm": 0.1274905502796173,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 397
+    },
+    {
+      "epoch": 0.539587852494577,
+      "grad_norm": 0.13629111647605896,
+      "learning_rate": 0.0001,
+      "loss": 0.231,
+      "step": 398
+    },
+    {
+      "epoch": 0.540943600867679,
+      "grad_norm": 0.10126534104347229,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 399
+    },
+    {
+      "epoch": 0.5422993492407809,
+      "grad_norm": 0.10935080051422119,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 400
+    },
+    {
+      "epoch": 0.5436550976138829,
+      "grad_norm": 0.10874802619218826,
+      "learning_rate": 0.0001,
+      "loss": 0.2027,
+      "step": 401
+    },
+    {
+      "epoch": 0.5450108459869848,
+      "grad_norm": 0.11695010960102081,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 402
+    },
+    {
+      "epoch": 0.5463665943600867,
+      "grad_norm": 0.09789639711380005,
+      "learning_rate": 0.0001,
+      "loss": 0.1984,
+      "step": 403
+    },
+    {
+      "epoch": 0.5477223427331888,
+      "grad_norm": 0.12741917371749878,
+      "learning_rate": 0.0001,
+      "loss": 0.212,
+      "step": 404
+    },
+    {
+      "epoch": 0.5490780911062907,
+      "grad_norm": 0.11735983937978745,
+      "learning_rate": 0.0001,
+      "loss": 0.2311,
+      "step": 405
+    },
+    {
+      "epoch": 0.5504338394793926,
+      "grad_norm": 0.1219860389828682,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 406
+    },
+    {
+      "epoch": 0.5517895878524945,
+      "grad_norm": 0.1132061630487442,
+      "learning_rate": 0.0001,
+      "loss": 0.2081,
+      "step": 407
+    },
+    {
+      "epoch": 0.5531453362255966,
+      "grad_norm": 0.09816819429397583,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 408
+    },
+    {
+      "epoch": 0.5545010845986985,
+      "grad_norm": 0.11391404271125793,
+      "learning_rate": 0.0001,
+      "loss": 0.2335,
+      "step": 409
+    },
+    {
+      "epoch": 0.5558568329718004,
+      "grad_norm": 0.10533520579338074,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 410
+    },
+    {
+      "epoch": 0.5572125813449024,
+      "grad_norm": 0.09347112476825714,
+      "learning_rate": 0.0001,
+      "loss": 0.192,
+      "step": 411
+    },
+    {
+      "epoch": 0.5585683297180043,
+      "grad_norm": 0.1317298859357834,
+      "learning_rate": 0.0001,
+      "loss": 0.2212,
+      "step": 412
+    },
+    {
+      "epoch": 0.5599240780911063,
+      "grad_norm": 0.11398331820964813,
+      "learning_rate": 0.0001,
+      "loss": 0.2034,
+      "step": 413
+    },
+    {
+      "epoch": 0.5612798264642083,
+      "grad_norm": 0.125030055642128,
+      "learning_rate": 0.0001,
+      "loss": 0.2175,
+      "step": 414
+    },
+    {
+      "epoch": 0.5626355748373102,
+      "grad_norm": 0.15805788338184357,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 415
+    },
+    {
+      "epoch": 0.5639913232104121,
+      "grad_norm": 0.12884917855262756,
+      "learning_rate": 0.0001,
+      "loss": 0.2085,
+      "step": 416
+    },
+    {
+      "epoch": 0.565347071583514,
+      "grad_norm": 0.1492464244365692,
+      "learning_rate": 0.0001,
+      "loss": 0.2269,
+      "step": 417
+    },
+    {
+      "epoch": 0.5667028199566161,
+      "grad_norm": 0.1406038999557495,
+      "learning_rate": 0.0001,
+      "loss": 0.2312,
+      "step": 418
+    },
+    {
+      "epoch": 0.568058568329718,
+      "grad_norm": 0.11515452712774277,
+      "learning_rate": 0.0001,
+      "loss": 0.2284,
+      "step": 419
+    },
+    {
+      "epoch": 0.56941431670282,
+      "grad_norm": 0.10425221920013428,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 420
+    },
+    {
+      "epoch": 0.5707700650759219,
+      "grad_norm": 0.12593768537044525,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 421
+    },
+    {
+      "epoch": 0.5721258134490239,
+      "grad_norm": 0.11307656019926071,
+      "learning_rate": 0.0001,
+      "loss": 0.2104,
+      "step": 422
+    },
+    {
+      "epoch": 0.5734815618221258,
+      "grad_norm": 0.11056561022996902,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 423
+    },
+    {
+      "epoch": 0.5748373101952278,
+      "grad_norm": 0.11113639920949936,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 424
+    },
+    {
+      "epoch": 0.5761930585683297,
+      "grad_norm": 0.1026131734251976,
+      "learning_rate": 0.0001,
+      "loss": 0.2249,
+      "step": 425
+    },
+    {
+      "epoch": 0.5775488069414316,
+      "grad_norm": 0.13202933967113495,
+      "learning_rate": 0.0001,
+      "loss": 0.2334,
+      "step": 426
+    },
+    {
+      "epoch": 0.5789045553145337,
+      "grad_norm": 0.10148321092128754,
+      "learning_rate": 0.0001,
+      "loss": 0.2108,
+      "step": 427
+    },
+    {
+      "epoch": 0.5802603036876356,
+      "grad_norm": 0.10881955176591873,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 428
+    },
+    {
+      "epoch": 0.5816160520607375,
+      "grad_norm": 0.10660913586616516,
+      "learning_rate": 0.0001,
+      "loss": 0.2071,
+      "step": 429
+    },
+    {
+      "epoch": 0.5829718004338394,
+      "grad_norm": 0.0939670130610466,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 430
+    },
+    {
+      "epoch": 0.5843275488069414,
+      "grad_norm": 0.10651350766420364,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 431
+    },
+    {
+      "epoch": 0.5856832971800434,
+      "grad_norm": 0.10108950734138489,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 432
+    },
+    {
+      "epoch": 0.5870390455531453,
+      "grad_norm": 0.12275184690952301,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 433
+    },
+    {
+      "epoch": 0.5883947939262473,
+      "grad_norm": 0.169072687625885,
+      "learning_rate": 0.0001,
+      "loss": 0.2056,
+      "step": 434
+    },
+    {
+      "epoch": 0.5897505422993492,
+      "grad_norm": 0.10997328162193298,
+      "learning_rate": 0.0001,
+      "loss": 0.2005,
+      "step": 435
+    },
+    {
+      "epoch": 0.5911062906724512,
+      "grad_norm": 0.10794641077518463,
+      "learning_rate": 0.0001,
+      "loss": 0.2135,
+      "step": 436
+    },
+    {
+      "epoch": 0.5924620390455532,
+      "grad_norm": 0.10882452875375748,
+      "learning_rate": 0.0001,
+      "loss": 0.2006,
+      "step": 437
+    },
+    {
+      "epoch": 0.5938177874186551,
+      "grad_norm": 0.11713463813066483,
+      "learning_rate": 0.0001,
+      "loss": 0.2191,
+      "step": 438
+    },
+    {
+      "epoch": 0.595173535791757,
+      "grad_norm": 0.12042079865932465,
+      "learning_rate": 0.0001,
+      "loss": 0.2266,
+      "step": 439
+    },
+    {
+      "epoch": 0.596529284164859,
+      "grad_norm": 0.10821802914142609,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 440
+    },
+    {
+      "epoch": 0.597885032537961,
+      "grad_norm": 0.10769280791282654,
+      "learning_rate": 0.0001,
+      "loss": 0.2195,
+      "step": 441
+    },
+    {
+      "epoch": 0.5992407809110629,
+      "grad_norm": 0.11067578196525574,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 442
+    },
+    {
+      "epoch": 0.6005965292841648,
+      "grad_norm": 0.11883726716041565,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 443
+    },
+    {
+      "epoch": 0.6019522776572668,
+      "grad_norm": 0.1317620873451233,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 444
+    },
+    {
+      "epoch": 0.6033080260303688,
+      "grad_norm": 0.10735867917537689,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 445
+    },
+    {
+      "epoch": 0.6046637744034707,
+      "grad_norm": 0.10964839160442352,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 446
+    },
+    {
+      "epoch": 0.6060195227765727,
+      "grad_norm": 0.1039125844836235,
+      "learning_rate": 0.0001,
+      "loss": 0.2175,
+      "step": 447
+    },
+    {
+      "epoch": 0.6073752711496746,
+      "grad_norm": 0.1113983616232872,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 448
+    },
+    {
+      "epoch": 0.6087310195227765,
+      "grad_norm": 0.11929844319820404,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 449
+    },
+    {
+      "epoch": 0.6100867678958786,
+      "grad_norm": 0.17209748923778534,
+      "learning_rate": 0.0001,
+      "loss": 0.2297,
+      "step": 450
+    },
+    {
+      "epoch": 0.6114425162689805,
+      "grad_norm": 0.11573749035596848,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 451
+    },
+    {
+      "epoch": 0.6127982646420824,
+      "grad_norm": 0.11317857354879379,
+      "learning_rate": 0.0001,
+      "loss": 0.2251,
+      "step": 452
+    },
+    {
+      "epoch": 0.6141540130151844,
+      "grad_norm": 0.12213891744613647,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 453
+    },
+    {
+      "epoch": 0.6155097613882863,
+      "grad_norm": 0.1496562361717224,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 454
+    },
+    {
+      "epoch": 0.6168655097613883,
+      "grad_norm": 0.1463765799999237,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 455
+    },
+    {
+      "epoch": 0.6182212581344902,
+      "grad_norm": 0.18038775026798248,
+      "learning_rate": 0.0001,
+      "loss": 0.2107,
+      "step": 456
+    },
+    {
+      "epoch": 0.6195770065075922,
+      "grad_norm": 0.11656726151704788,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 457
+    },
+    {
+      "epoch": 0.6209327548806941,
+      "grad_norm": 0.14198382198810577,
+      "learning_rate": 0.0001,
+      "loss": 0.2208,
+      "step": 458
+    },
+    {
+      "epoch": 0.6222885032537961,
+      "grad_norm": 0.08817528933286667,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 459
+    },
+    {
+      "epoch": 0.6236442516268981,
+      "grad_norm": 0.11271339654922485,
+      "learning_rate": 0.0001,
+      "loss": 0.23,
+      "step": 460
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.11104582995176315,
+      "learning_rate": 0.0001,
+      "loss": 0.2165,
+      "step": 461
+    },
+    {
+      "epoch": 0.6263557483731019,
+      "grad_norm": 0.10129480063915253,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 462
+    },
+    {
+      "epoch": 0.6277114967462039,
+      "grad_norm": 0.13278837502002716,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "step": 463
+    },
+    {
+      "epoch": 0.6290672451193059,
+      "grad_norm": 0.11076650768518448,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 464
+    },
+    {
+      "epoch": 0.6304229934924078,
+      "grad_norm": 0.0935707837343216,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 465
+    },
+    {
+      "epoch": 0.6317787418655098,
+      "grad_norm": 0.1178593710064888,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 466
+    },
+    {
+      "epoch": 0.6331344902386117,
+      "grad_norm": 0.11166414618492126,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 467
+    },
+    {
+      "epoch": 0.6344902386117137,
+      "grad_norm": 0.11557378619909286,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 468
+    },
+    {
+      "epoch": 0.6358459869848156,
+      "grad_norm": 0.11286355555057526,
+      "learning_rate": 0.0001,
+      "loss": 0.2078,
+      "step": 469
+    },
+    {
+      "epoch": 0.6372017353579176,
+      "grad_norm": 0.11957176774740219,
+      "learning_rate": 0.0001,
+      "loss": 0.2102,
+      "step": 470
+    },
+    {
+      "epoch": 0.6385574837310195,
+      "grad_norm": 0.09912893176078796,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 471
+    },
+    {
+      "epoch": 0.6399132321041214,
+      "grad_norm": 0.1189800277352333,
+      "learning_rate": 0.0001,
+      "loss": 0.2285,
+      "step": 472
+    },
+    {
+      "epoch": 0.6412689804772235,
+      "grad_norm": 0.11107879132032394,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 473
+    },
+    {
+      "epoch": 0.6426247288503254,
+      "grad_norm": 0.1265309900045395,
+      "learning_rate": 0.0001,
+      "loss": 0.2086,
+      "step": 474
+    },
+    {
+      "epoch": 0.6439804772234273,
+      "grad_norm": 0.12257790565490723,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 475
+    },
+    {
+      "epoch": 0.6453362255965293,
+      "grad_norm": 0.10021594911813736,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 476
+    },
+    {
+      "epoch": 0.6466919739696312,
+      "grad_norm": 0.16067621111869812,
+      "learning_rate": 0.0001,
+      "loss": 0.202,
+      "step": 477
+    },
+    {
+      "epoch": 0.6480477223427332,
+      "grad_norm": 0.10679979622364044,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 478
+    },
+    {
+      "epoch": 0.6494034707158352,
+      "grad_norm": 0.11437883228063583,
+      "learning_rate": 0.0001,
+      "loss": 0.225,
+      "step": 479
+    },
+    {
+      "epoch": 0.6507592190889371,
+      "grad_norm": 0.11093738675117493,
+      "learning_rate": 0.0001,
+      "loss": 0.2056,
+      "step": 480
+    },
+    {
+      "epoch": 0.652114967462039,
+      "grad_norm": 0.0941971093416214,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 481
+    },
+    {
+      "epoch": 0.653470715835141,
+      "grad_norm": 0.1056419312953949,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 482
+    },
+    {
+      "epoch": 0.654826464208243,
+      "grad_norm": 0.12441356480121613,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 483
+    },
+    {
+      "epoch": 0.6561822125813449,
+      "grad_norm": 0.1098334789276123,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 484
+    },
+    {
+      "epoch": 0.6575379609544468,
+      "grad_norm": 0.08955611288547516,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 485
+    },
+    {
+      "epoch": 0.6588937093275488,
+      "grad_norm": 0.10721218585968018,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 486
+    },
+    {
+      "epoch": 0.6602494577006508,
+      "grad_norm": 0.09911365807056427,
+      "learning_rate": 0.0001,
+      "loss": 0.2141,
+      "step": 487
+    },
+    {
+      "epoch": 0.6616052060737527,
+      "grad_norm": 0.1175253614783287,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 488
+    },
+    {
+      "epoch": 0.6629609544468547,
+      "grad_norm": 0.09751195460557938,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 489
+    },
+    {
+      "epoch": 0.6643167028199566,
+      "grad_norm": 0.11153285205364227,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 490
+    },
+    {
+      "epoch": 0.6656724511930586,
+      "grad_norm": 0.13056796789169312,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 491
+    },
+    {
+      "epoch": 0.6670281995661606,
+      "grad_norm": 0.1103479191660881,
+      "learning_rate": 0.0001,
+      "loss": 0.2131,
+      "step": 492
+    },
+    {
+      "epoch": 0.6683839479392625,
+      "grad_norm": 0.11970798671245575,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 493
+    },
+    {
+      "epoch": 0.6697396963123644,
+      "grad_norm": 0.12085957825183868,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 494
+    },
+    {
+      "epoch": 0.6710954446854663,
+      "grad_norm": 0.1281101107597351,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 495
+    },
+    {
+      "epoch": 0.6724511930585684,
+      "grad_norm": 0.14826293289661407,
+      "learning_rate": 0.0001,
+      "loss": 0.2188,
+      "step": 496
+    },
+    {
+      "epoch": 0.6738069414316703,
+      "grad_norm": 0.13201726973056793,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 497
+    },
+    {
+      "epoch": 0.6751626898047722,
+      "grad_norm": 0.10742480307817459,
+      "learning_rate": 0.0001,
+      "loss": 0.2115,
+      "step": 498
+    },
+    {
+      "epoch": 0.6765184381778742,
+      "grad_norm": 0.11474528163671494,
+      "learning_rate": 0.0001,
+      "loss": 0.2259,
+      "step": 499
+    },
+    {
+      "epoch": 0.6778741865509761,
+      "grad_norm": 0.12335482984781265,
+      "learning_rate": 0.0001,
+      "loss": 0.2091,
+      "step": 500
+    },
+    {
+      "epoch": 0.6778741865509761,
+      "eval_loss": 0.21124856173992157,
+      "eval_runtime": 585.9568,
+      "eval_samples_per_second": 4.72,
+      "eval_steps_per_second": 1.181,
+      "step": 500
+    },
+    {
+      "epoch": 0.6792299349240781,
+      "grad_norm": 0.109441377222538,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 501
+    },
+    {
+      "epoch": 0.68058568329718,
+      "grad_norm": 0.10464882850646973,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 502
+    },
+    {
+      "epoch": 0.681941431670282,
+      "grad_norm": 0.10190745443105698,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 503
+    },
+    {
+      "epoch": 0.6832971800433839,
+      "grad_norm": 0.11795409023761749,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 504
+    },
+    {
+      "epoch": 0.684652928416486,
+      "grad_norm": 0.10550260543823242,
+      "learning_rate": 0.0001,
+      "loss": 0.1982,
+      "step": 505
+    },
+    {
+      "epoch": 0.6860086767895879,
+      "grad_norm": 0.12333224713802338,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 506
+    },
+    {
+      "epoch": 0.6873644251626898,
+      "grad_norm": 0.09856431186199188,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 507
+    },
+    {
+      "epoch": 0.6887201735357917,
+      "grad_norm": 0.10570758581161499,
+      "learning_rate": 0.0001,
+      "loss": 0.2139,
+      "step": 508
+    },
+    {
+      "epoch": 0.6900759219088937,
+      "grad_norm": 0.11989443004131317,
+      "learning_rate": 0.0001,
+      "loss": 0.2106,
+      "step": 509
+    },
+    {
+      "epoch": 0.6914316702819957,
+      "grad_norm": 0.12116160988807678,
+      "learning_rate": 0.0001,
+      "loss": 0.2355,
+      "step": 510
+    },
+    {
+      "epoch": 0.6927874186550976,
+      "grad_norm": 0.11081567406654358,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 511
+    },
+    {
+      "epoch": 0.6941431670281996,
+      "grad_norm": 0.10212654620409012,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 512
+    },
+    {
+      "epoch": 0.6954989154013015,
+      "grad_norm": 0.11284127831459045,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 513
+    },
+    {
+      "epoch": 0.6968546637744034,
+      "grad_norm": 0.11569730192422867,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 514
+    },
+    {
+      "epoch": 0.6982104121475055,
+      "grad_norm": 0.1060398742556572,
+      "learning_rate": 0.0001,
+      "loss": 0.2107,
+      "step": 515
+    },
+    {
+      "epoch": 0.6995661605206074,
+      "grad_norm": 0.112233966588974,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 516
+    },
+    {
+      "epoch": 0.7009219088937093,
+      "grad_norm": 0.11053170263767242,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 517
+    },
+    {
+      "epoch": 0.7022776572668112,
+      "grad_norm": 0.10988939553499222,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 518
+    },
+    {
+      "epoch": 0.7036334056399133,
+      "grad_norm": 0.10733780264854431,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 519
+    },
+    {
+      "epoch": 0.7049891540130152,
+      "grad_norm": 0.0985184833407402,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 520
+    },
+    {
+      "epoch": 0.7063449023861171,
+      "grad_norm": 0.09226637333631516,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 521
+    },
+    {
+      "epoch": 0.7077006507592191,
+      "grad_norm": 0.09949634969234467,
+      "learning_rate": 0.0001,
+      "loss": 0.2243,
+      "step": 522
+    },
+    {
+      "epoch": 0.709056399132321,
+      "grad_norm": 0.10085625946521759,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 523
+    },
+    {
+      "epoch": 0.710412147505423,
+      "grad_norm": 0.10137657076120377,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 524
+    },
+    {
+      "epoch": 0.711767895878525,
+      "grad_norm": 0.09197983890771866,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 525
+    },
+    {
+      "epoch": 0.7131236442516269,
+      "grad_norm": 0.168605238199234,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 526
+    },
+    {
+      "epoch": 0.7144793926247288,
+      "grad_norm": 0.11544493585824966,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 527
+    },
+    {
+      "epoch": 0.7158351409978309,
+      "grad_norm": 0.10736843198537827,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 528
+    },
+    {
+      "epoch": 0.7171908893709328,
+      "grad_norm": 0.10598525404930115,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 529
+    },
+    {
+      "epoch": 0.7185466377440347,
+      "grad_norm": 0.12368134409189224,
+      "learning_rate": 0.0001,
+      "loss": 0.2141,
+      "step": 530
+    },
+    {
+      "epoch": 0.7199023861171366,
+      "grad_norm": 0.12551409006118774,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 531
+    },
+    {
+      "epoch": 0.7212581344902386,
+      "grad_norm": 0.10292831063270569,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 532
+    },
+    {
+      "epoch": 0.7226138828633406,
+      "grad_norm": 0.11200780421495438,
+      "learning_rate": 0.0001,
+      "loss": 0.2114,
+      "step": 533
+    },
+    {
+      "epoch": 0.7239696312364425,
+      "grad_norm": 0.13888825476169586,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 534
+    },
+    {
+      "epoch": 0.7253253796095445,
+      "grad_norm": 0.11513497680425644,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 535
+    },
+    {
+      "epoch": 0.7266811279826464,
+      "grad_norm": 0.10387038439512253,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 536
+    },
+    {
+      "epoch": 0.7280368763557483,
+      "grad_norm": 0.11427624523639679,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 537
+    },
+    {
+      "epoch": 0.7293926247288504,
+      "grad_norm": 0.18002212047576904,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 538
+    },
+    {
+      "epoch": 0.7307483731019523,
+      "grad_norm": 0.10382217913866043,
+      "learning_rate": 0.0001,
+      "loss": 0.2008,
+      "step": 539
+    },
+    {
+      "epoch": 0.7321041214750542,
+      "grad_norm": 0.17331746220588684,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 540
+    },
+    {
+      "epoch": 0.7334598698481561,
+      "grad_norm": 0.15686853229999542,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 541
+    },
+    {
+      "epoch": 0.7348156182212582,
+      "grad_norm": 0.13423365354537964,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 542
+    },
+    {
+      "epoch": 0.7361713665943601,
+      "grad_norm": 0.11132203787565231,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 543
+    },
+    {
+      "epoch": 0.737527114967462,
+      "grad_norm": 0.13134092092514038,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 544
+    },
+    {
+      "epoch": 0.738882863340564,
+      "grad_norm": 0.11013273149728775,
+      "learning_rate": 0.0001,
+      "loss": 0.2208,
+      "step": 545
+    },
+    {
+      "epoch": 0.7402386117136659,
+      "grad_norm": 0.1064901277422905,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 546
+    },
+    {
+      "epoch": 0.7415943600867679,
+      "grad_norm": 0.1148400530219078,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 547
+    },
+    {
+      "epoch": 0.7429501084598699,
+      "grad_norm": 0.12401867657899857,
+      "learning_rate": 0.0001,
+      "loss": 0.211,
+      "step": 548
+    },
+    {
+      "epoch": 0.7443058568329718,
+      "grad_norm": 0.10509403795003891,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 549
+    },
+    {
+      "epoch": 0.7456616052060737,
+      "grad_norm": 0.133061483502388,
+      "learning_rate": 0.0001,
+      "loss": 0.2313,
+      "step": 550
+    },
+    {
+      "epoch": 0.7470173535791758,
+      "grad_norm": 0.10771442204713821,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 551
+    },
+    {
+      "epoch": 0.7483731019522777,
+      "grad_norm": 0.10763978958129883,
+      "learning_rate": 0.0001,
+      "loss": 0.2196,
+      "step": 552
+    },
+    {
+      "epoch": 0.7497288503253796,
+      "grad_norm": 0.1287078708410263,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 553
+    },
+    {
+      "epoch": 0.7510845986984815,
+      "grad_norm": 0.10799828916788101,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "step": 554
+    },
+    {
+      "epoch": 0.7524403470715835,
+      "grad_norm": 0.24369190633296967,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 555
+    },
+    {
+      "epoch": 0.7537960954446855,
+      "grad_norm": 0.10731006413698196,
+      "learning_rate": 0.0001,
+      "loss": 0.2075,
+      "step": 556
+    },
+    {
+      "epoch": 0.7551518438177874,
+      "grad_norm": 0.11853396892547607,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 557
+    },
+    {
+      "epoch": 0.7565075921908894,
+      "grad_norm": 0.1504039168357849,
+      "learning_rate": 0.0001,
+      "loss": 0.2241,
+      "step": 558
+    },
+    {
+      "epoch": 0.7578633405639913,
+      "grad_norm": 0.11940452456474304,
+      "learning_rate": 0.0001,
+      "loss": 0.2098,
+      "step": 559
+    },
+    {
+      "epoch": 0.7592190889370932,
+      "grad_norm": 0.10573428124189377,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 560
+    },
+    {
+      "epoch": 0.7605748373101953,
+      "grad_norm": 0.12262751907110214,
+      "learning_rate": 0.0001,
+      "loss": 0.2124,
+      "step": 561
+    },
+    {
+      "epoch": 0.7619305856832972,
+      "grad_norm": 0.10280203074216843,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 562
+    },
+    {
+      "epoch": 0.7632863340563991,
+      "grad_norm": 0.13791029155254364,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 563
+    },
+    {
+      "epoch": 0.764642082429501,
+      "grad_norm": 0.2759683132171631,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 564
+    },
+    {
+      "epoch": 0.7659978308026031,
+      "grad_norm": 0.17915979027748108,
+      "learning_rate": 0.0001,
+      "loss": 0.2034,
+      "step": 565
+    },
+    {
+      "epoch": 0.767353579175705,
+      "grad_norm": 0.1250712275505066,
+      "learning_rate": 0.0001,
+      "loss": 0.2199,
+      "step": 566
+    },
+    {
+      "epoch": 0.7687093275488069,
+      "grad_norm": 0.10158663243055344,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 567
+    },
+    {
+      "epoch": 0.7700650759219089,
+      "grad_norm": 0.1283869743347168,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 568
+    },
+    {
+      "epoch": 0.7714208242950108,
+      "grad_norm": 0.11668034642934799,
+      "learning_rate": 0.0001,
+      "loss": 0.2168,
+      "step": 569
+    },
+    {
+      "epoch": 0.7727765726681128,
+      "grad_norm": 0.18928654491901398,
+      "learning_rate": 0.0001,
+      "loss": 0.2065,
+      "step": 570
+    },
+    {
+      "epoch": 0.7741323210412148,
+      "grad_norm": 0.11028747260570526,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 571
+    },
+    {
+      "epoch": 0.7754880694143167,
+      "grad_norm": 0.11906706541776657,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 572
+    },
+    {
+      "epoch": 0.7768438177874186,
+      "grad_norm": 0.10834264755249023,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 573
+    },
+    {
+      "epoch": 0.7781995661605207,
+      "grad_norm": 0.1415076106786728,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 574
+    },
+    {
+      "epoch": 0.7795553145336226,
+      "grad_norm": 0.122581347823143,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 575
+    },
+    {
+      "epoch": 0.7809110629067245,
+      "grad_norm": 0.16638000309467316,
+      "learning_rate": 0.0001,
+      "loss": 0.2249,
+      "step": 576
+    },
+    {
+      "epoch": 0.7822668112798264,
+      "grad_norm": 0.11046886444091797,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 577
+    },
+    {
+      "epoch": 0.7836225596529284,
+      "grad_norm": 0.10590071976184845,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 578
+    },
+    {
+      "epoch": 0.7849783080260304,
+      "grad_norm": 0.12335824221372604,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 579
+    },
+    {
+      "epoch": 0.7863340563991323,
+      "grad_norm": 0.10231521725654602,
+      "learning_rate": 0.0001,
+      "loss": 0.2179,
+      "step": 580
+    },
+    {
+      "epoch": 0.7876898047722343,
+      "grad_norm": 0.10867033898830414,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 581
+    },
+    {
+      "epoch": 0.7890455531453362,
+      "grad_norm": 0.10097082704305649,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 582
+    },
+    {
+      "epoch": 0.7904013015184381,
+      "grad_norm": 0.10306942462921143,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 583
+    },
+    {
+      "epoch": 0.7917570498915402,
+      "grad_norm": 0.11676366627216339,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 584
+    },
+    {
+      "epoch": 0.7931127982646421,
+      "grad_norm": 0.10839825123548508,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 585
+    },
+    {
+      "epoch": 0.794468546637744,
+      "grad_norm": 0.11621545255184174,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 586
+    },
+    {
+      "epoch": 0.795824295010846,
+      "grad_norm": 0.11346252262592316,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 587
+    },
+    {
+      "epoch": 0.797180043383948,
+      "grad_norm": 0.11613406985998154,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 588
+    },
+    {
+      "epoch": 0.7985357917570499,
+      "grad_norm": 0.11378610879182816,
+      "learning_rate": 0.0001,
+      "loss": 0.2164,
+      "step": 589
+    },
+    {
+      "epoch": 0.7998915401301518,
+      "grad_norm": 0.1336962729692459,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 590
+    },
+    {
+      "epoch": 0.8012472885032538,
+      "grad_norm": 0.12226283550262451,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 591
+    },
+    {
+      "epoch": 0.8026030368763557,
+      "grad_norm": 0.12707912921905518,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 592
+    },
+    {
+      "epoch": 0.8039587852494577,
+      "grad_norm": 0.10723856836557388,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 593
+    },
+    {
+      "epoch": 0.8053145336225597,
+      "grad_norm": 0.12285613268613815,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 594
+    },
+    {
+      "epoch": 0.8066702819956616,
+      "grad_norm": 0.11677780747413635,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 595
+    },
+    {
+      "epoch": 0.8080260303687635,
+      "grad_norm": 0.09618158638477325,
+      "learning_rate": 0.0001,
+      "loss": 0.2045,
+      "step": 596
+    },
+    {
+      "epoch": 0.8093817787418656,
+      "grad_norm": 0.13501878082752228,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 597
+    },
+    {
+      "epoch": 0.8107375271149675,
+      "grad_norm": 0.10393284261226654,
+      "learning_rate": 0.0001,
+      "loss": 0.2131,
+      "step": 598
+    },
+    {
+      "epoch": 0.8120932754880694,
+      "grad_norm": 0.10507247596979141,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 599
+    },
+    {
+      "epoch": 0.8134490238611713,
+      "grad_norm": 0.09469563513994217,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 600
+    },
+    {
+      "epoch": 0.8148047722342733,
+      "grad_norm": 0.11648718267679214,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 601
+    },
+    {
+      "epoch": 0.8161605206073753,
+      "grad_norm": 0.11656969040632248,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 602
+    },
+    {
+      "epoch": 0.8175162689804772,
+      "grad_norm": 0.10681107640266418,
+      "learning_rate": 0.0001,
+      "loss": 0.2194,
+      "step": 603
+    },
+    {
+      "epoch": 0.8188720173535792,
+      "grad_norm": 0.11960645765066147,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 604
+    },
+    {
+      "epoch": 0.8202277657266811,
+      "grad_norm": 0.14047929644584656,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 605
+    },
+    {
+      "epoch": 0.821583514099783,
+      "grad_norm": 0.13034816086292267,
+      "learning_rate": 0.0001,
+      "loss": 0.2171,
+      "step": 606
+    },
+    {
+      "epoch": 0.8229392624728851,
+      "grad_norm": 0.11235146224498749,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 607
+    },
+    {
+      "epoch": 0.824295010845987,
+      "grad_norm": 0.10121841728687286,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 608
+    },
+    {
+      "epoch": 0.8256507592190889,
+      "grad_norm": 0.11959599703550339,
+      "learning_rate": 0.0001,
+      "loss": 0.2009,
+      "step": 609
+    },
+    {
+      "epoch": 0.8270065075921909,
+      "grad_norm": 0.11458362638950348,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 610
+    },
+    {
+      "epoch": 0.8283622559652929,
+      "grad_norm": 0.10209733992815018,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 611
+    },
+    {
+      "epoch": 0.8297180043383948,
+      "grad_norm": 0.11466650664806366,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 612
+    },
+    {
+      "epoch": 0.8310737527114967,
+      "grad_norm": 0.1093161553144455,
+      "learning_rate": 0.0001,
+      "loss": 0.212,
+      "step": 613
+    },
+    {
+      "epoch": 0.8324295010845987,
+      "grad_norm": 0.11256524175405502,
+      "learning_rate": 0.0001,
+      "loss": 0.2221,
+      "step": 614
+    },
+    {
+      "epoch": 0.8337852494577006,
+      "grad_norm": 0.10143551975488663,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 615
+    },
+    {
+      "epoch": 0.8351409978308026,
+      "grad_norm": 0.10671805590391159,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 616
+    },
+    {
+      "epoch": 0.8364967462039046,
+      "grad_norm": 0.11868534982204437,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 617
+    },
+    {
+      "epoch": 0.8378524945770065,
+      "grad_norm": 0.13383151590824127,
+      "learning_rate": 0.0001,
+      "loss": 0.2009,
+      "step": 618
+    },
+    {
+      "epoch": 0.8392082429501084,
+      "grad_norm": 0.14050070941448212,
+      "learning_rate": 0.0001,
+      "loss": 0.2035,
+      "step": 619
+    },
+    {
+      "epoch": 0.8405639913232104,
+      "grad_norm": 0.1119464859366417,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 620
+    },
+    {
+      "epoch": 0.8419197396963124,
+      "grad_norm": 0.13229380548000336,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 621
+    },
+    {
+      "epoch": 0.8432754880694143,
+      "grad_norm": 0.11997786909341812,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 622
+    },
+    {
+      "epoch": 0.8446312364425163,
+      "grad_norm": 0.1286950260400772,
+      "learning_rate": 0.0001,
+      "loss": 0.2059,
+      "step": 623
+    },
+    {
+      "epoch": 0.8459869848156182,
+      "grad_norm": 0.1252814084291458,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 624
+    },
+    {
+      "epoch": 0.8473427331887202,
+      "grad_norm": 0.1247292011976242,
+      "learning_rate": 0.0001,
+      "loss": 0.214,
+      "step": 625
+    },
+    {
+      "epoch": 0.8486984815618221,
+      "grad_norm": 0.1519889086484909,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 626
+    },
+    {
+      "epoch": 0.8500542299349241,
+      "grad_norm": 0.11072339117527008,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 627
+    },
+    {
+      "epoch": 0.851409978308026,
+      "grad_norm": 0.1172877699136734,
+      "learning_rate": 0.0001,
+      "loss": 0.2032,
+      "step": 628
+    },
+    {
+      "epoch": 0.8527657266811279,
+      "grad_norm": 0.1210048496723175,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 629
+    },
+    {
+      "epoch": 0.85412147505423,
+      "grad_norm": 0.1261908859014511,
+      "learning_rate": 0.0001,
+      "loss": 0.2231,
+      "step": 630
+    },
+    {
+      "epoch": 0.8554772234273319,
+      "grad_norm": 0.10483667254447937,
+      "learning_rate": 0.0001,
+      "loss": 0.2073,
+      "step": 631
+    },
+    {
+      "epoch": 0.8568329718004338,
+      "grad_norm": 0.11037805676460266,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 632
+    },
+    {
+      "epoch": 0.8581887201735358,
+      "grad_norm": 0.10755746811628342,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 633
+    },
+    {
+      "epoch": 0.8595444685466378,
+      "grad_norm": 0.11525319516658783,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 634
+    },
+    {
+      "epoch": 0.8609002169197397,
+      "grad_norm": 0.10286539793014526,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 635
+    },
+    {
+      "epoch": 0.8622559652928417,
+      "grad_norm": 0.11918435245752335,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 636
+    },
+    {
+      "epoch": 0.8636117136659436,
+      "grad_norm": 0.11137478798627853,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 637
+    },
+    {
+      "epoch": 0.8649674620390455,
+      "grad_norm": 0.10425526648759842,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 638
+    },
+    {
+      "epoch": 0.8663232104121475,
+      "grad_norm": 0.10617603361606598,
+      "learning_rate": 0.0001,
+      "loss": 0.2009,
+      "step": 639
+    },
+    {
+      "epoch": 0.8676789587852495,
+      "grad_norm": 0.1009998768568039,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 640
+    },
+    {
+      "epoch": 0.8690347071583514,
+      "grad_norm": 0.1240236759185791,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 641
+    },
+    {
+      "epoch": 0.8703904555314533,
+      "grad_norm": 0.10027434676885605,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 642
+    },
+    {
+      "epoch": 0.8717462039045553,
+      "grad_norm": 0.11083299666643143,
+      "learning_rate": 0.0001,
+      "loss": 0.215,
+      "step": 643
+    },
+    {
+      "epoch": 0.8731019522776573,
+      "grad_norm": 0.12559041380882263,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 644
+    },
+    {
+      "epoch": 0.8744577006507592,
+      "grad_norm": 0.10607516020536423,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 645
+    },
+    {
+      "epoch": 0.8758134490238612,
+      "grad_norm": 0.11553547531366348,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 646
+    },
+    {
+      "epoch": 0.8771691973969631,
+      "grad_norm": 0.10643411427736282,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 647
+    },
+    {
+      "epoch": 0.8785249457700651,
+      "grad_norm": 0.1258634328842163,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 648
+    },
+    {
+      "epoch": 0.879880694143167,
+      "grad_norm": 0.10211330652236938,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 649
+    },
+    {
+      "epoch": 0.881236442516269,
+      "grad_norm": 0.11438219994306564,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 650
+    },
+    {
+      "epoch": 0.8825921908893709,
+      "grad_norm": 0.16057232022285461,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 651
+    },
+    {
+      "epoch": 0.8839479392624728,
+      "grad_norm": 0.12260377407073975,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 652
+    },
+    {
+      "epoch": 0.8853036876355749,
+      "grad_norm": 0.10950686037540436,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 653
+    },
+    {
+      "epoch": 0.8866594360086768,
+      "grad_norm": 0.12778618931770325,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 654
+    },
+    {
+      "epoch": 0.8880151843817787,
+      "grad_norm": 0.1354844868183136,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 655
+    },
+    {
+      "epoch": 0.8893709327548807,
+      "grad_norm": 0.10596424341201782,
+      "learning_rate": 0.0001,
+      "loss": 0.1987,
+      "step": 656
+    },
+    {
+      "epoch": 0.8907266811279827,
+      "grad_norm": 0.11473318189382553,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 657
+    },
+    {
+      "epoch": 0.8920824295010846,
+      "grad_norm": 0.11610270291566849,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 658
+    },
+    {
+      "epoch": 0.8934381778741866,
+      "grad_norm": 0.11700058728456497,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 659
+    },
+    {
+      "epoch": 0.8947939262472885,
+      "grad_norm": 0.11252418160438538,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 660
+    },
+    {
+      "epoch": 0.8961496746203904,
+      "grad_norm": 0.1183028444647789,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 661
+    },
+    {
+      "epoch": 0.8975054229934925,
+      "grad_norm": 0.11396466195583344,
+      "learning_rate": 0.0001,
+      "loss": 0.2197,
+      "step": 662
+    },
+    {
+      "epoch": 0.8988611713665944,
+      "grad_norm": 0.10009752959012985,
+      "learning_rate": 0.0001,
+      "loss": 0.2144,
+      "step": 663
+    },
+    {
+      "epoch": 0.9002169197396963,
+      "grad_norm": 0.09940466284751892,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 664
+    },
+    {
+      "epoch": 0.9015726681127982,
+      "grad_norm": 0.11077883839607239,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 665
+    },
+    {
+      "epoch": 0.9029284164859002,
+      "grad_norm": 0.10697755962610245,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 666
+    },
+    {
+      "epoch": 0.9042841648590022,
+      "grad_norm": 0.09719739109277725,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 667
+    },
+    {
+      "epoch": 0.9056399132321041,
+      "grad_norm": 0.13106481730937958,
+      "learning_rate": 0.0001,
+      "loss": 0.2183,
+      "step": 668
+    },
+    {
+      "epoch": 0.9069956616052061,
+      "grad_norm": 0.12257616221904755,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 669
+    },
+    {
+      "epoch": 0.908351409978308,
+      "grad_norm": 0.11231286078691483,
+      "learning_rate": 0.0001,
+      "loss": 0.1958,
+      "step": 670
+    },
+    {
+      "epoch": 0.90970715835141,
+      "grad_norm": 0.12029795348644257,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 671
+    },
+    {
+      "epoch": 0.911062906724512,
+      "grad_norm": 0.10609627515077591,
+      "learning_rate": 0.0001,
+      "loss": 0.1857,
+      "step": 672
+    },
+    {
+      "epoch": 0.9124186550976139,
+      "grad_norm": 0.11634498834609985,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 673
+    },
+    {
+      "epoch": 0.9137744034707158,
+      "grad_norm": 0.11392821371555328,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 674
+    },
+    {
+      "epoch": 0.9151301518438177,
+      "grad_norm": 0.2124888002872467,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 675
+    },
+    {
+      "epoch": 0.9164859002169198,
+      "grad_norm": 0.11601965129375458,
+      "learning_rate": 0.0001,
+      "loss": 0.2161,
+      "step": 676
+    },
+    {
+      "epoch": 0.9178416485900217,
+      "grad_norm": 0.13088494539260864,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 677
+    },
+    {
+      "epoch": 0.9191973969631236,
+      "grad_norm": 0.11482051759958267,
+      "learning_rate": 0.0001,
+      "loss": 0.1953,
+      "step": 678
+    },
+    {
+      "epoch": 0.9205531453362256,
+      "grad_norm": 0.10830912739038467,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 679
+    },
+    {
+      "epoch": 0.9219088937093276,
+      "grad_norm": 0.12834316492080688,
+      "learning_rate": 0.0001,
+      "loss": 0.2173,
+      "step": 680
+    },
+    {
+      "epoch": 0.9232646420824295,
+      "grad_norm": 0.11264733225107193,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 681
+    },
+    {
+      "epoch": 0.9246203904555315,
+      "grad_norm": 0.11361085623502731,
+      "learning_rate": 0.0001,
+      "loss": 0.1875,
+      "step": 682
+    },
+    {
+      "epoch": 0.9259761388286334,
+      "grad_norm": 0.1166103407740593,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 683
+    },
+    {
+      "epoch": 0.9273318872017353,
+      "grad_norm": 0.12536774575710297,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 684
+    },
+    {
+      "epoch": 0.9286876355748374,
+      "grad_norm": 0.1106983870267868,
+      "learning_rate": 0.0001,
+      "loss": 0.1992,
+      "step": 685
+    },
+    {
+      "epoch": 0.9300433839479393,
+      "grad_norm": 0.12899215519428253,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 686
+    },
+    {
+      "epoch": 0.9313991323210412,
+      "grad_norm": 0.12627775967121124,
+      "learning_rate": 0.0001,
+      "loss": 0.1949,
+      "step": 687
+    },
+    {
+      "epoch": 0.9327548806941431,
+      "grad_norm": 0.13036344945430756,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 688
+    },
+    {
+      "epoch": 0.9341106290672451,
+      "grad_norm": 0.11054501682519913,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 689
+    },
+    {
+      "epoch": 0.9354663774403471,
+      "grad_norm": 0.11302312463521957,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 690
+    },
+    {
+      "epoch": 0.936822125813449,
+      "grad_norm": 0.10723387449979782,
+      "learning_rate": 0.0001,
+      "loss": 0.1966,
+      "step": 691
+    },
+    {
+      "epoch": 0.938177874186551,
+      "grad_norm": 0.10424936562776566,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 692
+    },
+    {
+      "epoch": 0.9395336225596529,
+      "grad_norm": 0.10642514377832413,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 693
+    },
+    {
+      "epoch": 0.9408893709327549,
+      "grad_norm": 0.11069341748952866,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 694
+    },
+    {
+      "epoch": 0.9422451193058569,
+      "grad_norm": 0.11237718164920807,
+      "learning_rate": 0.0001,
+      "loss": 0.2204,
+      "step": 695
+    },
+    {
+      "epoch": 0.9436008676789588,
+      "grad_norm": 0.12448445707559586,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 696
+    },
+    {
+      "epoch": 0.9449566160520607,
+      "grad_norm": 0.09546726942062378,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 697
+    },
+    {
+      "epoch": 0.9463123644251626,
+      "grad_norm": 0.11963894963264465,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 698
+    },
+    {
+      "epoch": 0.9476681127982647,
+      "grad_norm": 0.10406365245580673,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 699
+    },
+    {
+      "epoch": 0.9490238611713666,
+      "grad_norm": 0.08631463348865509,
+      "learning_rate": 0.0001,
+      "loss": 0.1895,
+      "step": 700
+    },
+    {
+      "epoch": 0.9503796095444685,
+      "grad_norm": 0.15357080101966858,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 701
+    },
+    {
+      "epoch": 0.9517353579175705,
+      "grad_norm": 0.12241730839014053,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 702
+    },
+    {
+      "epoch": 0.9530911062906724,
+      "grad_norm": 0.11405344307422638,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 703
+    },
+    {
+      "epoch": 0.9544468546637744,
+      "grad_norm": 0.10434224456548691,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 704
+    },
+    {
+      "epoch": 0.9558026030368764,
+      "grad_norm": 0.1137109324336052,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 705
+    },
+    {
+      "epoch": 0.9571583514099783,
+      "grad_norm": 0.11673258244991302,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 706
+    },
+    {
+      "epoch": 0.9585140997830802,
+      "grad_norm": 0.1113823652267456,
+      "learning_rate": 0.0001,
+      "loss": 0.2117,
+      "step": 707
+    },
+    {
+      "epoch": 0.9598698481561823,
+      "grad_norm": 0.11658234149217606,
+      "learning_rate": 0.0001,
+      "loss": 0.1991,
+      "step": 708
+    },
+    {
+      "epoch": 0.9612255965292842,
+      "grad_norm": 0.1245894581079483,
+      "learning_rate": 0.0001,
+      "loss": 0.2052,
+      "step": 709
+    },
+    {
+      "epoch": 0.9625813449023861,
+      "grad_norm": 0.1347448080778122,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 710
+    },
+    {
+      "epoch": 0.963937093275488,
+      "grad_norm": 0.10326442122459412,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 711
+    },
+    {
+      "epoch": 0.96529284164859,
+      "grad_norm": 0.11431329697370529,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 712
+    },
+    {
+      "epoch": 0.966648590021692,
+      "grad_norm": 0.12800860404968262,
+      "learning_rate": 0.0001,
+      "loss": 0.2122,
+      "step": 713
+    },
+    {
+      "epoch": 0.9680043383947939,
+      "grad_norm": 0.12824051082134247,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 714
+    },
+    {
+      "epoch": 0.9693600867678959,
+      "grad_norm": 0.12753504514694214,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 715
+    },
+    {
+      "epoch": 0.9707158351409978,
+      "grad_norm": 0.11565020680427551,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 716
+    },
+    {
+      "epoch": 0.9720715835140998,
+      "grad_norm": 0.12578970193862915,
+      "learning_rate": 0.0001,
+      "loss": 0.2156,
+      "step": 717
+    },
+    {
+      "epoch": 0.9734273318872018,
+      "grad_norm": 0.12153230607509613,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 718
+    },
+    {
+      "epoch": 0.9747830802603037,
+      "grad_norm": 0.12290175259113312,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 719
+    },
+    {
+      "epoch": 0.9761388286334056,
+      "grad_norm": 0.11235173046588898,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 720
+    },
+    {
+      "epoch": 0.9774945770065075,
+      "grad_norm": 0.11295671761035919,
+      "learning_rate": 0.0001,
+      "loss": 0.2043,
+      "step": 721
+    },
+    {
+      "epoch": 0.9788503253796096,
+      "grad_norm": 0.19828207790851593,
+      "learning_rate": 0.0001,
+      "loss": 0.2277,
+      "step": 722
+    },
+    {
+      "epoch": 0.9802060737527115,
+      "grad_norm": 0.11920096725225449,
+      "learning_rate": 0.0001,
+      "loss": 0.2117,
+      "step": 723
+    },
+    {
+      "epoch": 0.9815618221258134,
+      "grad_norm": 0.11569735407829285,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 724
+    },
+    {
+      "epoch": 0.9829175704989154,
+      "grad_norm": 0.11042783409357071,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 725
+    },
+    {
+      "epoch": 0.9842733188720173,
+      "grad_norm": 0.09161633998155594,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 726
+    },
+    {
+      "epoch": 0.9856290672451193,
+      "grad_norm": 0.11520474404096603,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 727
+    },
+    {
+      "epoch": 0.9869848156182213,
+      "grad_norm": 0.1269569993019104,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 728
+    },
+    {
+      "epoch": 0.9883405639913232,
+      "grad_norm": 0.12018048763275146,
+      "learning_rate": 0.0001,
+      "loss": 0.2098,
+      "step": 729
+    },
+    {
+      "epoch": 0.9896963123644251,
+      "grad_norm": 0.0950358584523201,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 730
+    },
+    {
+      "epoch": 0.9910520607375272,
+      "grad_norm": 0.11683353036642075,
+      "learning_rate": 0.0001,
+      "loss": 0.2129,
+      "step": 731
+    },
+    {
+      "epoch": 0.9924078091106291,
+      "grad_norm": 0.11652163416147232,
+      "learning_rate": 0.0001,
+      "loss": 0.2129,
+      "step": 732
+    },
+    {
+      "epoch": 0.993763557483731,
+      "grad_norm": 0.11725040525197983,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 733
+    },
+    {
+      "epoch": 0.995119305856833,
+      "grad_norm": 0.10410426557064056,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 734
+    },
+    {
+      "epoch": 0.9964750542299349,
+      "grad_norm": 0.11244272440671921,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 735
+    },
+    {
+      "epoch": 0.9978308026030369,
+      "grad_norm": 0.10956812649965286,
+      "learning_rate": 0.0001,
+      "loss": 0.2137,
+      "step": 736
+    },
+    {
+      "epoch": 0.9991865509761388,
+      "grad_norm": 0.11602611094713211,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 737
+    },
+    {
+      "epoch": 1.0005422993492408,
+      "grad_norm": 0.25436535477638245,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 738
+    },
+    {
+      "epoch": 1.0018980477223427,
+      "grad_norm": 0.10693640261888504,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 739
+    },
+    {
+      "epoch": 1.0032537960954446,
+      "grad_norm": 0.09938878566026688,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 740
+    },
+    {
+      "epoch": 1.0046095444685466,
+      "grad_norm": 0.10401073098182678,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 741
+    },
+    {
+      "epoch": 1.0059652928416485,
+      "grad_norm": 0.11641883850097656,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 742
+    },
+    {
+      "epoch": 1.0073210412147506,
+      "grad_norm": 0.10613098740577698,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 743
+    },
+    {
+      "epoch": 1.0086767895878526,
+      "grad_norm": 0.10632190853357315,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 744
+    },
+    {
+      "epoch": 1.0100325379609545,
+      "grad_norm": 0.12229351699352264,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 745
+    },
+    {
+      "epoch": 1.0113882863340564,
+      "grad_norm": 0.11870548129081726,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 746
+    },
+    {
+      "epoch": 1.0127440347071583,
+      "grad_norm": 0.11921311169862747,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 747
+    },
+    {
+      "epoch": 1.0140997830802603,
+      "grad_norm": 0.11286763101816177,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 748
+    },
+    {
+      "epoch": 1.0154555314533622,
+      "grad_norm": 0.11586420238018036,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 749
+    },
+    {
+      "epoch": 1.0168112798264641,
+      "grad_norm": 0.10945212841033936,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 750
+    },
+    {
+      "epoch": 1.018167028199566,
+      "grad_norm": 0.1315394639968872,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 751
+    },
+    {
+      "epoch": 1.0195227765726682,
+      "grad_norm": 0.11502466350793839,
+      "learning_rate": 0.0001,
+      "loss": 0.1816,
+      "step": 752
+    },
+    {
+      "epoch": 1.0208785249457701,
+      "grad_norm": 0.13596606254577637,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 753
+    },
+    {
+      "epoch": 1.022234273318872,
+      "grad_norm": 0.12724058330059052,
+      "learning_rate": 0.0001,
+      "loss": 0.2125,
+      "step": 754
+    },
+    {
+      "epoch": 1.023590021691974,
+      "grad_norm": 0.11776183545589447,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 755
+    },
+    {
+      "epoch": 1.024945770065076,
+      "grad_norm": 0.12062231451272964,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 756
+    },
+    {
+      "epoch": 1.0263015184381779,
+      "grad_norm": 0.11406990885734558,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 757
+    },
+    {
+      "epoch": 1.0276572668112798,
+      "grad_norm": 0.10735568404197693,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 758
+    },
+    {
+      "epoch": 1.0290130151843817,
+      "grad_norm": 0.12077546119689941,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 759
+    },
+    {
+      "epoch": 1.0303687635574836,
+      "grad_norm": 0.14352481067180634,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 760
+    },
+    {
+      "epoch": 1.0317245119305858,
+      "grad_norm": 0.10997813194990158,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 761
+    },
+    {
+      "epoch": 1.0330802603036877,
+      "grad_norm": 0.1143399029970169,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 762
+    },
+    {
+      "epoch": 1.0344360086767896,
+      "grad_norm": 0.11430211365222931,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 763
+    },
+    {
+      "epoch": 1.0357917570498916,
+      "grad_norm": 0.11984989792108536,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 764
+    },
+    {
+      "epoch": 1.0371475054229935,
+      "grad_norm": 0.1228814646601677,
+      "learning_rate": 0.0001,
+      "loss": 0.1981,
+      "step": 765
+    },
+    {
+      "epoch": 1.0385032537960954,
+      "grad_norm": 0.10160529613494873,
+      "learning_rate": 0.0001,
+      "loss": 0.1857,
+      "step": 766
+    },
+    {
+      "epoch": 1.0398590021691974,
+      "grad_norm": 0.11860557645559311,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 767
+    },
+    {
+      "epoch": 1.0412147505422993,
+      "grad_norm": 0.11132457107305527,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 768
+    },
+    {
+      "epoch": 1.0425704989154012,
+      "grad_norm": 0.12177620083093643,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 769
+    },
+    {
+      "epoch": 1.0439262472885034,
+      "grad_norm": 0.14306630194187164,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 770
+    },
+    {
+      "epoch": 1.0452819956616053,
+      "grad_norm": 0.12264086306095123,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 771
+    },
+    {
+      "epoch": 1.0466377440347072,
+      "grad_norm": 0.11370858550071716,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 772
+    },
+    {
+      "epoch": 1.0479934924078091,
+      "grad_norm": 0.11763013899326324,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 773
+    },
+    {
+      "epoch": 1.049349240780911,
+      "grad_norm": 0.12041058391332626,
+      "learning_rate": 0.0001,
+      "loss": 0.2058,
+      "step": 774
+    },
+    {
+      "epoch": 1.050704989154013,
+      "grad_norm": 0.12122388184070587,
+      "learning_rate": 0.0001,
+      "loss": 0.1898,
+      "step": 775
+    },
+    {
+      "epoch": 1.052060737527115,
+      "grad_norm": 0.11808788031339645,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 776
+    },
+    {
+      "epoch": 1.0534164859002169,
+      "grad_norm": 0.1098509132862091,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 777
+    },
+    {
+      "epoch": 1.0547722342733188,
+      "grad_norm": 0.1229412853717804,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 778
+    },
+    {
+      "epoch": 1.0561279826464207,
+      "grad_norm": 0.11343245208263397,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 779
+    },
+    {
+      "epoch": 1.0574837310195229,
+      "grad_norm": 0.15121108293533325,
+      "learning_rate": 0.0001,
+      "loss": 0.2031,
+      "step": 780
+    },
+    {
+      "epoch": 1.0588394793926248,
+      "grad_norm": 0.2177276462316513,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 781
+    },
+    {
+      "epoch": 1.0601952277657267,
+      "grad_norm": 0.13165955245494843,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 782
+    },
+    {
+      "epoch": 1.0615509761388287,
+      "grad_norm": 0.11355244368314743,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 783
+    },
+    {
+      "epoch": 1.0629067245119306,
+      "grad_norm": 0.13550937175750732,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 784
+    },
+    {
+      "epoch": 1.0642624728850325,
+      "grad_norm": 0.11278457194566727,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 785
+    },
+    {
+      "epoch": 1.0656182212581344,
+      "grad_norm": 0.13036085665225983,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 786
+    },
+    {
+      "epoch": 1.0669739696312364,
+      "grad_norm": 0.12656573951244354,
+      "learning_rate": 0.0001,
+      "loss": 0.1787,
+      "step": 787
+    },
+    {
+      "epoch": 1.0683297180043383,
+      "grad_norm": 0.1420610100030899,
+      "learning_rate": 0.0001,
+      "loss": 0.1939,
+      "step": 788
+    },
+    {
+      "epoch": 1.0696854663774404,
+      "grad_norm": 0.1299659162759781,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 789
+    },
+    {
+      "epoch": 1.0710412147505424,
+      "grad_norm": 0.1357019692659378,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 790
+    },
+    {
+      "epoch": 1.0723969631236443,
+      "grad_norm": 0.1231936663389206,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 791
+    },
+    {
+      "epoch": 1.0737527114967462,
+      "grad_norm": 0.09797357767820358,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 792
+    },
+    {
+      "epoch": 1.0751084598698482,
+      "grad_norm": 0.10265905410051346,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 793
+    },
+    {
+      "epoch": 1.07646420824295,
+      "grad_norm": 0.10287607461214066,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 794
+    },
+    {
+      "epoch": 1.077819956616052,
+      "grad_norm": 0.10395827144384384,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 795
+    },
+    {
+      "epoch": 1.079175704989154,
+      "grad_norm": 0.11927088350057602,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 796
+    },
+    {
+      "epoch": 1.0805314533622559,
+      "grad_norm": 0.11181191354990005,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 797
+    },
+    {
+      "epoch": 1.081887201735358,
+      "grad_norm": 0.12188813835382462,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 798
+    },
+    {
+      "epoch": 1.08324295010846,
+      "grad_norm": 0.11425131559371948,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 799
+    },
+    {
+      "epoch": 1.0845986984815619,
+      "grad_norm": 0.10277964174747467,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 800
+    },
+    {
+      "epoch": 1.0859544468546638,
+      "grad_norm": 0.12257596105337143,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 801
+    },
+    {
+      "epoch": 1.0873101952277657,
+      "grad_norm": 0.1264577955007553,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 802
+    },
+    {
+      "epoch": 1.0886659436008677,
+      "grad_norm": 0.1235210970044136,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 803
+    },
+    {
+      "epoch": 1.0900216919739696,
+      "grad_norm": 0.12020499259233475,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 804
+    },
+    {
+      "epoch": 1.0913774403470715,
+      "grad_norm": 0.12127955257892609,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 805
+    },
+    {
+      "epoch": 1.0927331887201734,
+      "grad_norm": 0.13456392288208008,
+      "learning_rate": 0.0001,
+      "loss": 0.2118,
+      "step": 806
+    },
+    {
+      "epoch": 1.0940889370932756,
+      "grad_norm": 0.12651638686656952,
+      "learning_rate": 0.0001,
+      "loss": 0.2037,
+      "step": 807
+    },
+    {
+      "epoch": 1.0954446854663775,
+      "grad_norm": 0.11540345102548599,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 808
+    },
+    {
+      "epoch": 1.0968004338394794,
+      "grad_norm": 0.11138421297073364,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 809
+    },
+    {
+      "epoch": 1.0981561822125814,
+      "grad_norm": 0.10703103989362717,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 810
+    },
+    {
+      "epoch": 1.0995119305856833,
+      "grad_norm": 0.1276537925004959,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 811
+    },
+    {
+      "epoch": 1.1008676789587852,
+      "grad_norm": 0.11813724040985107,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 812
+    },
+    {
+      "epoch": 1.1022234273318872,
+      "grad_norm": 0.12671397626399994,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 813
+    },
+    {
+      "epoch": 1.103579175704989,
+      "grad_norm": 0.11774583905935287,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 814
+    },
+    {
+      "epoch": 1.104934924078091,
+      "grad_norm": 0.10728626698255539,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 815
+    },
+    {
+      "epoch": 1.106290672451193,
+      "grad_norm": 0.12491397559642792,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 816
+    },
+    {
+      "epoch": 1.107646420824295,
+      "grad_norm": 0.1744874119758606,
+      "learning_rate": 0.0001,
+      "loss": 0.1955,
+      "step": 817
+    },
+    {
+      "epoch": 1.109002169197397,
+      "grad_norm": 0.11541765928268433,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 818
+    },
+    {
+      "epoch": 1.110357917570499,
+      "grad_norm": 0.10669142752885818,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 819
+    },
+    {
+      "epoch": 1.1117136659436009,
+      "grad_norm": 0.11726786196231842,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 820
+    },
+    {
+      "epoch": 1.1130694143167028,
+      "grad_norm": 0.11508400738239288,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 821
+    },
+    {
+      "epoch": 1.1144251626898047,
+      "grad_norm": 0.0930793359875679,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 822
+    },
+    {
+      "epoch": 1.1157809110629067,
+      "grad_norm": 0.12295538187026978,
+      "learning_rate": 0.0001,
+      "loss": 0.1938,
+      "step": 823
+    },
+    {
+      "epoch": 1.1171366594360086,
+      "grad_norm": 0.11898976564407349,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 824
+    },
+    {
+      "epoch": 1.1184924078091107,
+      "grad_norm": 0.10814597457647324,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 825
+    },
+    {
+      "epoch": 1.1198481561822127,
+      "grad_norm": 0.14314916729927063,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 826
+    },
+    {
+      "epoch": 1.1212039045553146,
+      "grad_norm": 0.1266772449016571,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 827
+    },
+    {
+      "epoch": 1.1225596529284165,
+      "grad_norm": 0.12375739961862564,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 828
+    },
+    {
+      "epoch": 1.1239154013015185,
+      "grad_norm": 0.11660799384117126,
+      "learning_rate": 0.0001,
+      "loss": 0.2041,
+      "step": 829
+    },
+    {
+      "epoch": 1.1252711496746204,
+      "grad_norm": 0.12489171326160431,
+      "learning_rate": 0.0001,
+      "loss": 0.1788,
+      "step": 830
+    },
+    {
+      "epoch": 1.1266268980477223,
+      "grad_norm": 0.12043704837560654,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 831
+    },
+    {
+      "epoch": 1.1279826464208242,
+      "grad_norm": 0.11040499061346054,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 832
+    },
+    {
+      "epoch": 1.1293383947939262,
+      "grad_norm": 0.12426988035440445,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 833
+    },
+    {
+      "epoch": 1.130694143167028,
+      "grad_norm": 0.10827081650495529,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 834
+    },
+    {
+      "epoch": 1.1320498915401302,
+      "grad_norm": 0.11265697330236435,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 835
+    },
+    {
+      "epoch": 1.1334056399132322,
+      "grad_norm": 0.12313784658908844,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 836
+    },
+    {
+      "epoch": 1.134761388286334,
+      "grad_norm": 0.12942351400852203,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 837
+    },
+    {
+      "epoch": 1.136117136659436,
+      "grad_norm": 0.1544887125492096,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 838
+    },
+    {
+      "epoch": 1.137472885032538,
+      "grad_norm": 0.11129739880561829,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 839
+    },
+    {
+      "epoch": 1.13882863340564,
+      "grad_norm": 0.1632719337940216,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 840
+    },
+    {
+      "epoch": 1.1401843817787418,
+      "grad_norm": 0.11265744268894196,
+      "learning_rate": 0.0001,
+      "loss": 0.2126,
+      "step": 841
+    },
+    {
+      "epoch": 1.1415401301518437,
+      "grad_norm": 0.12824206054210663,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 842
+    },
+    {
+      "epoch": 1.1428958785249457,
+      "grad_norm": 0.19684740900993347,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 843
+    },
+    {
+      "epoch": 1.1442516268980478,
+      "grad_norm": 0.11099156737327576,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 844
+    },
+    {
+      "epoch": 1.1456073752711498,
+      "grad_norm": 0.11781927198171616,
+      "learning_rate": 0.0001,
+      "loss": 0.2089,
+      "step": 845
+    },
+    {
+      "epoch": 1.1469631236442517,
+      "grad_norm": 0.12977120280265808,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 846
+    },
+    {
+      "epoch": 1.1483188720173536,
+      "grad_norm": 0.1544928252696991,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 847
+    },
+    {
+      "epoch": 1.1496746203904555,
+      "grad_norm": 0.13819129765033722,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 848
+    },
+    {
+      "epoch": 1.1510303687635575,
+      "grad_norm": 0.13939248025417328,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 849
+    },
+    {
+      "epoch": 1.1523861171366594,
+      "grad_norm": 0.1409015953540802,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 850
+    },
+    {
+      "epoch": 1.1537418655097613,
+      "grad_norm": 0.14717864990234375,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 851
+    },
+    {
+      "epoch": 1.1550976138828633,
+      "grad_norm": 0.13812538981437683,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 852
+    },
+    {
+      "epoch": 1.1564533622559652,
+      "grad_norm": 0.1089063510298729,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 853
+    },
+    {
+      "epoch": 1.1578091106290673,
+      "grad_norm": 0.1296411007642746,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 854
+    },
+    {
+      "epoch": 1.1591648590021693,
+      "grad_norm": 0.11074882000684738,
+      "learning_rate": 0.0001,
+      "loss": 0.1991,
+      "step": 855
+    },
+    {
+      "epoch": 1.1605206073752712,
+      "grad_norm": 0.128693088889122,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 856
+    },
+    {
+      "epoch": 1.1618763557483731,
+      "grad_norm": 0.13630498945713043,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 857
+    },
+    {
+      "epoch": 1.163232104121475,
+      "grad_norm": 0.11853824555873871,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 858
+    },
+    {
+      "epoch": 1.164587852494577,
+      "grad_norm": 0.13688893616199493,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 859
+    },
+    {
+      "epoch": 1.165943600867679,
+      "grad_norm": 0.12337955832481384,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 860
+    },
+    {
+      "epoch": 1.1672993492407808,
+      "grad_norm": 0.1359425038099289,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 861
+    },
+    {
+      "epoch": 1.168655097613883,
+      "grad_norm": 0.11356756091117859,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 862
+    },
+    {
+      "epoch": 1.170010845986985,
+      "grad_norm": 0.1301959604024887,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 863
+    },
+    {
+      "epoch": 1.1713665943600868,
+      "grad_norm": 0.1659461259841919,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 864
+    },
+    {
+      "epoch": 1.1727223427331888,
+      "grad_norm": 0.14047051966190338,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 865
+    },
+    {
+      "epoch": 1.1740780911062907,
+      "grad_norm": 0.11875968426465988,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 866
+    },
+    {
+      "epoch": 1.1754338394793926,
+      "grad_norm": 0.11229258030653,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 867
+    },
+    {
+      "epoch": 1.1767895878524945,
+      "grad_norm": 0.10771773755550385,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 868
+    },
+    {
+      "epoch": 1.1781453362255965,
+      "grad_norm": 0.13326172530651093,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 869
+    },
+    {
+      "epoch": 1.1795010845986984,
+      "grad_norm": 0.11325906217098236,
+      "learning_rate": 0.0001,
+      "loss": 0.2065,
+      "step": 870
+    },
+    {
+      "epoch": 1.1808568329718003,
+      "grad_norm": 0.1327616423368454,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 871
+    },
+    {
+      "epoch": 1.1822125813449025,
+      "grad_norm": 0.11589329689741135,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 872
+    },
+    {
+      "epoch": 1.1835683297180044,
+      "grad_norm": 0.7335811853408813,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 873
+    },
+    {
+      "epoch": 1.1849240780911063,
+      "grad_norm": 0.10128708928823471,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 874
+    },
+    {
+      "epoch": 1.1862798264642083,
+      "grad_norm": 0.14725837111473083,
+      "learning_rate": 0.0001,
+      "loss": 0.2053,
+      "step": 875
+    },
+    {
+      "epoch": 1.1876355748373102,
+      "grad_norm": 0.14333541691303253,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 876
+    },
+    {
+      "epoch": 1.1889913232104121,
+      "grad_norm": 0.16398996114730835,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 877
+    },
+    {
+      "epoch": 1.190347071583514,
+      "grad_norm": 0.11339271813631058,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 878
+    },
+    {
+      "epoch": 1.191702819956616,
+      "grad_norm": 0.12222304195165634,
+      "learning_rate": 0.0001,
+      "loss": 0.2041,
+      "step": 879
+    },
+    {
+      "epoch": 1.1930585683297181,
+      "grad_norm": 0.11465858668088913,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 880
+    },
+    {
+      "epoch": 1.19441431670282,
+      "grad_norm": 0.16641555726528168,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 881
+    },
+    {
+      "epoch": 1.195770065075922,
+      "grad_norm": 0.12705209851264954,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 882
+    },
+    {
+      "epoch": 1.197125813449024,
+      "grad_norm": 0.11217901110649109,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 883
+    },
+    {
+      "epoch": 1.1984815618221258,
+      "grad_norm": 0.1257891058921814,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 884
+    },
+    {
+      "epoch": 1.1998373101952278,
+      "grad_norm": 0.132249653339386,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 885
+    },
+    {
+      "epoch": 1.2011930585683297,
+      "grad_norm": 0.1220329999923706,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 886
+    },
+    {
+      "epoch": 1.2025488069414316,
+      "grad_norm": 0.15027636289596558,
+      "learning_rate": 0.0001,
+      "loss": 0.1992,
+      "step": 887
+    },
+    {
+      "epoch": 1.2039045553145336,
+      "grad_norm": 0.11561386287212372,
+      "learning_rate": 0.0001,
+      "loss": 0.1912,
+      "step": 888
+    },
+    {
+      "epoch": 1.2052603036876355,
+      "grad_norm": 0.1185799315571785,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 889
+    },
+    {
+      "epoch": 1.2066160520607374,
+      "grad_norm": 0.1488744169473648,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 890
+    },
+    {
+      "epoch": 1.2079718004338396,
+      "grad_norm": 0.14555370807647705,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 891
+    },
+    {
+      "epoch": 1.2093275488069415,
+      "grad_norm": 0.11113309860229492,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 892
+    },
+    {
+      "epoch": 1.2106832971800434,
+      "grad_norm": 0.10961677879095078,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 893
+    },
+    {
+      "epoch": 1.2120390455531453,
+      "grad_norm": 0.13135626912117004,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 894
+    },
+    {
+      "epoch": 1.2133947939262473,
+      "grad_norm": 0.1027296632528305,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 895
+    },
+    {
+      "epoch": 1.2147505422993492,
+      "grad_norm": 0.10646037012338638,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 896
+    },
+    {
+      "epoch": 1.2161062906724511,
+      "grad_norm": 0.133303701877594,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 897
+    },
+    {
+      "epoch": 1.217462039045553,
+      "grad_norm": 0.1276165097951889,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 898
+    },
+    {
+      "epoch": 1.2188177874186552,
+      "grad_norm": 0.11786617338657379,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 899
+    },
+    {
+      "epoch": 1.2201735357917571,
+      "grad_norm": 0.12070033699274063,
+      "learning_rate": 0.0001,
+      "loss": 0.205,
+      "step": 900
+    },
+    {
+      "epoch": 1.221529284164859,
+      "grad_norm": 0.12070579081773758,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 901
+    },
+    {
+      "epoch": 1.222885032537961,
+      "grad_norm": 0.11606718599796295,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 902
+    },
+    {
+      "epoch": 1.224240780911063,
+      "grad_norm": 0.15249618887901306,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 903
+    },
+    {
+      "epoch": 1.2255965292841648,
+      "grad_norm": 0.13006334006786346,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 904
+    },
+    {
+      "epoch": 1.2269522776572668,
+      "grad_norm": 0.13756072521209717,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 905
+    },
+    {
+      "epoch": 1.2283080260303687,
+      "grad_norm": 0.15109659731388092,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 906
+    },
+    {
+      "epoch": 1.2296637744034706,
+      "grad_norm": 0.11546261608600616,
+      "learning_rate": 0.0001,
+      "loss": 0.1923,
+      "step": 907
+    },
+    {
+      "epoch": 1.2310195227765726,
+      "grad_norm": 0.11288850754499435,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 908
+    },
+    {
+      "epoch": 1.2323752711496747,
+      "grad_norm": 0.13695698976516724,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 909
+    },
+    {
+      "epoch": 1.2337310195227766,
+      "grad_norm": 0.11048099398612976,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 910
+    },
+    {
+      "epoch": 1.2350867678958786,
+      "grad_norm": 0.11988060921430588,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 911
+    },
+    {
+      "epoch": 1.2364425162689805,
+      "grad_norm": 0.1545763462781906,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 912
+    },
+    {
+      "epoch": 1.2377982646420824,
+      "grad_norm": 0.10802659392356873,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 913
+    },
+    {
+      "epoch": 1.2391540130151844,
+      "grad_norm": 0.12360113114118576,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 914
+    },
+    {
+      "epoch": 1.2405097613882863,
+      "grad_norm": 0.1242009624838829,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 915
+    },
+    {
+      "epoch": 1.2418655097613882,
+      "grad_norm": 0.126487597823143,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 916
+    },
+    {
+      "epoch": 1.2432212581344904,
+      "grad_norm": 0.12998108565807343,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 917
+    },
+    {
+      "epoch": 1.2445770065075923,
+      "grad_norm": 0.12627403438091278,
+      "learning_rate": 0.0001,
+      "loss": 0.1944,
+      "step": 918
+    },
+    {
+      "epoch": 1.2459327548806942,
+      "grad_norm": 0.12328945845365524,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 919
+    },
+    {
+      "epoch": 1.2472885032537961,
+      "grad_norm": 0.1290748417377472,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 920
+    },
+    {
+      "epoch": 1.248644251626898,
+      "grad_norm": 0.1166953295469284,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 921
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.1032584086060524,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 922
+    },
+    {
+      "epoch": 1.251355748373102,
+      "grad_norm": 0.12523934245109558,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 923
+    },
+    {
+      "epoch": 1.2527114967462039,
+      "grad_norm": 0.12179497629404068,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 924
+    },
+    {
+      "epoch": 1.2540672451193058,
+      "grad_norm": 0.13713449239730835,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 925
+    },
+    {
+      "epoch": 1.2554229934924077,
+      "grad_norm": 0.11815124750137329,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 926
+    },
+    {
+      "epoch": 1.2567787418655096,
+      "grad_norm": 0.1333799958229065,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 927
+    },
+    {
+      "epoch": 1.2581344902386118,
+      "grad_norm": 0.11289018392562866,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 928
+    },
+    {
+      "epoch": 1.2594902386117137,
+      "grad_norm": 0.1333857923746109,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 929
+    },
+    {
+      "epoch": 1.2608459869848156,
+      "grad_norm": 0.11712473630905151,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 930
+    },
+    {
+      "epoch": 1.2622017353579176,
+      "grad_norm": 0.12506066262722015,
+      "learning_rate": 0.0001,
+      "loss": 0.1911,
+      "step": 931
+    },
+    {
+      "epoch": 1.2635574837310195,
+      "grad_norm": 0.14353911578655243,
+      "learning_rate": 0.0001,
+      "loss": 0.2005,
+      "step": 932
+    },
+    {
+      "epoch": 1.2649132321041214,
+      "grad_norm": 0.12906968593597412,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 933
+    },
+    {
+      "epoch": 1.2662689804772234,
+      "grad_norm": 0.10110311955213547,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 934
+    },
+    {
+      "epoch": 1.2676247288503255,
+      "grad_norm": 0.1439683735370636,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 935
+    },
+    {
+      "epoch": 1.2689804772234274,
+      "grad_norm": 0.136976957321167,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 936
+    },
+    {
+      "epoch": 1.2703362255965294,
+      "grad_norm": 0.11722557246685028,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 937
+    },
+    {
+      "epoch": 1.2716919739696313,
+      "grad_norm": 0.11081776767969131,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 938
+    },
+    {
+      "epoch": 1.2730477223427332,
+      "grad_norm": 0.11922642588615417,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 939
+    },
+    {
+      "epoch": 1.2744034707158352,
+      "grad_norm": 0.12547357380390167,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 940
+    },
+    {
+      "epoch": 1.275759219088937,
+      "grad_norm": 0.15054965019226074,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 941
+    },
+    {
+      "epoch": 1.277114967462039,
+      "grad_norm": 0.15004533529281616,
+      "learning_rate": 0.0001,
+      "loss": 0.183,
+      "step": 942
+    },
+    {
+      "epoch": 1.278470715835141,
+      "grad_norm": 0.1321701556444168,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 943
+    },
+    {
+      "epoch": 1.2798264642082429,
+      "grad_norm": 0.1444014459848404,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 944
+    },
+    {
+      "epoch": 1.2811822125813448,
+      "grad_norm": 0.12540584802627563,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 945
+    },
+    {
+      "epoch": 1.2825379609544467,
+      "grad_norm": 0.13105201721191406,
+      "learning_rate": 0.0001,
+      "loss": 0.1953,
+      "step": 946
+    },
+    {
+      "epoch": 1.2838937093275489,
+      "grad_norm": 0.12747126817703247,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 947
+    },
+    {
+      "epoch": 1.2852494577006508,
+      "grad_norm": 0.12107202410697937,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 948
+    },
+    {
+      "epoch": 1.2866052060737527,
+      "grad_norm": 0.11618299782276154,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 949
+    },
+    {
+      "epoch": 1.2879609544468547,
+      "grad_norm": 0.1385001838207245,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 950
+    },
+    {
+      "epoch": 1.2893167028199566,
+      "grad_norm": 0.13121256232261658,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 951
+    },
+    {
+      "epoch": 1.2906724511930585,
+      "grad_norm": 0.12440761923789978,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 952
+    },
+    {
+      "epoch": 1.2920281995661604,
+      "grad_norm": 0.10734440386295319,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 953
+    },
+    {
+      "epoch": 1.2933839479392626,
+      "grad_norm": 0.13095004856586456,
+      "learning_rate": 0.0001,
+      "loss": 0.2175,
+      "step": 954
+    },
+    {
+      "epoch": 1.2947396963123645,
+      "grad_norm": 0.13218335807323456,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 955
+    },
+    {
+      "epoch": 1.2960954446854664,
+      "grad_norm": 0.1265275776386261,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 956
+    },
+    {
+      "epoch": 1.2974511930585684,
+      "grad_norm": 0.12137480080127716,
+      "learning_rate": 0.0001,
+      "loss": 0.201,
+      "step": 957
+    },
+    {
+      "epoch": 1.2988069414316703,
+      "grad_norm": 0.12581804394721985,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 958
+    },
+    {
+      "epoch": 1.3001626898047722,
+      "grad_norm": 0.127769336104393,
+      "learning_rate": 0.0001,
+      "loss": 0.2052,
+      "step": 959
+    },
+    {
+      "epoch": 1.3015184381778742,
+      "grad_norm": 0.11478791385889053,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 960
+    },
+    {
+      "epoch": 1.302874186550976,
+      "grad_norm": 0.11588724702596664,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 961
+    },
+    {
+      "epoch": 1.304229934924078,
+      "grad_norm": 0.13547374308109283,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 962
+    },
+    {
+      "epoch": 1.30558568329718,
+      "grad_norm": 0.12892450392246246,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 963
+    },
+    {
+      "epoch": 1.3069414316702819,
+      "grad_norm": 0.11946334689855576,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 964
+    },
+    {
+      "epoch": 1.308297180043384,
+      "grad_norm": 0.10716400295495987,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 965
+    },
+    {
+      "epoch": 1.309652928416486,
+      "grad_norm": 0.1252261996269226,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 966
+    },
+    {
+      "epoch": 1.3110086767895879,
+      "grad_norm": 0.1143805980682373,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 967
+    },
+    {
+      "epoch": 1.3123644251626898,
+      "grad_norm": 0.140336811542511,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 968
+    },
+    {
+      "epoch": 1.3137201735357917,
+      "grad_norm": 0.11809618026018143,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 969
+    },
+    {
+      "epoch": 1.3150759219088937,
+      "grad_norm": 0.10512138158082962,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 970
+    },
+    {
+      "epoch": 1.3164316702819956,
+      "grad_norm": 0.11192452907562256,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 971
+    },
+    {
+      "epoch": 1.3177874186550977,
+      "grad_norm": 0.11674557626247406,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 972
+    },
+    {
+      "epoch": 1.3191431670281997,
+      "grad_norm": 0.11132452636957169,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 973
+    },
+    {
+      "epoch": 1.3204989154013016,
+      "grad_norm": 0.10949098318815231,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 974
+    },
+    {
+      "epoch": 1.3218546637744035,
+      "grad_norm": 0.1156991571187973,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 975
+    },
+    {
+      "epoch": 1.3232104121475055,
+      "grad_norm": 0.10592202842235565,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 976
+    },
+    {
+      "epoch": 1.3245661605206074,
+      "grad_norm": 0.12329782545566559,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 977
+    },
+    {
+      "epoch": 1.3259219088937093,
+      "grad_norm": 0.16654783487319946,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 978
+    },
+    {
+      "epoch": 1.3272776572668112,
+      "grad_norm": 0.12002754211425781,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 979
+    },
+    {
+      "epoch": 1.3286334056399132,
+      "grad_norm": 0.1073680967092514,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 980
+    },
+    {
+      "epoch": 1.329989154013015,
+      "grad_norm": 0.12832270562648773,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 981
+    },
+    {
+      "epoch": 1.331344902386117,
+      "grad_norm": 0.1141325831413269,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 982
+    },
+    {
+      "epoch": 1.332700650759219,
+      "grad_norm": 0.11039267480373383,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 983
+    },
+    {
+      "epoch": 1.334056399132321,
+      "grad_norm": 0.11766202747821808,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 984
+    },
+    {
+      "epoch": 1.335412147505423,
+      "grad_norm": 0.11707600951194763,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 985
+    },
+    {
+      "epoch": 1.336767895878525,
+      "grad_norm": 0.12010547518730164,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 986
+    },
+    {
+      "epoch": 1.3381236442516269,
+      "grad_norm": 0.12517651915550232,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 987
+    },
+    {
+      "epoch": 1.3394793926247288,
+      "grad_norm": 0.11202540248632431,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 988
+    },
+    {
+      "epoch": 1.3408351409978307,
+      "grad_norm": 0.10568049550056458,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 989
+    },
+    {
+      "epoch": 1.342190889370933,
+      "grad_norm": 0.115880586206913,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 990
+    },
+    {
+      "epoch": 1.3435466377440348,
+      "grad_norm": 0.15287692844867706,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 991
+    },
+    {
+      "epoch": 1.3449023861171367,
+      "grad_norm": 0.1133360043168068,
+      "learning_rate": 0.0001,
+      "loss": 0.2077,
+      "step": 992
+    },
+    {
+      "epoch": 1.3462581344902387,
+      "grad_norm": 0.13320140540599823,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 993
+    },
+    {
+      "epoch": 1.3476138828633406,
+      "grad_norm": 0.10174473375082016,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 994
+    },
+    {
+      "epoch": 1.3489696312364425,
+      "grad_norm": 0.44916167855262756,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 995
+    },
+    {
+      "epoch": 1.3503253796095445,
+      "grad_norm": 0.11276068538427353,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 996
+    },
+    {
+      "epoch": 1.3516811279826464,
+      "grad_norm": 0.11648362129926682,
+      "learning_rate": 0.0001,
+      "loss": 0.2044,
+      "step": 997
+    },
+    {
+      "epoch": 1.3530368763557483,
+      "grad_norm": 0.11647507548332214,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 998
+    },
+    {
+      "epoch": 1.3543926247288502,
+      "grad_norm": 0.11464931070804596,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 999
+    },
+    {
+      "epoch": 1.3557483731019522,
+      "grad_norm": 0.10750597715377808,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1000
+    },
+    {
+      "epoch": 1.3557483731019522,
+      "eval_loss": 0.20068813860416412,
+      "eval_runtime": 587.3199,
+      "eval_samples_per_second": 4.71,
+      "eval_steps_per_second": 1.178,
+      "step": 1000
+    },
+    {
+      "epoch": 1.357104121475054,
+      "grad_norm": 0.11371137201786041,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 1001
+    },
+    {
+      "epoch": 1.3584598698481563,
+      "grad_norm": 0.1218474954366684,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1002
+    },
+    {
+      "epoch": 1.3598156182212582,
+      "grad_norm": 0.1285208910703659,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 1003
+    },
+    {
+      "epoch": 1.36117136659436,
+      "grad_norm": 0.10109011828899384,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1004
+    },
+    {
+      "epoch": 1.362527114967462,
+      "grad_norm": 0.12855185568332672,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1005
+    },
+    {
+      "epoch": 1.363882863340564,
+      "grad_norm": 0.1162986010313034,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 1006
+    },
+    {
+      "epoch": 1.365238611713666,
+      "grad_norm": 0.11368726938962936,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1007
+    },
+    {
+      "epoch": 1.3665943600867678,
+      "grad_norm": 0.1013002023100853,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 1008
+    },
+    {
+      "epoch": 1.36795010845987,
+      "grad_norm": 0.12230715155601501,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1009
+    },
+    {
+      "epoch": 1.369305856832972,
+      "grad_norm": 0.11484050005674362,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 1010
+    },
+    {
+      "epoch": 1.3706616052060738,
+      "grad_norm": 0.12114618718624115,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 1011
+    },
+    {
+      "epoch": 1.3720173535791758,
+      "grad_norm": 0.09783415496349335,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1012
+    },
+    {
+      "epoch": 1.3733731019522777,
+      "grad_norm": 0.11668910086154938,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 1013
+    },
+    {
+      "epoch": 1.3747288503253796,
+      "grad_norm": 0.12314967066049576,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1014
+    },
+    {
+      "epoch": 1.3760845986984815,
+      "grad_norm": 0.10992521792650223,
+      "learning_rate": 0.0001,
+      "loss": 0.1881,
+      "step": 1015
+    },
+    {
+      "epoch": 1.3774403470715835,
+      "grad_norm": 0.11317230015993118,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1016
+    },
+    {
+      "epoch": 1.3787960954446854,
+      "grad_norm": 0.11415086686611176,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 1017
+    },
+    {
+      "epoch": 1.3801518438177873,
+      "grad_norm": 0.12152314931154251,
+      "learning_rate": 0.0001,
+      "loss": 0.1791,
+      "step": 1018
+    },
+    {
+      "epoch": 1.3815075921908893,
+      "grad_norm": 0.1265685260295868,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 1019
+    },
+    {
+      "epoch": 1.3828633405639914,
+      "grad_norm": 0.11083249747753143,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 1020
+    },
+    {
+      "epoch": 1.3842190889370933,
+      "grad_norm": 0.13765420019626617,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1021
+    },
+    {
+      "epoch": 1.3855748373101953,
+      "grad_norm": 0.11946640908718109,
+      "learning_rate": 0.0001,
+      "loss": 0.2107,
+      "step": 1022
+    },
+    {
+      "epoch": 1.3869305856832972,
+      "grad_norm": 0.13332737982273102,
+      "learning_rate": 0.0001,
+      "loss": 0.2022,
+      "step": 1023
+    },
+    {
+      "epoch": 1.3882863340563991,
+      "grad_norm": 0.11524049192667007,
+      "learning_rate": 0.0001,
+      "loss": 0.1787,
+      "step": 1024
+    },
+    {
+      "epoch": 1.389642082429501,
+      "grad_norm": 0.12142997235059738,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 1025
+    },
+    {
+      "epoch": 1.390997830802603,
+      "grad_norm": 0.11735180765390396,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1026
+    },
+    {
+      "epoch": 1.3923535791757051,
+      "grad_norm": 0.11266755312681198,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 1027
+    },
+    {
+      "epoch": 1.393709327548807,
+      "grad_norm": 0.11281787604093552,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1028
+    },
+    {
+      "epoch": 1.395065075921909,
+      "grad_norm": 0.11209619790315628,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1029
+    },
+    {
+      "epoch": 1.396420824295011,
+      "grad_norm": 0.108651302754879,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1030
+    },
+    {
+      "epoch": 1.3977765726681128,
+      "grad_norm": 0.12406530231237411,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1031
+    },
+    {
+      "epoch": 1.3991323210412148,
+      "grad_norm": 0.12167002260684967,
+      "learning_rate": 0.0001,
+      "loss": 0.2019,
+      "step": 1032
+    },
+    {
+      "epoch": 1.4004880694143167,
+      "grad_norm": 0.1461237519979477,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 1033
+    },
+    {
+      "epoch": 1.4018438177874186,
+      "grad_norm": 0.12844736874103546,
+      "learning_rate": 0.0001,
+      "loss": 0.2082,
+      "step": 1034
+    },
+    {
+      "epoch": 1.4031995661605206,
+      "grad_norm": 0.12697617709636688,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 1035
+    },
+    {
+      "epoch": 1.4045553145336225,
+      "grad_norm": 0.12123944610357285,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1036
+    },
+    {
+      "epoch": 1.4059110629067244,
+      "grad_norm": 0.13650144636631012,
+      "learning_rate": 0.0001,
+      "loss": 0.2085,
+      "step": 1037
+    },
+    {
+      "epoch": 1.4072668112798263,
+      "grad_norm": 0.11150817573070526,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 1038
+    },
+    {
+      "epoch": 1.4086225596529285,
+      "grad_norm": 0.12959595024585724,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 1039
+    },
+    {
+      "epoch": 1.4099783080260304,
+      "grad_norm": 0.1053675040602684,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 1040
+    },
+    {
+      "epoch": 1.4113340563991323,
+      "grad_norm": 0.13261812925338745,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1041
+    },
+    {
+      "epoch": 1.4126898047722343,
+      "grad_norm": 0.12131708115339279,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 1042
+    },
+    {
+      "epoch": 1.4140455531453362,
+      "grad_norm": 0.11687289923429489,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 1043
+    },
+    {
+      "epoch": 1.4154013015184381,
+      "grad_norm": 0.12361282855272293,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1044
+    },
+    {
+      "epoch": 1.41675704989154,
+      "grad_norm": 0.10678700357675552,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1045
+    },
+    {
+      "epoch": 1.4181127982646422,
+      "grad_norm": 0.1320444643497467,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 1046
+    },
+    {
+      "epoch": 1.4194685466377441,
+      "grad_norm": 0.12514331936836243,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 1047
+    },
+    {
+      "epoch": 1.420824295010846,
+      "grad_norm": 0.1211230605840683,
+      "learning_rate": 0.0001,
+      "loss": 0.2075,
+      "step": 1048
+    },
+    {
+      "epoch": 1.422180043383948,
+      "grad_norm": 0.12863247096538544,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1049
+    },
+    {
+      "epoch": 1.42353579175705,
+      "grad_norm": 0.24209816753864288,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 1050
+    },
+    {
+      "epoch": 1.4248915401301518,
+      "grad_norm": 0.12444008141756058,
+      "learning_rate": 0.0001,
+      "loss": 0.2024,
+      "step": 1051
+    },
+    {
+      "epoch": 1.4262472885032538,
+      "grad_norm": 0.13908544182777405,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1052
+    },
+    {
+      "epoch": 1.4276030368763557,
+      "grad_norm": 0.11489854753017426,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 1053
+    },
+    {
+      "epoch": 1.4289587852494576,
+      "grad_norm": 0.11523835361003876,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 1054
+    },
+    {
+      "epoch": 1.4303145336225596,
+      "grad_norm": 0.11590521782636642,
+      "learning_rate": 0.0001,
+      "loss": 0.1908,
+      "step": 1055
+    },
+    {
+      "epoch": 1.4316702819956615,
+      "grad_norm": 0.2362946718931198,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 1056
+    },
+    {
+      "epoch": 1.4330260303687636,
+      "grad_norm": 0.11213434487581253,
+      "learning_rate": 0.0001,
+      "loss": 0.1813,
+      "step": 1057
+    },
+    {
+      "epoch": 1.4343817787418656,
+      "grad_norm": 0.12066798657178879,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 1058
+    },
+    {
+      "epoch": 1.4357375271149675,
+      "grad_norm": 0.1135016530752182,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 1059
+    },
+    {
+      "epoch": 1.4370932754880694,
+      "grad_norm": 0.13168586790561676,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 1060
+    },
+    {
+      "epoch": 1.4384490238611713,
+      "grad_norm": 0.1198778823018074,
+      "learning_rate": 0.0001,
+      "loss": 0.2064,
+      "step": 1061
+    },
+    {
+      "epoch": 1.4398047722342733,
+      "grad_norm": 0.11412026733160019,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1062
+    },
+    {
+      "epoch": 1.4411605206073752,
+      "grad_norm": 0.13722017407417297,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1063
+    },
+    {
+      "epoch": 1.4425162689804774,
+      "grad_norm": 0.16059622168540955,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 1064
+    },
+    {
+      "epoch": 1.4438720173535793,
+      "grad_norm": 0.12309937924146652,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 1065
+    },
+    {
+      "epoch": 1.4452277657266812,
+      "grad_norm": 0.12493696808815002,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 1066
+    },
+    {
+      "epoch": 1.4465835140997831,
+      "grad_norm": 0.1121676042675972,
+      "learning_rate": 0.0001,
+      "loss": 0.2011,
+      "step": 1067
+    },
+    {
+      "epoch": 1.447939262472885,
+      "grad_norm": 0.10178512334823608,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 1068
+    },
+    {
+      "epoch": 1.449295010845987,
+      "grad_norm": 0.12241014838218689,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1069
+    },
+    {
+      "epoch": 1.450650759219089,
+      "grad_norm": 0.1257627010345459,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 1070
+    },
+    {
+      "epoch": 1.4520065075921909,
+      "grad_norm": 0.11385004222393036,
+      "learning_rate": 0.0001,
+      "loss": 0.1982,
+      "step": 1071
+    },
+    {
+      "epoch": 1.4533622559652928,
+      "grad_norm": 0.1106705293059349,
+      "learning_rate": 0.0001,
+      "loss": 0.1912,
+      "step": 1072
+    },
+    {
+      "epoch": 1.4547180043383947,
+      "grad_norm": 0.1041717529296875,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1073
+    },
+    {
+      "epoch": 1.4560737527114966,
+      "grad_norm": 0.11672812700271606,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 1074
+    },
+    {
+      "epoch": 1.4574295010845986,
+      "grad_norm": 0.1090116798877716,
+      "learning_rate": 0.0001,
+      "loss": 0.1875,
+      "step": 1075
+    },
+    {
+      "epoch": 1.4587852494577007,
+      "grad_norm": 0.12387195974588394,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1076
+    },
+    {
+      "epoch": 1.4601409978308026,
+      "grad_norm": 0.1254459023475647,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 1077
+    },
+    {
+      "epoch": 1.4614967462039046,
+      "grad_norm": 0.1265246719121933,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 1078
+    },
+    {
+      "epoch": 1.4628524945770065,
+      "grad_norm": 0.12745743989944458,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 1079
+    },
+    {
+      "epoch": 1.4642082429501084,
+      "grad_norm": 0.12497106194496155,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 1080
+    },
+    {
+      "epoch": 1.4655639913232104,
+      "grad_norm": 0.142815962433815,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 1081
+    },
+    {
+      "epoch": 1.4669197396963123,
+      "grad_norm": 0.14819836616516113,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 1082
+    },
+    {
+      "epoch": 1.4682754880694144,
+      "grad_norm": 0.13954699039459229,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1083
+    },
+    {
+      "epoch": 1.4696312364425164,
+      "grad_norm": 0.12082433700561523,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1084
+    },
+    {
+      "epoch": 1.4709869848156183,
+      "grad_norm": 0.12150903791189194,
+      "learning_rate": 0.0001,
+      "loss": 0.1848,
+      "step": 1085
+    },
+    {
+      "epoch": 1.4723427331887202,
+      "grad_norm": 0.11382132768630981,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 1086
+    },
+    {
+      "epoch": 1.4736984815618221,
+      "grad_norm": 0.15556840598583221,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1087
+    },
+    {
+      "epoch": 1.475054229934924,
+      "grad_norm": 0.11654501408338547,
+      "learning_rate": 0.0001,
+      "loss": 0.1896,
+      "step": 1088
+    },
+    {
+      "epoch": 1.476409978308026,
+      "grad_norm": 0.12764589488506317,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1089
+    },
+    {
+      "epoch": 1.477765726681128,
+      "grad_norm": 0.11568386107683182,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 1090
+    },
+    {
+      "epoch": 1.4791214750542299,
+      "grad_norm": 0.1260610967874527,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 1091
+    },
+    {
+      "epoch": 1.4804772234273318,
+      "grad_norm": 0.10844352841377258,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 1092
+    },
+    {
+      "epoch": 1.4818329718004337,
+      "grad_norm": 0.13766804337501526,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 1093
+    },
+    {
+      "epoch": 1.4831887201735359,
+      "grad_norm": 0.11210589855909348,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 1094
+    },
+    {
+      "epoch": 1.4845444685466378,
+      "grad_norm": 0.11904489994049072,
+      "learning_rate": 0.0001,
+      "loss": 0.1898,
+      "step": 1095
+    },
+    {
+      "epoch": 1.4859002169197397,
+      "grad_norm": 0.10001026093959808,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1096
+    },
+    {
+      "epoch": 1.4872559652928417,
+      "grad_norm": 0.12232402712106705,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1097
+    },
+    {
+      "epoch": 1.4886117136659436,
+      "grad_norm": 0.11793755739927292,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 1098
+    },
+    {
+      "epoch": 1.4899674620390455,
+      "grad_norm": 0.10772337019443512,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1099
+    },
+    {
+      "epoch": 1.4913232104121474,
+      "grad_norm": 0.10974647849798203,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 1100
+    },
+    {
+      "epoch": 1.4926789587852496,
+      "grad_norm": 0.10750234127044678,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1101
+    },
+    {
+      "epoch": 1.4940347071583515,
+      "grad_norm": 0.11225265264511108,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 1102
+    },
+    {
+      "epoch": 1.4953904555314534,
+      "grad_norm": 0.10960083454847336,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 1103
+    },
+    {
+      "epoch": 1.4967462039045554,
+      "grad_norm": 0.12764433026313782,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1104
+    },
+    {
+      "epoch": 1.4981019522776573,
+      "grad_norm": 0.11526694148778915,
+      "learning_rate": 0.0001,
+      "loss": 0.1905,
+      "step": 1105
+    },
+    {
+      "epoch": 1.4994577006507592,
+      "grad_norm": 0.12409050762653351,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1106
+    },
+    {
+      "epoch": 1.5008134490238612,
+      "grad_norm": 0.11837679147720337,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1107
+    },
+    {
+      "epoch": 1.502169197396963,
+      "grad_norm": 0.11508496850728989,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1108
+    },
+    {
+      "epoch": 1.503524945770065,
+      "grad_norm": 0.11298589408397675,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 1109
+    },
+    {
+      "epoch": 1.504880694143167,
+      "grad_norm": 0.10936693847179413,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1110
+    },
+    {
+      "epoch": 1.5062364425162689,
+      "grad_norm": 0.10975781828165054,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 1111
+    },
+    {
+      "epoch": 1.5075921908893708,
+      "grad_norm": 0.113121896982193,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1112
+    },
+    {
+      "epoch": 1.5089479392624727,
+      "grad_norm": 0.11344970762729645,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1113
+    },
+    {
+      "epoch": 1.5103036876355749,
+      "grad_norm": 0.10153965651988983,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1114
+    },
+    {
+      "epoch": 1.5116594360086768,
+      "grad_norm": 0.11901721358299255,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1115
+    },
+    {
+      "epoch": 1.5130151843817787,
+      "grad_norm": 0.11148644983768463,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1116
+    },
+    {
+      "epoch": 1.5143709327548807,
+      "grad_norm": 0.12646125257015228,
+      "learning_rate": 0.0001,
+      "loss": 0.2008,
+      "step": 1117
+    },
+    {
+      "epoch": 1.5157266811279828,
+      "grad_norm": 0.12142078578472137,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 1118
+    },
+    {
+      "epoch": 1.5170824295010847,
+      "grad_norm": 0.1442292332649231,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 1119
+    },
+    {
+      "epoch": 1.5184381778741867,
+      "grad_norm": 0.10648629069328308,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1120
+    },
+    {
+      "epoch": 1.5197939262472886,
+      "grad_norm": 0.12487778067588806,
+      "learning_rate": 0.0001,
+      "loss": 0.1996,
+      "step": 1121
+    },
+    {
+      "epoch": 1.5211496746203905,
+      "grad_norm": 0.13072475790977478,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1122
+    },
+    {
+      "epoch": 1.5225054229934925,
+      "grad_norm": 0.11304986476898193,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1123
+    },
+    {
+      "epoch": 1.5238611713665944,
+      "grad_norm": 0.11299838125705719,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 1124
+    },
+    {
+      "epoch": 1.5252169197396963,
+      "grad_norm": 0.11574121564626694,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1125
+    },
+    {
+      "epoch": 1.5265726681127982,
+      "grad_norm": 0.10599964112043381,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 1126
+    },
+    {
+      "epoch": 1.5279284164859002,
+      "grad_norm": 0.1266956925392151,
+      "learning_rate": 0.0001,
+      "loss": 0.1866,
+      "step": 1127
+    },
+    {
+      "epoch": 1.529284164859002,
+      "grad_norm": 0.12131784111261368,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1128
+    },
+    {
+      "epoch": 1.530639913232104,
+      "grad_norm": 0.11221503466367722,
+      "learning_rate": 0.0001,
+      "loss": 0.1983,
+      "step": 1129
+    },
+    {
+      "epoch": 1.531995661605206,
+      "grad_norm": 0.10776128619909286,
+      "learning_rate": 0.0001,
+      "loss": 0.2017,
+      "step": 1130
+    },
+    {
+      "epoch": 1.5333514099783079,
+      "grad_norm": 0.12102995067834854,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 1131
+    },
+    {
+      "epoch": 1.53470715835141,
+      "grad_norm": 0.1076129898428917,
+      "learning_rate": 0.0001,
+      "loss": 0.1908,
+      "step": 1132
+    },
+    {
+      "epoch": 1.536062906724512,
+      "grad_norm": 0.12766781449317932,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 1133
+    },
+    {
+      "epoch": 1.5374186550976139,
+      "grad_norm": 0.12022371590137482,
+      "learning_rate": 0.0001,
+      "loss": 0.1802,
+      "step": 1134
+    },
+    {
+      "epoch": 1.5387744034707158,
+      "grad_norm": 0.12678341567516327,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1135
+    },
+    {
+      "epoch": 1.5401301518438177,
+      "grad_norm": 0.12946011126041412,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1136
+    },
+    {
+      "epoch": 1.54148590021692,
+      "grad_norm": 0.1307421773672104,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1137
+    },
+    {
+      "epoch": 1.5428416485900218,
+      "grad_norm": 0.15584075450897217,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1138
+    },
+    {
+      "epoch": 1.5441973969631237,
+      "grad_norm": 0.1196863204240799,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1139
+    },
+    {
+      "epoch": 1.5455531453362257,
+      "grad_norm": 0.11646710336208344,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 1140
+    },
+    {
+      "epoch": 1.5469088937093276,
+      "grad_norm": 0.12545618414878845,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 1141
+    },
+    {
+      "epoch": 1.5482646420824295,
+      "grad_norm": 0.13654284179210663,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 1142
+    },
+    {
+      "epoch": 1.5496203904555315,
+      "grad_norm": 0.15370215475559235,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1143
+    },
+    {
+      "epoch": 1.5509761388286334,
+      "grad_norm": 0.15176653861999512,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 1144
+    },
+    {
+      "epoch": 1.5523318872017353,
+      "grad_norm": 0.13959647715091705,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1145
+    },
+    {
+      "epoch": 1.5536876355748372,
+      "grad_norm": 0.10847233235836029,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 1146
+    },
+    {
+      "epoch": 1.5550433839479392,
+      "grad_norm": 0.12198203802108765,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1147
+    },
+    {
+      "epoch": 1.556399132321041,
+      "grad_norm": 0.10308860242366791,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 1148
+    },
+    {
+      "epoch": 1.557754880694143,
+      "grad_norm": 0.11321444064378738,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1149
+    },
+    {
+      "epoch": 1.559110629067245,
+      "grad_norm": 0.11665124446153641,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 1150
+    },
+    {
+      "epoch": 1.560466377440347,
+      "grad_norm": 0.1418260782957077,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 1151
+    },
+    {
+      "epoch": 1.561822125813449,
+      "grad_norm": 0.11121200770139694,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 1152
+    },
+    {
+      "epoch": 1.563177874186551,
+      "grad_norm": 0.10029853135347366,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1153
+    },
+    {
+      "epoch": 1.564533622559653,
+      "grad_norm": 0.12499584257602692,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 1154
+    },
+    {
+      "epoch": 1.565889370932755,
+      "grad_norm": 0.11592911183834076,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 1155
+    },
+    {
+      "epoch": 1.567245119305857,
+      "grad_norm": 0.23644152283668518,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1156
+    },
+    {
+      "epoch": 1.568600867678959,
+      "grad_norm": 0.13098058104515076,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 1157
+    },
+    {
+      "epoch": 1.5699566160520608,
+      "grad_norm": 0.1155550554394722,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1158
+    },
+    {
+      "epoch": 1.5713123644251628,
+      "grad_norm": 0.10914117842912674,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 1159
+    },
+    {
+      "epoch": 1.5726681127982647,
+      "grad_norm": 0.10114935040473938,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 1160
+    },
+    {
+      "epoch": 1.5740238611713666,
+      "grad_norm": 0.13062158226966858,
+      "learning_rate": 0.0001,
+      "loss": 0.1953,
+      "step": 1161
+    },
+    {
+      "epoch": 1.5753796095444685,
+      "grad_norm": 0.09537646919488907,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 1162
+    },
+    {
+      "epoch": 1.5767353579175705,
+      "grad_norm": 0.11090555787086487,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1163
+    },
+    {
+      "epoch": 1.5780911062906724,
+      "grad_norm": 0.11305756866931915,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 1164
+    },
+    {
+      "epoch": 1.5794468546637743,
+      "grad_norm": 0.10634257644414902,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1165
+    },
+    {
+      "epoch": 1.5808026030368763,
+      "grad_norm": 0.11837907135486603,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 1166
+    },
+    {
+      "epoch": 1.5821583514099782,
+      "grad_norm": 0.10559927672147751,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1167
+    },
+    {
+      "epoch": 1.58351409978308,
+      "grad_norm": 0.11520858854055405,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 1168
+    },
+    {
+      "epoch": 1.5848698481561823,
+      "grad_norm": 0.143623948097229,
+      "learning_rate": 0.0001,
+      "loss": 0.2045,
+      "step": 1169
+    },
+    {
+      "epoch": 1.5862255965292842,
+      "grad_norm": 0.13498550653457642,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1170
+    },
+    {
+      "epoch": 1.5875813449023861,
+      "grad_norm": 0.130161315202713,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1171
+    },
+    {
+      "epoch": 1.588937093275488,
+      "grad_norm": 0.1406034678220749,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1172
+    },
+    {
+      "epoch": 1.59029284164859,
+      "grad_norm": 0.1035957857966423,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1173
+    },
+    {
+      "epoch": 1.5916485900216921,
+      "grad_norm": 0.13619080185890198,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 1174
+    },
+    {
+      "epoch": 1.593004338394794,
+      "grad_norm": 0.10073598474264145,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 1175
+    },
+    {
+      "epoch": 1.594360086767896,
+      "grad_norm": 0.10985419899225235,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 1176
+    },
+    {
+      "epoch": 1.595715835140998,
+      "grad_norm": 0.15524499118328094,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1177
+    },
+    {
+      "epoch": 1.5970715835140998,
+      "grad_norm": 0.13092532753944397,
+      "learning_rate": 0.0001,
+      "loss": 0.2059,
+      "step": 1178
+    },
+    {
+      "epoch": 1.5984273318872018,
+      "grad_norm": 0.08907661586999893,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 1179
+    },
+    {
+      "epoch": 1.5997830802603037,
+      "grad_norm": 0.11854741722345352,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1180
+    },
+    {
+      "epoch": 1.6011388286334056,
+      "grad_norm": 0.11499210447072983,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1181
+    },
+    {
+      "epoch": 1.6024945770065075,
+      "grad_norm": 0.11054805666208267,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 1182
+    },
+    {
+      "epoch": 1.6038503253796095,
+      "grad_norm": 0.10097283124923706,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 1183
+    },
+    {
+      "epoch": 1.6052060737527114,
+      "grad_norm": 0.11178874224424362,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 1184
+    },
+    {
+      "epoch": 1.6065618221258133,
+      "grad_norm": 0.2234911471605301,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 1185
+    },
+    {
+      "epoch": 1.6079175704989153,
+      "grad_norm": 0.11132630705833435,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 1186
+    },
+    {
+      "epoch": 1.6092733188720172,
+      "grad_norm": 0.11684749275445938,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 1187
+    },
+    {
+      "epoch": 1.6106290672451193,
+      "grad_norm": 0.10871650278568268,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 1188
+    },
+    {
+      "epoch": 1.6119848156182213,
+      "grad_norm": 0.12698756158351898,
+      "learning_rate": 0.0001,
+      "loss": 0.1954,
+      "step": 1189
+    },
+    {
+      "epoch": 1.6133405639913232,
+      "grad_norm": 0.12337315082550049,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 1190
+    },
+    {
+      "epoch": 1.6146963123644251,
+      "grad_norm": 0.11370830237865448,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1191
+    },
+    {
+      "epoch": 1.6160520607375273,
+      "grad_norm": 0.13150843977928162,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 1192
+    },
+    {
+      "epoch": 1.6174078091106292,
+      "grad_norm": 0.13284140825271606,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 1193
+    },
+    {
+      "epoch": 1.6187635574837311,
+      "grad_norm": 0.11413049697875977,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1194
+    },
+    {
+      "epoch": 1.620119305856833,
+      "grad_norm": 0.13481928408145905,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 1195
+    },
+    {
+      "epoch": 1.621475054229935,
+      "grad_norm": 0.131308913230896,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 1196
+    },
+    {
+      "epoch": 1.622830802603037,
+      "grad_norm": 0.13984714448451996,
+      "learning_rate": 0.0001,
+      "loss": 0.1983,
+      "step": 1197
+    },
+    {
+      "epoch": 1.6241865509761388,
+      "grad_norm": 0.13194681704044342,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1198
+    },
+    {
+      "epoch": 1.6255422993492408,
+      "grad_norm": 0.11310645192861557,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 1199
+    },
+    {
+      "epoch": 1.6268980477223427,
+      "grad_norm": 0.12191493064165115,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1200
+    },
+    {
+      "epoch": 1.6282537960954446,
+      "grad_norm": 0.12168656289577484,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 1201
+    },
+    {
+      "epoch": 1.6296095444685466,
+      "grad_norm": 0.10614129900932312,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1202
+    },
+    {
+      "epoch": 1.6309652928416485,
+      "grad_norm": 0.12460123747587204,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 1203
+    },
+    {
+      "epoch": 1.6323210412147504,
+      "grad_norm": 0.11379490792751312,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 1204
+    },
+    {
+      "epoch": 1.6336767895878523,
+      "grad_norm": 0.12374304980039597,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1205
+    },
+    {
+      "epoch": 1.6350325379609545,
+      "grad_norm": 0.11722256243228912,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1206
+    },
+    {
+      "epoch": 1.6363882863340564,
+      "grad_norm": 0.11973638087511063,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 1207
+    },
+    {
+      "epoch": 1.6377440347071583,
+      "grad_norm": 0.10788135230541229,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1208
+    },
+    {
+      "epoch": 1.6390997830802603,
+      "grad_norm": 0.10746819525957108,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1209
+    },
+    {
+      "epoch": 1.6404555314533622,
+      "grad_norm": 0.13546374440193176,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 1210
+    },
+    {
+      "epoch": 1.6418112798264644,
+      "grad_norm": 0.11627724766731262,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1211
+    },
+    {
+      "epoch": 1.6431670281995663,
+      "grad_norm": 0.15665557980537415,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 1212
+    },
+    {
+      "epoch": 1.6445227765726682,
+      "grad_norm": 0.1165895015001297,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1213
+    },
+    {
+      "epoch": 1.6458785249457701,
+      "grad_norm": 0.13061290979385376,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1214
+    },
+    {
+      "epoch": 1.647234273318872,
+      "grad_norm": 0.1269635111093521,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 1215
+    },
+    {
+      "epoch": 1.648590021691974,
+      "grad_norm": 0.11976515501737595,
+      "learning_rate": 0.0001,
+      "loss": 0.1786,
+      "step": 1216
+    },
+    {
+      "epoch": 1.649945770065076,
+      "grad_norm": 0.11105464398860931,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 1217
+    },
+    {
+      "epoch": 1.6513015184381779,
+      "grad_norm": 0.12581010162830353,
+      "learning_rate": 0.0001,
+      "loss": 0.1983,
+      "step": 1218
+    },
+    {
+      "epoch": 1.6526572668112798,
+      "grad_norm": 0.1230585128068924,
+      "learning_rate": 0.0001,
+      "loss": 0.1888,
+      "step": 1219
+    },
+    {
+      "epoch": 1.6540130151843817,
+      "grad_norm": 0.11903957277536392,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1220
+    },
+    {
+      "epoch": 1.6553687635574836,
+      "grad_norm": 0.14139562845230103,
+      "learning_rate": 0.0001,
+      "loss": 0.2131,
+      "step": 1221
+    },
+    {
+      "epoch": 1.6567245119305856,
+      "grad_norm": 0.1173047348856926,
+      "learning_rate": 0.0001,
+      "loss": 0.2111,
+      "step": 1222
+    },
+    {
+      "epoch": 1.6580802603036875,
+      "grad_norm": 0.11156854778528214,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1223
+    },
+    {
+      "epoch": 1.6594360086767896,
+      "grad_norm": 0.12952075898647308,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 1224
+    },
+    {
+      "epoch": 1.6607917570498916,
+      "grad_norm": 0.10950185358524323,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1225
+    },
+    {
+      "epoch": 1.6621475054229935,
+      "grad_norm": 0.12041502445936203,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 1226
+    },
+    {
+      "epoch": 1.6635032537960954,
+      "grad_norm": 0.1415816694498062,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1227
+    },
+    {
+      "epoch": 1.6648590021691974,
+      "grad_norm": 0.12750586867332458,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1228
+    },
+    {
+      "epoch": 1.6662147505422995,
+      "grad_norm": 0.10851788520812988,
+      "learning_rate": 0.0001,
+      "loss": 0.1955,
+      "step": 1229
+    },
+    {
+      "epoch": 1.6675704989154014,
+      "grad_norm": 0.11707224696874619,
+      "learning_rate": 0.0001,
+      "loss": 0.1918,
+      "step": 1230
+    },
+    {
+      "epoch": 1.6689262472885034,
+      "grad_norm": 0.11676563322544098,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1231
+    },
+    {
+      "epoch": 1.6702819956616053,
+      "grad_norm": 0.10895233601331711,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1232
+    },
+    {
+      "epoch": 1.6716377440347072,
+      "grad_norm": 0.11895432323217392,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1233
+    },
+    {
+      "epoch": 1.6729934924078091,
+      "grad_norm": 0.13438113033771515,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1234
+    },
+    {
+      "epoch": 1.674349240780911,
+      "grad_norm": 0.1250566691160202,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 1235
+    },
+    {
+      "epoch": 1.675704989154013,
+      "grad_norm": 0.14335834980010986,
+      "learning_rate": 0.0001,
+      "loss": 0.1895,
+      "step": 1236
+    },
+    {
+      "epoch": 1.677060737527115,
+      "grad_norm": 0.11735735088586807,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1237
+    },
+    {
+      "epoch": 1.6784164859002169,
+      "grad_norm": 0.1468167006969452,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 1238
+    },
+    {
+      "epoch": 1.6797722342733188,
+      "grad_norm": 0.1282537281513214,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 1239
+    },
+    {
+      "epoch": 1.6811279826464207,
+      "grad_norm": 0.12504026293754578,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 1240
+    },
+    {
+      "epoch": 1.6824837310195226,
+      "grad_norm": 0.11548586934804916,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1241
+    },
+    {
+      "epoch": 1.6838394793926246,
+      "grad_norm": 0.10258237272500992,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 1242
+    },
+    {
+      "epoch": 1.6851952277657267,
+      "grad_norm": 0.1402621865272522,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 1243
+    },
+    {
+      "epoch": 1.6865509761388287,
+      "grad_norm": 0.11828981339931488,
+      "learning_rate": 0.0001,
+      "loss": 0.1941,
+      "step": 1244
+    },
+    {
+      "epoch": 1.6879067245119306,
+      "grad_norm": 0.10513028502464294,
+      "learning_rate": 0.0001,
+      "loss": 0.1907,
+      "step": 1245
+    },
+    {
+      "epoch": 1.6892624728850325,
+      "grad_norm": 0.14414629340171814,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 1246
+    },
+    {
+      "epoch": 1.6906182212581344,
+      "grad_norm": 0.11393895000219345,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 1247
+    },
+    {
+      "epoch": 1.6919739696312366,
+      "grad_norm": 0.11836519092321396,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1248
+    },
+    {
+      "epoch": 1.6933297180043385,
+      "grad_norm": 0.13914765417575836,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 1249
+    },
+    {
+      "epoch": 1.6946854663774404,
+      "grad_norm": 0.13182178139686584,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 1250
+    },
+    {
+      "epoch": 1.6960412147505424,
+      "grad_norm": 0.12725161015987396,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 1251
+    },
+    {
+      "epoch": 1.6973969631236443,
+      "grad_norm": 0.14727424085140228,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 1252
+    },
+    {
+      "epoch": 1.6987527114967462,
+      "grad_norm": 0.1276342123746872,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 1253
+    },
+    {
+      "epoch": 1.7001084598698482,
+      "grad_norm": 0.11127869039773941,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 1254
+    },
+    {
+      "epoch": 1.70146420824295,
+      "grad_norm": 0.12156875431537628,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 1255
+    },
+    {
+      "epoch": 1.702819956616052,
+      "grad_norm": 0.12866662442684174,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1256
+    },
+    {
+      "epoch": 1.704175704989154,
+      "grad_norm": 0.1236620843410492,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1257
+    },
+    {
+      "epoch": 1.7055314533622559,
+      "grad_norm": 0.12631292641162872,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1258
+    },
+    {
+      "epoch": 1.7068872017353578,
+      "grad_norm": 0.11388317495584488,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1259
+    },
+    {
+      "epoch": 1.7082429501084597,
+      "grad_norm": 0.12774516642093658,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 1260
+    },
+    {
+      "epoch": 1.7095986984815619,
+      "grad_norm": 0.14858609437942505,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 1261
+    },
+    {
+      "epoch": 1.7109544468546638,
+      "grad_norm": 0.12573033571243286,
+      "learning_rate": 0.0001,
+      "loss": 0.1999,
+      "step": 1262
+    },
+    {
+      "epoch": 1.7123101952277657,
+      "grad_norm": 0.12592986226081848,
+      "learning_rate": 0.0001,
+      "loss": 0.2002,
+      "step": 1263
+    },
+    {
+      "epoch": 1.7136659436008677,
+      "grad_norm": 0.11726995557546616,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 1264
+    },
+    {
+      "epoch": 1.7150216919739696,
+      "grad_norm": 0.18719233572483063,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 1265
+    },
+    {
+      "epoch": 1.7163774403470717,
+      "grad_norm": 0.11020854860544205,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1266
+    },
+    {
+      "epoch": 1.7177331887201737,
+      "grad_norm": 0.11451730877161026,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1267
+    },
+    {
+      "epoch": 1.7190889370932756,
+      "grad_norm": 0.11634907871484756,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 1268
+    },
+    {
+      "epoch": 1.7204446854663775,
+      "grad_norm": 0.10934798419475555,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1269
+    },
+    {
+      "epoch": 1.7218004338394794,
+      "grad_norm": 0.13651181757450104,
+      "learning_rate": 0.0001,
+      "loss": 0.1877,
+      "step": 1270
+    },
+    {
+      "epoch": 1.7231561822125814,
+      "grad_norm": 0.11215245723724365,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1271
+    },
+    {
+      "epoch": 1.7245119305856833,
+      "grad_norm": 0.14050449430942535,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1272
+    },
+    {
+      "epoch": 1.7258676789587852,
+      "grad_norm": 0.13667066395282745,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1273
+    },
+    {
+      "epoch": 1.7272234273318872,
+      "grad_norm": 0.14513789117336273,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 1274
+    },
+    {
+      "epoch": 1.728579175704989,
+      "grad_norm": 0.12057226896286011,
+      "learning_rate": 0.0001,
+      "loss": 0.1992,
+      "step": 1275
+    },
+    {
+      "epoch": 1.729934924078091,
+      "grad_norm": 0.12683060765266418,
+      "learning_rate": 0.0001,
+      "loss": 0.2016,
+      "step": 1276
+    },
+    {
+      "epoch": 1.731290672451193,
+      "grad_norm": 0.11656946688890457,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 1277
+    },
+    {
+      "epoch": 1.7326464208242949,
+      "grad_norm": 0.12544988095760345,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 1278
+    },
+    {
+      "epoch": 1.7340021691973968,
+      "grad_norm": 0.14030902087688446,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 1279
+    },
+    {
+      "epoch": 1.735357917570499,
+      "grad_norm": 0.11102521419525146,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1280
+    },
+    {
+      "epoch": 1.7367136659436009,
+      "grad_norm": 0.11298868805170059,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 1281
+    },
+    {
+      "epoch": 1.7380694143167028,
+      "grad_norm": 0.12173949182033539,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 1282
+    },
+    {
+      "epoch": 1.7394251626898047,
+      "grad_norm": 0.1329275667667389,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1283
+    },
+    {
+      "epoch": 1.7407809110629069,
+      "grad_norm": 0.13701599836349487,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 1284
+    },
+    {
+      "epoch": 1.7421366594360088,
+      "grad_norm": 0.12843158841133118,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1285
+    },
+    {
+      "epoch": 1.7434924078091107,
+      "grad_norm": 0.11957607418298721,
+      "learning_rate": 0.0001,
+      "loss": 0.1894,
+      "step": 1286
+    },
+    {
+      "epoch": 1.7448481561822127,
+      "grad_norm": 0.11455661803483963,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 1287
+    },
+    {
+      "epoch": 1.7462039045553146,
+      "grad_norm": 0.0998346358537674,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1288
+    },
+    {
+      "epoch": 1.7475596529284165,
+      "grad_norm": 0.1376727670431137,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 1289
+    },
+    {
+      "epoch": 1.7489154013015185,
+      "grad_norm": 0.11547600477933884,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1290
+    },
+    {
+      "epoch": 1.7502711496746204,
+      "grad_norm": 0.1405201405286789,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1291
+    },
+    {
+      "epoch": 1.7516268980477223,
+      "grad_norm": 0.1275051236152649,
+      "learning_rate": 0.0001,
+      "loss": 0.1899,
+      "step": 1292
+    },
+    {
+      "epoch": 1.7529826464208242,
+      "grad_norm": 0.11436183750629425,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1293
+    },
+    {
+      "epoch": 1.7543383947939262,
+      "grad_norm": 0.10620865970849991,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1294
+    },
+    {
+      "epoch": 1.755694143167028,
+      "grad_norm": 0.13133226335048676,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 1295
+    },
+    {
+      "epoch": 1.75704989154013,
+      "grad_norm": 0.1284555345773697,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 1296
+    },
+    {
+      "epoch": 1.758405639913232,
+      "grad_norm": 0.13066278398036957,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1297
+    },
+    {
+      "epoch": 1.759761388286334,
+      "grad_norm": 0.11201050877571106,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1298
+    },
+    {
+      "epoch": 1.761117136659436,
+      "grad_norm": 0.13508011400699615,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 1299
+    },
+    {
+      "epoch": 1.762472885032538,
+      "grad_norm": 0.14097465574741364,
+      "learning_rate": 0.0001,
+      "loss": 0.2241,
+      "step": 1300
+    },
+    {
+      "epoch": 1.76382863340564,
+      "grad_norm": 0.12292483448982239,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 1301
+    },
+    {
+      "epoch": 1.7651843817787418,
+      "grad_norm": 0.11659924685955048,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1302
+    },
+    {
+      "epoch": 1.766540130151844,
+      "grad_norm": 0.10861191898584366,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1303
+    },
+    {
+      "epoch": 1.767895878524946,
+      "grad_norm": 0.12461470067501068,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1304
+    },
+    {
+      "epoch": 1.7692516268980478,
+      "grad_norm": 0.1242380440235138,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 1305
+    },
+    {
+      "epoch": 1.7706073752711498,
+      "grad_norm": 0.12022621184587479,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 1306
+    },
+    {
+      "epoch": 1.7719631236442517,
+      "grad_norm": 0.11899636685848236,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 1307
+    },
+    {
+      "epoch": 1.7733188720173536,
+      "grad_norm": 0.1083032563328743,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1308
+    },
+    {
+      "epoch": 1.7746746203904555,
+      "grad_norm": 0.13143344223499298,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1309
+    },
+    {
+      "epoch": 1.7760303687635575,
+      "grad_norm": 0.13755397498607635,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1310
+    },
+    {
+      "epoch": 1.7773861171366594,
+      "grad_norm": 0.1269962340593338,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1311
+    },
+    {
+      "epoch": 1.7787418655097613,
+      "grad_norm": 0.1279396116733551,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1312
+    },
+    {
+      "epoch": 1.7800976138828633,
+      "grad_norm": 0.1349986493587494,
+      "learning_rate": 0.0001,
+      "loss": 0.2113,
+      "step": 1313
+    },
+    {
+      "epoch": 1.7814533622559652,
+      "grad_norm": 0.12752625346183777,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 1314
+    },
+    {
+      "epoch": 1.782809110629067,
+      "grad_norm": 0.11171070486307144,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1315
+    },
+    {
+      "epoch": 1.784164859002169,
+      "grad_norm": 0.1097131296992302,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1316
+    },
+    {
+      "epoch": 1.7855206073752712,
+      "grad_norm": 0.12325100600719452,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 1317
+    },
+    {
+      "epoch": 1.7868763557483731,
+      "grad_norm": 0.1229829490184784,
+      "learning_rate": 0.0001,
+      "loss": 0.1885,
+      "step": 1318
+    },
+    {
+      "epoch": 1.788232104121475,
+      "grad_norm": 0.12175952643156052,
+      "learning_rate": 0.0001,
+      "loss": 0.1895,
+      "step": 1319
+    },
+    {
+      "epoch": 1.789587852494577,
+      "grad_norm": 0.1383509337902069,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 1320
+    },
+    {
+      "epoch": 1.7909436008676791,
+      "grad_norm": 0.12128841131925583,
+      "learning_rate": 0.0001,
+      "loss": 0.192,
+      "step": 1321
+    },
+    {
+      "epoch": 1.792299349240781,
+      "grad_norm": 0.11734595149755478,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1322
+    },
+    {
+      "epoch": 1.793655097613883,
+      "grad_norm": 0.14504635334014893,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1323
+    },
+    {
+      "epoch": 1.795010845986985,
+      "grad_norm": 0.14949192106723785,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 1324
+    },
+    {
+      "epoch": 1.7963665943600868,
+      "grad_norm": 0.11390583962202072,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 1325
+    },
+    {
+      "epoch": 1.7977223427331888,
+      "grad_norm": 0.12815044820308685,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 1326
+    },
+    {
+      "epoch": 1.7990780911062907,
+      "grad_norm": 0.10549808293581009,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 1327
+    },
+    {
+      "epoch": 1.8004338394793926,
+      "grad_norm": 0.11771165579557419,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 1328
+    },
+    {
+      "epoch": 1.8017895878524945,
+      "grad_norm": 0.11718256026506424,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1329
+    },
+    {
+      "epoch": 1.8031453362255965,
+      "grad_norm": 0.12123684585094452,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1330
+    },
+    {
+      "epoch": 1.8045010845986984,
+      "grad_norm": 0.1800345778465271,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 1331
+    },
+    {
+      "epoch": 1.8058568329718003,
+      "grad_norm": 0.14188869297504425,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 1332
+    },
+    {
+      "epoch": 1.8072125813449023,
+      "grad_norm": 0.12359337508678436,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1333
+    },
+    {
+      "epoch": 1.8085683297180042,
+      "grad_norm": 0.11946556717157364,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1334
+    },
+    {
+      "epoch": 1.8099240780911063,
+      "grad_norm": 0.13021409511566162,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 1335
+    },
+    {
+      "epoch": 1.8112798264642083,
+      "grad_norm": 0.11907394230365753,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 1336
+    },
+    {
+      "epoch": 1.8126355748373102,
+      "grad_norm": 0.12994050979614258,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 1337
+    },
+    {
+      "epoch": 1.8139913232104121,
+      "grad_norm": 0.1035689115524292,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 1338
+    },
+    {
+      "epoch": 1.815347071583514,
+      "grad_norm": 0.11135593056678772,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1339
+    },
+    {
+      "epoch": 1.8167028199566162,
+      "grad_norm": 0.11486642807722092,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 1340
+    },
+    {
+      "epoch": 1.8180585683297181,
+      "grad_norm": 0.14242038130760193,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 1341
+    },
+    {
+      "epoch": 1.81941431670282,
+      "grad_norm": 0.12918058037757874,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1342
+    },
+    {
+      "epoch": 1.820770065075922,
+      "grad_norm": 0.12029004096984863,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1343
+    },
+    {
+      "epoch": 1.822125813449024,
+      "grad_norm": 0.26643913984298706,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1344
+    },
+    {
+      "epoch": 1.8234815618221258,
+      "grad_norm": 0.09846089035272598,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 1345
+    },
+    {
+      "epoch": 1.8248373101952278,
+      "grad_norm": 0.10248425602912903,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1346
+    },
+    {
+      "epoch": 1.8261930585683297,
+      "grad_norm": 0.12435374408960342,
+      "learning_rate": 0.0001,
+      "loss": 0.2072,
+      "step": 1347
+    },
+    {
+      "epoch": 1.8275488069414316,
+      "grad_norm": 0.13368810713291168,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 1348
+    },
+    {
+      "epoch": 1.8289045553145336,
+      "grad_norm": 0.1289181113243103,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1349
+    },
+    {
+      "epoch": 1.8302603036876355,
+      "grad_norm": 0.10527734458446503,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 1350
+    },
+    {
+      "epoch": 1.8316160520607374,
+      "grad_norm": 0.11010352522134781,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 1351
+    },
+    {
+      "epoch": 1.8329718004338393,
+      "grad_norm": 0.12538234889507294,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 1352
+    },
+    {
+      "epoch": 1.8343275488069413,
+      "grad_norm": 0.12591451406478882,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 1353
+    },
+    {
+      "epoch": 1.8356832971800434,
+      "grad_norm": 0.10815133154392242,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1354
+    },
+    {
+      "epoch": 1.8370390455531453,
+      "grad_norm": 0.21128854155540466,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1355
+    },
+    {
+      "epoch": 1.8383947939262473,
+      "grad_norm": 0.10303530097007751,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 1356
+    },
+    {
+      "epoch": 1.8397505422993492,
+      "grad_norm": 0.11695558577775955,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1357
+    },
+    {
+      "epoch": 1.8411062906724514,
+      "grad_norm": 0.1205521896481514,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1358
+    },
+    {
+      "epoch": 1.8424620390455533,
+      "grad_norm": 0.11526507884263992,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1359
+    },
+    {
+      "epoch": 1.8438177874186552,
+      "grad_norm": 0.11606362462043762,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1360
+    },
+    {
+      "epoch": 1.8451735357917571,
+      "grad_norm": 0.13574472069740295,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1361
+    },
+    {
+      "epoch": 1.846529284164859,
+      "grad_norm": 0.1354644000530243,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 1362
+    },
+    {
+      "epoch": 1.847885032537961,
+      "grad_norm": 0.12681134045124054,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 1363
+    },
+    {
+      "epoch": 1.849240780911063,
+      "grad_norm": 0.11985041946172714,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 1364
+    },
+    {
+      "epoch": 1.8505965292841648,
+      "grad_norm": 0.10991258919239044,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1365
+    },
+    {
+      "epoch": 1.8519522776572668,
+      "grad_norm": 0.12147706001996994,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1366
+    },
+    {
+      "epoch": 1.8533080260303687,
+      "grad_norm": 0.12858495116233826,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1367
+    },
+    {
+      "epoch": 1.8546637744034706,
+      "grad_norm": 0.09493502229452133,
+      "learning_rate": 0.0001,
+      "loss": 0.1508,
+      "step": 1368
+    },
+    {
+      "epoch": 1.8560195227765726,
+      "grad_norm": 0.11763230711221695,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1369
+    },
+    {
+      "epoch": 1.8573752711496745,
+      "grad_norm": 0.13290619850158691,
+      "learning_rate": 0.0001,
+      "loss": 0.192,
+      "step": 1370
+    },
+    {
+      "epoch": 1.8587310195227764,
+      "grad_norm": 0.12978313863277435,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 1371
+    },
+    {
+      "epoch": 1.8600867678958786,
+      "grad_norm": 0.12653270363807678,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 1372
+    },
+    {
+      "epoch": 1.8614425162689805,
+      "grad_norm": 0.14263620972633362,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 1373
+    },
+    {
+      "epoch": 1.8627982646420824,
+      "grad_norm": 0.1400168240070343,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1374
+    },
+    {
+      "epoch": 1.8641540130151844,
+      "grad_norm": 0.17182239890098572,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 1375
+    },
+    {
+      "epoch": 1.8655097613882863,
+      "grad_norm": 0.15442930161952972,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 1376
+    },
+    {
+      "epoch": 1.8668655097613884,
+      "grad_norm": 0.10439960658550262,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1377
+    },
+    {
+      "epoch": 1.8682212581344904,
+      "grad_norm": 0.1089179515838623,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1378
+    },
+    {
+      "epoch": 1.8695770065075923,
+      "grad_norm": 0.1283685564994812,
+      "learning_rate": 0.0001,
+      "loss": 0.2022,
+      "step": 1379
+    },
+    {
+      "epoch": 1.8709327548806942,
+      "grad_norm": 0.10410815477371216,
+      "learning_rate": 0.0001,
+      "loss": 0.1759,
+      "step": 1380
+    },
+    {
+      "epoch": 1.8722885032537961,
+      "grad_norm": 0.10370563715696335,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 1381
+    },
+    {
+      "epoch": 1.873644251626898,
+      "grad_norm": 0.15194657444953918,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1382
+    },
+    {
+      "epoch": 1.875,
+      "grad_norm": 0.1234624832868576,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 1383
+    },
+    {
+      "epoch": 1.876355748373102,
+      "grad_norm": 0.12981152534484863,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 1384
+    },
+    {
+      "epoch": 1.8777114967462039,
+      "grad_norm": 0.1425439864397049,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 1385
+    },
+    {
+      "epoch": 1.8790672451193058,
+      "grad_norm": 0.1259755641222,
+      "learning_rate": 0.0001,
+      "loss": 0.2043,
+      "step": 1386
+    },
+    {
+      "epoch": 1.8804229934924077,
+      "grad_norm": 0.15680141746997833,
+      "learning_rate": 0.0001,
+      "loss": 0.1978,
+      "step": 1387
+    },
+    {
+      "epoch": 1.8817787418655096,
+      "grad_norm": 0.11902142316102982,
+      "learning_rate": 0.0001,
+      "loss": 0.1915,
+      "step": 1388
+    },
+    {
+      "epoch": 1.8831344902386116,
+      "grad_norm": 0.12371156364679337,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 1389
+    },
+    {
+      "epoch": 1.8844902386117137,
+      "grad_norm": 0.13596239686012268,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1390
+    },
+    {
+      "epoch": 1.8858459869848156,
+      "grad_norm": 0.13073337078094482,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1391
+    },
+    {
+      "epoch": 1.8872017353579176,
+      "grad_norm": 0.1322539895772934,
+      "learning_rate": 0.0001,
+      "loss": 0.1959,
+      "step": 1392
+    },
+    {
+      "epoch": 1.8885574837310195,
+      "grad_norm": 0.12941564619541168,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 1393
+    },
+    {
+      "epoch": 1.8899132321041214,
+      "grad_norm": 0.11447383463382721,
+      "learning_rate": 0.0001,
+      "loss": 0.1904,
+      "step": 1394
+    },
+    {
+      "epoch": 1.8912689804772236,
+      "grad_norm": 0.11330213397741318,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1395
+    },
+    {
+      "epoch": 1.8926247288503255,
+      "grad_norm": 0.14319045841693878,
+      "learning_rate": 0.0001,
+      "loss": 0.1923,
+      "step": 1396
+    },
+    {
+      "epoch": 1.8939804772234274,
+      "grad_norm": 0.12432601302862167,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 1397
+    },
+    {
+      "epoch": 1.8953362255965294,
+      "grad_norm": 0.11268299072980881,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1398
+    },
+    {
+      "epoch": 1.8966919739696313,
+      "grad_norm": 0.1071951761841774,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 1399
+    },
+    {
+      "epoch": 1.8980477223427332,
+      "grad_norm": 0.10361482203006744,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 1400
+    },
+    {
+      "epoch": 1.8994034707158352,
+      "grad_norm": 0.13007575273513794,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 1401
+    },
+    {
+      "epoch": 1.900759219088937,
+      "grad_norm": 0.10795672237873077,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1402
+    },
+    {
+      "epoch": 1.902114967462039,
+      "grad_norm": 0.10109420120716095,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 1403
+    },
+    {
+      "epoch": 1.903470715835141,
+      "grad_norm": 0.11790471524000168,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1404
+    },
+    {
+      "epoch": 1.9048264642082429,
+      "grad_norm": 0.1457929164171219,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 1405
+    },
+    {
+      "epoch": 1.9061822125813448,
+      "grad_norm": 0.11985544115304947,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 1406
+    },
+    {
+      "epoch": 1.9075379609544467,
+      "grad_norm": 0.10876721888780594,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1407
+    },
+    {
+      "epoch": 1.9088937093275486,
+      "grad_norm": 0.12687523663043976,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1408
+    },
+    {
+      "epoch": 1.9102494577006508,
+      "grad_norm": 0.13659699261188507,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 1409
+    },
+    {
+      "epoch": 1.9116052060737527,
+      "grad_norm": 0.15051300823688507,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1410
+    },
+    {
+      "epoch": 1.9129609544468547,
+      "grad_norm": 0.12626850605010986,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 1411
+    },
+    {
+      "epoch": 1.9143167028199566,
+      "grad_norm": 0.12241556495428085,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1412
+    },
+    {
+      "epoch": 1.9156724511930587,
+      "grad_norm": 0.21131230890750885,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1413
+    },
+    {
+      "epoch": 1.9170281995661607,
+      "grad_norm": 0.12205749750137329,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1414
+    },
+    {
+      "epoch": 1.9183839479392626,
+      "grad_norm": 0.09610819071531296,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1415
+    },
+    {
+      "epoch": 1.9197396963123645,
+      "grad_norm": 0.11469210684299469,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 1416
+    },
+    {
+      "epoch": 1.9210954446854664,
+      "grad_norm": 0.11049284040927887,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1417
+    },
+    {
+      "epoch": 1.9224511930585684,
+      "grad_norm": 0.11098019778728485,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1418
+    },
+    {
+      "epoch": 1.9238069414316703,
+      "grad_norm": 0.09846334159374237,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1419
+    },
+    {
+      "epoch": 1.9251626898047722,
+      "grad_norm": 0.10705545544624329,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 1420
+    },
+    {
+      "epoch": 1.9265184381778742,
+      "grad_norm": 0.1294596940279007,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 1421
+    },
+    {
+      "epoch": 1.927874186550976,
+      "grad_norm": 0.1301390528678894,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 1422
+    },
+    {
+      "epoch": 1.929229934924078,
+      "grad_norm": 0.11075222492218018,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1423
+    },
+    {
+      "epoch": 1.93058568329718,
+      "grad_norm": 0.11881387233734131,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 1424
+    },
+    {
+      "epoch": 1.9319414316702819,
+      "grad_norm": 0.12189192324876785,
+      "learning_rate": 0.0001,
+      "loss": 0.1965,
+      "step": 1425
+    },
+    {
+      "epoch": 1.9332971800433838,
+      "grad_norm": 0.12628984451293945,
+      "learning_rate": 0.0001,
+      "loss": 0.1979,
+      "step": 1426
+    },
+    {
+      "epoch": 1.934652928416486,
+      "grad_norm": 0.12321808934211731,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 1427
+    },
+    {
+      "epoch": 1.9360086767895879,
+      "grad_norm": 0.11550022661685944,
+      "learning_rate": 0.0001,
+      "loss": 0.2,
+      "step": 1428
+    },
+    {
+      "epoch": 1.9373644251626898,
+      "grad_norm": 0.10776833444833755,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 1429
+    },
+    {
+      "epoch": 1.9387201735357917,
+      "grad_norm": 0.13376648724079132,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 1430
+    },
+    {
+      "epoch": 1.9400759219088937,
+      "grad_norm": 0.11048057675361633,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 1431
+    },
+    {
+      "epoch": 1.9414316702819958,
+      "grad_norm": 0.12688471376895905,
+      "learning_rate": 0.0001,
+      "loss": 0.1851,
+      "step": 1432
+    },
+    {
+      "epoch": 1.9427874186550977,
+      "grad_norm": 0.11357000470161438,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 1433
+    },
+    {
+      "epoch": 1.9441431670281997,
+      "grad_norm": 0.11683011054992676,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1434
+    },
+    {
+      "epoch": 1.9454989154013016,
+      "grad_norm": 0.10903217643499374,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1435
+    },
+    {
+      "epoch": 1.9468546637744035,
+      "grad_norm": 0.11643415689468384,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 1436
+    },
+    {
+      "epoch": 1.9482104121475055,
+      "grad_norm": 0.10832429677248001,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 1437
+    },
+    {
+      "epoch": 1.9495661605206074,
+      "grad_norm": 0.13552255928516388,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1438
+    },
+    {
+      "epoch": 1.9509219088937093,
+      "grad_norm": 0.12309954315423965,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 1439
+    },
+    {
+      "epoch": 1.9522776572668112,
+      "grad_norm": 0.12863095104694366,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1440
+    },
+    {
+      "epoch": 1.9536334056399132,
+      "grad_norm": 0.13045132160186768,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 1441
+    },
+    {
+      "epoch": 1.954989154013015,
+      "grad_norm": 0.12640979886054993,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1442
+    },
+    {
+      "epoch": 1.956344902386117,
+      "grad_norm": 0.1256365180015564,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 1443
+    },
+    {
+      "epoch": 1.957700650759219,
+      "grad_norm": 0.12674441933631897,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1444
+    },
+    {
+      "epoch": 1.9590563991323209,
+      "grad_norm": 0.1102062538266182,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1445
+    },
+    {
+      "epoch": 1.960412147505423,
+      "grad_norm": 0.128218874335289,
+      "learning_rate": 0.0001,
+      "loss": 0.1854,
+      "step": 1446
+    },
+    {
+      "epoch": 1.961767895878525,
+      "grad_norm": 0.11969443410634995,
+      "learning_rate": 0.0001,
+      "loss": 0.1947,
+      "step": 1447
+    },
+    {
+      "epoch": 1.9631236442516269,
+      "grad_norm": 0.10714039951562881,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 1448
+    },
+    {
+      "epoch": 1.9644793926247288,
+      "grad_norm": 0.10589423030614853,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1449
+    },
+    {
+      "epoch": 1.965835140997831,
+      "grad_norm": 0.12388709932565689,
+      "learning_rate": 0.0001,
+      "loss": 0.1865,
+      "step": 1450
+    },
+    {
+      "epoch": 1.967190889370933,
+      "grad_norm": 0.11404157429933548,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1451
+    },
+    {
+      "epoch": 1.9685466377440348,
+      "grad_norm": 0.1288078874349594,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1452
+    },
+    {
+      "epoch": 1.9699023861171367,
+      "grad_norm": 0.12317921221256256,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1453
+    },
+    {
+      "epoch": 1.9712581344902387,
+      "grad_norm": 0.12275351583957672,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1454
+    },
+    {
+      "epoch": 1.9726138828633406,
+      "grad_norm": 0.10890765488147736,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 1455
+    },
+    {
+      "epoch": 1.9739696312364425,
+      "grad_norm": 0.14234666526317596,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 1456
+    },
+    {
+      "epoch": 1.9753253796095445,
+      "grad_norm": 0.11150006204843521,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 1457
+    },
+    {
+      "epoch": 1.9766811279826464,
+      "grad_norm": 0.1094827875494957,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1458
+    },
+    {
+      "epoch": 1.9780368763557483,
+      "grad_norm": 0.11834603548049927,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1459
+    },
+    {
+      "epoch": 1.9793926247288502,
+      "grad_norm": 0.11959917098283768,
+      "learning_rate": 0.0001,
+      "loss": 0.1884,
+      "step": 1460
+    },
+    {
+      "epoch": 1.9807483731019522,
+      "grad_norm": 0.14127543568611145,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 1461
+    },
+    {
+      "epoch": 1.982104121475054,
+      "grad_norm": 0.13468819856643677,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 1462
+    },
+    {
+      "epoch": 1.983459869848156,
+      "grad_norm": 0.11925316601991653,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1463
+    },
+    {
+      "epoch": 1.9848156182212582,
+      "grad_norm": 0.14282701909542084,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1464
+    },
+    {
+      "epoch": 1.98617136659436,
+      "grad_norm": 0.15285570919513702,
+      "learning_rate": 0.0001,
+      "loss": 0.2023,
+      "step": 1465
+    },
+    {
+      "epoch": 1.987527114967462,
+      "grad_norm": 0.11243706941604614,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1466
+    },
+    {
+      "epoch": 1.988882863340564,
+      "grad_norm": 0.12018000334501266,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 1467
+    },
+    {
+      "epoch": 1.990238611713666,
+      "grad_norm": 0.10517760366201401,
+      "learning_rate": 0.0001,
+      "loss": 0.2028,
+      "step": 1468
+    },
+    {
+      "epoch": 1.991594360086768,
+      "grad_norm": 0.09495973587036133,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 1469
+    },
+    {
+      "epoch": 1.99295010845987,
+      "grad_norm": 0.11628373712301254,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 1470
+    },
+    {
+      "epoch": 1.994305856832972,
+      "grad_norm": 0.11972565203905106,
+      "learning_rate": 0.0001,
+      "loss": 0.1933,
+      "step": 1471
+    },
+    {
+      "epoch": 1.9956616052060738,
+      "grad_norm": 0.10712572932243347,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1472
+    },
+    {
+      "epoch": 1.9970173535791758,
+      "grad_norm": 0.09603525698184967,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 1473
+    },
+    {
+      "epoch": 1.9983731019522777,
+      "grad_norm": 0.10339102149009705,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1474
+    },
+    {
+      "epoch": 1.9997288503253796,
+      "grad_norm": 0.10534688830375671,
+      "learning_rate": 0.0001,
+      "loss": 0.1985,
+      "step": 1475
+    },
+    {
+      "epoch": 2.0010845986984815,
+      "grad_norm": 0.11211862415075302,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1476
+    },
+    {
+      "epoch": 2.0024403470715835,
+      "grad_norm": 0.11722870171070099,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1477
+    },
+    {
+      "epoch": 2.0037960954446854,
+      "grad_norm": 0.10969678312540054,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1478
+    },
+    {
+      "epoch": 2.0051518438177873,
+      "grad_norm": 0.12350711971521378,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 1479
+    },
+    {
+      "epoch": 2.0065075921908893,
+      "grad_norm": 0.1109645888209343,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1480
+    },
+    {
+      "epoch": 2.007863340563991,
+      "grad_norm": 0.14247243106365204,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1481
+    },
+    {
+      "epoch": 2.009219088937093,
+      "grad_norm": 0.12506955862045288,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 1482
+    },
+    {
+      "epoch": 2.010574837310195,
+      "grad_norm": 0.12036751955747604,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 1483
+    },
+    {
+      "epoch": 2.011930585683297,
+      "grad_norm": 0.12250246852636337,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 1484
+    },
+    {
+      "epoch": 2.0132863340563993,
+      "grad_norm": 0.13561339676380157,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1485
+    },
+    {
+      "epoch": 2.0146420824295013,
+      "grad_norm": 0.13706760108470917,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 1486
+    },
+    {
+      "epoch": 2.015997830802603,
+      "grad_norm": 0.14557357132434845,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 1487
+    },
+    {
+      "epoch": 2.017353579175705,
+      "grad_norm": 0.29200097918510437,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 1488
+    },
+    {
+      "epoch": 2.018709327548807,
+      "grad_norm": 0.1385810375213623,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1489
+    },
+    {
+      "epoch": 2.020065075921909,
+      "grad_norm": 0.1447155773639679,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 1490
+    },
+    {
+      "epoch": 2.021420824295011,
+      "grad_norm": 0.14343003928661346,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1491
+    },
+    {
+      "epoch": 2.022776572668113,
+      "grad_norm": 0.1451655477285385,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1492
+    },
+    {
+      "epoch": 2.0241323210412148,
+      "grad_norm": 0.11695728451013565,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1493
+    },
+    {
+      "epoch": 2.0254880694143167,
+      "grad_norm": 0.14340044558048248,
+      "learning_rate": 0.0001,
+      "loss": 0.1917,
+      "step": 1494
+    },
+    {
+      "epoch": 2.0268438177874186,
+      "grad_norm": 0.12146680057048798,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1495
+    },
+    {
+      "epoch": 2.0281995661605206,
+      "grad_norm": 0.12590985000133514,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1496
+    },
+    {
+      "epoch": 2.0295553145336225,
+      "grad_norm": 0.14505374431610107,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 1497
+    },
+    {
+      "epoch": 2.0309110629067244,
+      "grad_norm": 0.1317855715751648,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 1498
+    },
+    {
+      "epoch": 2.0322668112798263,
+      "grad_norm": 0.12209803611040115,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1499
+    },
+    {
+      "epoch": 2.0336225596529283,
+      "grad_norm": 0.12174764275550842,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1500
+    },
+    {
+      "epoch": 2.0336225596529283,
+      "eval_loss": 0.1940242350101471,
+      "eval_runtime": 598.977,
+      "eval_samples_per_second": 4.618,
+      "eval_steps_per_second": 1.155,
+      "step": 1500
+    },
+    {
+      "epoch": 2.03497830802603,
+      "grad_norm": 0.15083366632461548,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1501
+    },
+    {
+      "epoch": 2.036334056399132,
+      "grad_norm": 0.11443629115819931,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1502
+    },
+    {
+      "epoch": 2.037689804772234,
+      "grad_norm": 0.13465768098831177,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1503
+    },
+    {
+      "epoch": 2.0390455531453364,
+      "grad_norm": 0.11749093234539032,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 1504
+    },
+    {
+      "epoch": 2.0404013015184383,
+      "grad_norm": 0.16848647594451904,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 1505
+    },
+    {
+      "epoch": 2.0417570498915403,
+      "grad_norm": 0.1512487679719925,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1506
+    },
+    {
+      "epoch": 2.043112798264642,
+      "grad_norm": 0.14480343461036682,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 1507
+    },
+    {
+      "epoch": 2.044468546637744,
+      "grad_norm": 0.14325043559074402,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1508
+    },
+    {
+      "epoch": 2.045824295010846,
+      "grad_norm": 0.15253564715385437,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 1509
+    },
+    {
+      "epoch": 2.047180043383948,
+      "grad_norm": 0.11457974463701248,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1510
+    },
+    {
+      "epoch": 2.04853579175705,
+      "grad_norm": 0.15782511234283447,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 1511
+    },
+    {
+      "epoch": 2.049891540130152,
+      "grad_norm": 0.37062159180641174,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 1512
+    },
+    {
+      "epoch": 2.0512472885032538,
+      "grad_norm": 0.14218860864639282,
+      "learning_rate": 0.0001,
+      "loss": 0.1942,
+      "step": 1513
+    },
+    {
+      "epoch": 2.0526030368763557,
+      "grad_norm": 0.12189324200153351,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1514
+    },
+    {
+      "epoch": 2.0539587852494576,
+      "grad_norm": 0.12311328202486038,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1515
+    },
+    {
+      "epoch": 2.0553145336225596,
+      "grad_norm": 0.12938576936721802,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1516
+    },
+    {
+      "epoch": 2.0566702819956615,
+      "grad_norm": 0.1667119264602661,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 1517
+    },
+    {
+      "epoch": 2.0580260303687634,
+      "grad_norm": 0.11717395484447479,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1518
+    },
+    {
+      "epoch": 2.0593817787418653,
+      "grad_norm": 0.14932990074157715,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1519
+    },
+    {
+      "epoch": 2.0607375271149673,
+      "grad_norm": 0.13962507247924805,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 1520
+    },
+    {
+      "epoch": 2.062093275488069,
+      "grad_norm": 0.12420637905597687,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1521
+    },
+    {
+      "epoch": 2.0634490238611716,
+      "grad_norm": 0.12071668356657028,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 1522
+    },
+    {
+      "epoch": 2.0648047722342735,
+      "grad_norm": 0.16795822978019714,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 1523
+    },
+    {
+      "epoch": 2.0661605206073754,
+      "grad_norm": 0.12975743412971497,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1524
+    },
+    {
+      "epoch": 2.0675162689804774,
+      "grad_norm": 0.13956806063652039,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 1525
+    },
+    {
+      "epoch": 2.0688720173535793,
+      "grad_norm": 0.1577308028936386,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 1526
+    },
+    {
+      "epoch": 2.070227765726681,
+      "grad_norm": 0.13579098880290985,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1527
+    },
+    {
+      "epoch": 2.071583514099783,
+      "grad_norm": 0.1431436389684677,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1528
+    },
+    {
+      "epoch": 2.072939262472885,
+      "grad_norm": 0.11545071750879288,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1529
+    },
+    {
+      "epoch": 2.074295010845987,
+      "grad_norm": 0.1387283354997635,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 1530
+    },
+    {
+      "epoch": 2.075650759219089,
+      "grad_norm": 0.15424039959907532,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1531
+    },
+    {
+      "epoch": 2.077006507592191,
+      "grad_norm": 0.15335212647914886,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1532
+    },
+    {
+      "epoch": 2.078362255965293,
+      "grad_norm": 0.13307146728038788,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 1533
+    },
+    {
+      "epoch": 2.0797180043383947,
+      "grad_norm": 0.3623206913471222,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 1534
+    },
+    {
+      "epoch": 2.0810737527114966,
+      "grad_norm": 0.16389431059360504,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1535
+    },
+    {
+      "epoch": 2.0824295010845986,
+      "grad_norm": 0.13593287765979767,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1536
+    },
+    {
+      "epoch": 2.0837852494577005,
+      "grad_norm": 0.13062967360019684,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 1537
+    },
+    {
+      "epoch": 2.0851409978308024,
+      "grad_norm": 0.13697993755340576,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 1538
+    },
+    {
+      "epoch": 2.0864967462039044,
+      "grad_norm": 0.1638040691614151,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1539
+    },
+    {
+      "epoch": 2.0878524945770067,
+      "grad_norm": 0.1512344628572464,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1540
+    },
+    {
+      "epoch": 2.0892082429501087,
+      "grad_norm": 0.12879016995429993,
+      "learning_rate": 0.0001,
+      "loss": 0.1779,
+      "step": 1541
+    },
+    {
+      "epoch": 2.0905639913232106,
+      "grad_norm": 0.12636736035346985,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 1542
+    },
+    {
+      "epoch": 2.0919197396963125,
+      "grad_norm": 0.1380431056022644,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1543
+    },
+    {
+      "epoch": 2.0932754880694144,
+      "grad_norm": 0.13902105391025543,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 1544
+    },
+    {
+      "epoch": 2.0946312364425164,
+      "grad_norm": 0.13652832806110382,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 1545
+    },
+    {
+      "epoch": 2.0959869848156183,
+      "grad_norm": 0.23625411093235016,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 1546
+    },
+    {
+      "epoch": 2.09734273318872,
+      "grad_norm": 0.15604355931282043,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 1547
+    },
+    {
+      "epoch": 2.098698481561822,
+      "grad_norm": 0.11908724159002304,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 1548
+    },
+    {
+      "epoch": 2.100054229934924,
+      "grad_norm": 0.12767139077186584,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 1549
+    },
+    {
+      "epoch": 2.101409978308026,
+      "grad_norm": 0.1439056545495987,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 1550
+    },
+    {
+      "epoch": 2.102765726681128,
+      "grad_norm": 0.14219361543655396,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1551
+    },
+    {
+      "epoch": 2.10412147505423,
+      "grad_norm": 0.14059534668922424,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1552
+    },
+    {
+      "epoch": 2.105477223427332,
+      "grad_norm": 0.18150627613067627,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 1553
+    },
+    {
+      "epoch": 2.1068329718004337,
+      "grad_norm": 0.15568111836910248,
+      "learning_rate": 0.0001,
+      "loss": 0.1702,
+      "step": 1554
+    },
+    {
+      "epoch": 2.1081887201735356,
+      "grad_norm": 0.13531972467899323,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 1555
+    },
+    {
+      "epoch": 2.1095444685466376,
+      "grad_norm": 0.14124904572963715,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 1556
+    },
+    {
+      "epoch": 2.1109002169197395,
+      "grad_norm": 0.12770290672779083,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 1557
+    },
+    {
+      "epoch": 2.1122559652928414,
+      "grad_norm": 0.1708056479692459,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1558
+    },
+    {
+      "epoch": 2.113611713665944,
+      "grad_norm": 0.15470866858959198,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1559
+    },
+    {
+      "epoch": 2.1149674620390457,
+      "grad_norm": 0.14157503843307495,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1560
+    },
+    {
+      "epoch": 2.1163232104121477,
+      "grad_norm": 0.11347147822380066,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 1561
+    },
+    {
+      "epoch": 2.1176789587852496,
+      "grad_norm": 0.1300053596496582,
+      "learning_rate": 0.0001,
+      "loss": 0.1706,
+      "step": 1562
+    },
+    {
+      "epoch": 2.1190347071583515,
+      "grad_norm": 0.1272808015346527,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 1563
+    },
+    {
+      "epoch": 2.1203904555314534,
+      "grad_norm": 0.15705911815166473,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 1564
+    },
+    {
+      "epoch": 2.1217462039045554,
+      "grad_norm": 0.1795680969953537,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 1565
+    },
+    {
+      "epoch": 2.1231019522776573,
+      "grad_norm": 0.11919550597667694,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 1566
+    },
+    {
+      "epoch": 2.1244577006507592,
+      "grad_norm": 0.1326528787612915,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 1567
+    },
+    {
+      "epoch": 2.125813449023861,
+      "grad_norm": 0.14050917327404022,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1568
+    },
+    {
+      "epoch": 2.127169197396963,
+      "grad_norm": 0.11937253177165985,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 1569
+    },
+    {
+      "epoch": 2.128524945770065,
+      "grad_norm": 0.1364632397890091,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 1570
+    },
+    {
+      "epoch": 2.129880694143167,
+      "grad_norm": 0.12543657422065735,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1571
+    },
+    {
+      "epoch": 2.131236442516269,
+      "grad_norm": 0.12024430185556412,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 1572
+    },
+    {
+      "epoch": 2.132592190889371,
+      "grad_norm": 0.12795285880565643,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1573
+    },
+    {
+      "epoch": 2.1339479392624727,
+      "grad_norm": 0.14104045927524567,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 1574
+    },
+    {
+      "epoch": 2.1353036876355747,
+      "grad_norm": 0.150103360414505,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 1575
+    },
+    {
+      "epoch": 2.1366594360086766,
+      "grad_norm": 0.15485058724880219,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 1576
+    },
+    {
+      "epoch": 2.1380151843817785,
+      "grad_norm": 0.15760836005210876,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 1577
+    },
+    {
+      "epoch": 2.139370932754881,
+      "grad_norm": 0.12862347066402435,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 1578
+    },
+    {
+      "epoch": 2.140726681127983,
+      "grad_norm": 0.14734147489070892,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1579
+    },
+    {
+      "epoch": 2.1420824295010847,
+      "grad_norm": 0.13999058306217194,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 1580
+    },
+    {
+      "epoch": 2.1434381778741867,
+      "grad_norm": 0.1420711874961853,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 1581
+    },
+    {
+      "epoch": 2.1447939262472886,
+      "grad_norm": 0.13666944205760956,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 1582
+    },
+    {
+      "epoch": 2.1461496746203905,
+      "grad_norm": 0.12495048344135284,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 1583
+    },
+    {
+      "epoch": 2.1475054229934925,
+      "grad_norm": 0.1323455274105072,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1584
+    },
+    {
+      "epoch": 2.1488611713665944,
+      "grad_norm": 0.1265854835510254,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1585
+    },
+    {
+      "epoch": 2.1502169197396963,
+      "grad_norm": 0.1382724642753601,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1586
+    },
+    {
+      "epoch": 2.1515726681127982,
+      "grad_norm": 0.20584627985954285,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 1587
+    },
+    {
+      "epoch": 2.1529284164859,
+      "grad_norm": 0.14046695828437805,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 1588
+    },
+    {
+      "epoch": 2.154284164859002,
+      "grad_norm": 0.13364088535308838,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 1589
+    },
+    {
+      "epoch": 2.155639913232104,
+      "grad_norm": 0.1610957384109497,
+      "learning_rate": 0.0001,
+      "loss": 0.1853,
+      "step": 1590
+    },
+    {
+      "epoch": 2.156995661605206,
+      "grad_norm": 0.12944139540195465,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 1591
+    },
+    {
+      "epoch": 2.158351409978308,
+      "grad_norm": 0.15146706998348236,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 1592
+    },
+    {
+      "epoch": 2.15970715835141,
+      "grad_norm": 0.13295045495033264,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 1593
+    },
+    {
+      "epoch": 2.1610629067245117,
+      "grad_norm": 0.13657963275909424,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 1594
+    },
+    {
+      "epoch": 2.162418655097614,
+      "grad_norm": 0.13421840965747833,
+      "learning_rate": 0.0001,
+      "loss": 0.1848,
+      "step": 1595
+    },
+    {
+      "epoch": 2.163774403470716,
+      "grad_norm": 0.11724752187728882,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1596
+    },
+    {
+      "epoch": 2.165130151843818,
+      "grad_norm": 0.12780649960041046,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 1597
+    },
+    {
+      "epoch": 2.16648590021692,
+      "grad_norm": 0.1341710239648819,
+      "learning_rate": 0.0001,
+      "loss": 0.1735,
+      "step": 1598
+    },
+    {
+      "epoch": 2.167841648590022,
+      "grad_norm": 0.1329096406698227,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1599
+    },
+    {
+      "epoch": 2.1691973969631237,
+      "grad_norm": 0.13396957516670227,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1600
+    },
+    {
+      "epoch": 2.1705531453362257,
+      "grad_norm": 0.1353560984134674,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1601
+    },
+    {
+      "epoch": 2.1719088937093276,
+      "grad_norm": 0.12873829901218414,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1602
+    },
+    {
+      "epoch": 2.1732646420824295,
+      "grad_norm": 0.1359226405620575,
+      "learning_rate": 0.0001,
+      "loss": 0.1863,
+      "step": 1603
+    },
+    {
+      "epoch": 2.1746203904555315,
+      "grad_norm": 0.14013107120990753,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 1604
+    },
+    {
+      "epoch": 2.1759761388286334,
+      "grad_norm": 0.1416664719581604,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1605
+    },
+    {
+      "epoch": 2.1773318872017353,
+      "grad_norm": 0.12688012421131134,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 1606
+    },
+    {
+      "epoch": 2.1786876355748372,
+      "grad_norm": 0.15436887741088867,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 1607
+    },
+    {
+      "epoch": 2.180043383947939,
+      "grad_norm": 0.15290120244026184,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1608
+    },
+    {
+      "epoch": 2.181399132321041,
+      "grad_norm": 0.1752799153327942,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 1609
+    },
+    {
+      "epoch": 2.182754880694143,
+      "grad_norm": 0.15672698616981506,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 1610
+    },
+    {
+      "epoch": 2.184110629067245,
+      "grad_norm": 0.1511913239955902,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 1611
+    },
+    {
+      "epoch": 2.185466377440347,
+      "grad_norm": 0.13836145401000977,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1612
+    },
+    {
+      "epoch": 2.186822125813449,
+      "grad_norm": 0.15657475590705872,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 1613
+    },
+    {
+      "epoch": 2.188177874186551,
+      "grad_norm": 0.16395945847034454,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 1614
+    },
+    {
+      "epoch": 2.189533622559653,
+      "grad_norm": 0.16202005743980408,
+      "learning_rate": 0.0001,
+      "loss": 0.1779,
+      "step": 1615
+    },
+    {
+      "epoch": 2.190889370932755,
+      "grad_norm": 0.13093328475952148,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 1616
+    },
+    {
+      "epoch": 2.192245119305857,
+      "grad_norm": 0.1303802877664566,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1617
+    },
+    {
+      "epoch": 2.193600867678959,
+      "grad_norm": 0.1909562200307846,
+      "learning_rate": 0.0001,
+      "loss": 0.1961,
+      "step": 1618
+    },
+    {
+      "epoch": 2.194956616052061,
+      "grad_norm": 0.12153103202581406,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 1619
+    },
+    {
+      "epoch": 2.1963123644251628,
+      "grad_norm": 0.12234237045049667,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 1620
+    },
+    {
+      "epoch": 2.1976681127982647,
+      "grad_norm": 0.12391461431980133,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 1621
+    },
+    {
+      "epoch": 2.1990238611713666,
+      "grad_norm": 0.12571696937084198,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 1622
+    },
+    {
+      "epoch": 2.2003796095444685,
+      "grad_norm": 0.1604662984609604,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1623
+    },
+    {
+      "epoch": 2.2017353579175705,
+      "grad_norm": 0.13143956661224365,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1624
+    },
+    {
+      "epoch": 2.2030911062906724,
+      "grad_norm": 0.14480189979076385,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1625
+    },
+    {
+      "epoch": 2.2044468546637743,
+      "grad_norm": 0.14590312540531158,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1626
+    },
+    {
+      "epoch": 2.2058026030368763,
+      "grad_norm": 0.15697380900382996,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 1627
+    },
+    {
+      "epoch": 2.207158351409978,
+      "grad_norm": 0.15544751286506653,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 1628
+    },
+    {
+      "epoch": 2.20851409978308,
+      "grad_norm": 0.13963334262371063,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 1629
+    },
+    {
+      "epoch": 2.209869848156182,
+      "grad_norm": 0.14501529932022095,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 1630
+    },
+    {
+      "epoch": 2.211225596529284,
+      "grad_norm": 0.2369701862335205,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1631
+    },
+    {
+      "epoch": 2.212581344902386,
+      "grad_norm": 0.13367590308189392,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 1632
+    },
+    {
+      "epoch": 2.2139370932754883,
+      "grad_norm": 0.13435053825378418,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 1633
+    },
+    {
+      "epoch": 2.21529284164859,
+      "grad_norm": 0.1346767395734787,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1634
+    },
+    {
+      "epoch": 2.216648590021692,
+      "grad_norm": 0.1625145673751831,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 1635
+    },
+    {
+      "epoch": 2.218004338394794,
+      "grad_norm": 0.13845720887184143,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1636
+    },
+    {
+      "epoch": 2.219360086767896,
+      "grad_norm": 0.17217110097408295,
+      "learning_rate": 0.0001,
+      "loss": 0.1951,
+      "step": 1637
+    },
+    {
+      "epoch": 2.220715835140998,
+      "grad_norm": 0.14031730592250824,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1638
+    },
+    {
+      "epoch": 2.2220715835141,
+      "grad_norm": 0.14364181458950043,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1639
+    },
+    {
+      "epoch": 2.2234273318872018,
+      "grad_norm": 0.12490963190793991,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1640
+    },
+    {
+      "epoch": 2.2247830802603037,
+      "grad_norm": 0.14568261802196503,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 1641
+    },
+    {
+      "epoch": 2.2261388286334056,
+      "grad_norm": 0.14572761952877045,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 1642
+    },
+    {
+      "epoch": 2.2274945770065075,
+      "grad_norm": 0.1437234878540039,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 1643
+    },
+    {
+      "epoch": 2.2288503253796095,
+      "grad_norm": 0.12780772149562836,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 1644
+    },
+    {
+      "epoch": 2.2302060737527114,
+      "grad_norm": 0.1284794807434082,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 1645
+    },
+    {
+      "epoch": 2.2315618221258133,
+      "grad_norm": 0.14983277022838593,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 1646
+    },
+    {
+      "epoch": 2.2329175704989153,
+      "grad_norm": 0.13619579374790192,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 1647
+    },
+    {
+      "epoch": 2.234273318872017,
+      "grad_norm": 0.1373264640569687,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1648
+    },
+    {
+      "epoch": 2.235629067245119,
+      "grad_norm": 0.12539522349834442,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1649
+    },
+    {
+      "epoch": 2.2369848156182215,
+      "grad_norm": 0.13389815390110016,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1650
+    },
+    {
+      "epoch": 2.2383405639913234,
+      "grad_norm": 0.1448618769645691,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 1651
+    },
+    {
+      "epoch": 2.2396963123644253,
+      "grad_norm": 0.12419396638870239,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 1652
+    },
+    {
+      "epoch": 2.2410520607375273,
+      "grad_norm": 0.17443470656871796,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 1653
+    },
+    {
+      "epoch": 2.242407809110629,
+      "grad_norm": 0.12900389730930328,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1654
+    },
+    {
+      "epoch": 2.243763557483731,
+      "grad_norm": 0.14614033699035645,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1655
+    },
+    {
+      "epoch": 2.245119305856833,
+      "grad_norm": 0.11778552830219269,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 1656
+    },
+    {
+      "epoch": 2.246475054229935,
+      "grad_norm": 0.13851110637187958,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 1657
+    },
+    {
+      "epoch": 2.247830802603037,
+      "grad_norm": 0.11600729823112488,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1658
+    },
+    {
+      "epoch": 2.249186550976139,
+      "grad_norm": 0.11106248944997787,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 1659
+    },
+    {
+      "epoch": 2.2505422993492408,
+      "grad_norm": 0.12387838959693909,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1660
+    },
+    {
+      "epoch": 2.2518980477223427,
+      "grad_norm": 0.1518625169992447,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 1661
+    },
+    {
+      "epoch": 2.2532537960954446,
+      "grad_norm": 0.12530219554901123,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 1662
+    },
+    {
+      "epoch": 2.2546095444685466,
+      "grad_norm": 0.12507042288780212,
+      "learning_rate": 0.0001,
+      "loss": 0.183,
+      "step": 1663
+    },
+    {
+      "epoch": 2.2559652928416485,
+      "grad_norm": 0.1251320093870163,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 1664
+    },
+    {
+      "epoch": 2.2573210412147504,
+      "grad_norm": 0.14066371321678162,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1665
+    },
+    {
+      "epoch": 2.2586767895878523,
+      "grad_norm": 0.1421237736940384,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 1666
+    },
+    {
+      "epoch": 2.2600325379609543,
+      "grad_norm": 0.12262992560863495,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 1667
+    },
+    {
+      "epoch": 2.261388286334056,
+      "grad_norm": 0.15946927666664124,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 1668
+    },
+    {
+      "epoch": 2.2627440347071586,
+      "grad_norm": 0.14757443964481354,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 1669
+    },
+    {
+      "epoch": 2.2640997830802605,
+      "grad_norm": 0.15576466917991638,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 1670
+    },
+    {
+      "epoch": 2.2654555314533624,
+      "grad_norm": 0.1359878033399582,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 1671
+    },
+    {
+      "epoch": 2.2668112798264644,
+      "grad_norm": 0.13674959540367126,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 1672
+    },
+    {
+      "epoch": 2.2681670281995663,
+      "grad_norm": 0.13368159532546997,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 1673
+    },
+    {
+      "epoch": 2.269522776572668,
+      "grad_norm": 0.14111898839473724,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 1674
+    },
+    {
+      "epoch": 2.27087852494577,
+      "grad_norm": 0.12549975514411926,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 1675
+    },
+    {
+      "epoch": 2.272234273318872,
+      "grad_norm": 0.11545778810977936,
+      "learning_rate": 0.0001,
+      "loss": 0.1876,
+      "step": 1676
+    },
+    {
+      "epoch": 2.273590021691974,
+      "grad_norm": 0.1180974617600441,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 1677
+    },
+    {
+      "epoch": 2.274945770065076,
+      "grad_norm": 0.1125197783112526,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 1678
+    },
+    {
+      "epoch": 2.276301518438178,
+      "grad_norm": 0.14856475591659546,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1679
+    },
+    {
+      "epoch": 2.27765726681128,
+      "grad_norm": 0.12645962834358215,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1680
+    },
+    {
+      "epoch": 2.2790130151843817,
+      "grad_norm": 0.13138733804225922,
+      "learning_rate": 0.0001,
+      "loss": 0.1712,
+      "step": 1681
+    },
+    {
+      "epoch": 2.2803687635574836,
+      "grad_norm": 0.12731800973415375,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 1682
+    },
+    {
+      "epoch": 2.2817245119305856,
+      "grad_norm": 0.13199755549430847,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 1683
+    },
+    {
+      "epoch": 2.2830802603036875,
+      "grad_norm": 0.16242188215255737,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 1684
+    },
+    {
+      "epoch": 2.2844360086767894,
+      "grad_norm": 0.1725577712059021,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 1685
+    },
+    {
+      "epoch": 2.2857917570498913,
+      "grad_norm": 0.13155874609947205,
+      "learning_rate": 0.0001,
+      "loss": 0.1765,
+      "step": 1686
+    },
+    {
+      "epoch": 2.2871475054229933,
+      "grad_norm": 0.24537064135074615,
+      "learning_rate": 0.0001,
+      "loss": 0.1759,
+      "step": 1687
+    },
+    {
+      "epoch": 2.2885032537960956,
+      "grad_norm": 0.15291744470596313,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "step": 1688
+    },
+    {
+      "epoch": 2.2898590021691976,
+      "grad_norm": 0.13214854896068573,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 1689
+    },
+    {
+      "epoch": 2.2912147505422995,
+      "grad_norm": 0.1289874166250229,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1690
+    },
+    {
+      "epoch": 2.2925704989154014,
+      "grad_norm": 0.12239929288625717,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1691
+    },
+    {
+      "epoch": 2.2939262472885034,
+      "grad_norm": 0.13659486174583435,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1692
+    },
+    {
+      "epoch": 2.2952819956616053,
+      "grad_norm": 0.19577306509017944,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1693
+    },
+    {
+      "epoch": 2.296637744034707,
+      "grad_norm": 0.12895137071609497,
+      "learning_rate": 0.0001,
+      "loss": 0.1605,
+      "step": 1694
+    },
+    {
+      "epoch": 2.297993492407809,
+      "grad_norm": 0.13768333196640015,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 1695
+    },
+    {
+      "epoch": 2.299349240780911,
+      "grad_norm": 0.13093388080596924,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 1696
+    },
+    {
+      "epoch": 2.300704989154013,
+      "grad_norm": 0.11495552957057953,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1697
+    },
+    {
+      "epoch": 2.302060737527115,
+      "grad_norm": 0.12672379612922668,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 1698
+    },
+    {
+      "epoch": 2.303416485900217,
+      "grad_norm": 0.11887643486261368,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 1699
+    },
+    {
+      "epoch": 2.304772234273319,
+      "grad_norm": 0.18476954102516174,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1700
+    },
+    {
+      "epoch": 2.3061279826464207,
+      "grad_norm": 0.1330435574054718,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 1701
+    },
+    {
+      "epoch": 2.3074837310195226,
+      "grad_norm": 0.13997763395309448,
+      "learning_rate": 0.0001,
+      "loss": 0.1605,
+      "step": 1702
+    },
+    {
+      "epoch": 2.3088394793926246,
+      "grad_norm": 0.14067688584327698,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 1703
+    },
+    {
+      "epoch": 2.3101952277657265,
+      "grad_norm": 0.12718264758586884,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 1704
+    },
+    {
+      "epoch": 2.311550976138829,
+      "grad_norm": 0.1252772957086563,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 1705
+    },
+    {
+      "epoch": 2.3129067245119304,
+      "grad_norm": 0.15276820957660675,
+      "learning_rate": 0.0001,
+      "loss": 0.1809,
+      "step": 1706
+    },
+    {
+      "epoch": 2.3142624728850327,
+      "grad_norm": 0.14265884459018707,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1707
+    },
+    {
+      "epoch": 2.3156182212581347,
+      "grad_norm": 0.16426727175712585,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1708
+    },
+    {
+      "epoch": 2.3169739696312366,
+      "grad_norm": 0.14440706372261047,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 1709
+    },
+    {
+      "epoch": 2.3183297180043385,
+      "grad_norm": 0.13933925330638885,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 1710
+    },
+    {
+      "epoch": 2.3196854663774404,
+      "grad_norm": 0.1536855548620224,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 1711
+    },
+    {
+      "epoch": 2.3210412147505424,
+      "grad_norm": 0.144095316529274,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1712
+    },
+    {
+      "epoch": 2.3223969631236443,
+      "grad_norm": 0.1279783695936203,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1713
+    },
+    {
+      "epoch": 2.3237527114967462,
+      "grad_norm": 0.146341934800148,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 1714
+    },
+    {
+      "epoch": 2.325108459869848,
+      "grad_norm": 0.1271177977323532,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 1715
+    },
+    {
+      "epoch": 2.32646420824295,
+      "grad_norm": 0.1421731859445572,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1716
+    },
+    {
+      "epoch": 2.327819956616052,
+      "grad_norm": 0.17044255137443542,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 1717
+    },
+    {
+      "epoch": 2.329175704989154,
+      "grad_norm": 0.1398533582687378,
+      "learning_rate": 0.0001,
+      "loss": 0.189,
+      "step": 1718
+    },
+    {
+      "epoch": 2.330531453362256,
+      "grad_norm": 0.1496542990207672,
+      "learning_rate": 0.0001,
+      "loss": 0.1835,
+      "step": 1719
+    },
+    {
+      "epoch": 2.331887201735358,
+      "grad_norm": 0.12644293904304504,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 1720
+    },
+    {
+      "epoch": 2.3332429501084597,
+      "grad_norm": 0.1502489447593689,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1721
+    },
+    {
+      "epoch": 2.3345986984815617,
+      "grad_norm": 0.13845519721508026,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 1722
+    },
+    {
+      "epoch": 2.3359544468546636,
+      "grad_norm": 0.12244993448257446,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 1723
+    },
+    {
+      "epoch": 2.337310195227766,
+      "grad_norm": 0.14362798631191254,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 1724
+    },
+    {
+      "epoch": 2.3386659436008674,
+      "grad_norm": 0.13733485341072083,
+      "learning_rate": 0.0001,
+      "loss": 0.1874,
+      "step": 1725
+    },
+    {
+      "epoch": 2.34002169197397,
+      "grad_norm": 0.12415989488363266,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 1726
+    },
+    {
+      "epoch": 2.3413774403470717,
+      "grad_norm": 0.12717168033123016,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1727
+    },
+    {
+      "epoch": 2.3427331887201737,
+      "grad_norm": 0.13738910853862762,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 1728
+    },
+    {
+      "epoch": 2.3440889370932756,
+      "grad_norm": 0.11867024004459381,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 1729
+    },
+    {
+      "epoch": 2.3454446854663775,
+      "grad_norm": 0.15322217345237732,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 1730
+    },
+    {
+      "epoch": 2.3468004338394794,
+      "grad_norm": 0.15615007281303406,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 1731
+    },
+    {
+      "epoch": 2.3481561822125814,
+      "grad_norm": 0.14995476603507996,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 1732
+    },
+    {
+      "epoch": 2.3495119305856833,
+      "grad_norm": 0.1390848308801651,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1733
+    },
+    {
+      "epoch": 2.3508676789587852,
+      "grad_norm": 0.16655781865119934,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 1734
+    },
+    {
+      "epoch": 2.352223427331887,
+      "grad_norm": 0.15455470979213715,
+      "learning_rate": 0.0001,
+      "loss": 0.1781,
+      "step": 1735
+    },
+    {
+      "epoch": 2.353579175704989,
+      "grad_norm": 0.14578616619110107,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 1736
+    },
+    {
+      "epoch": 2.354934924078091,
+      "grad_norm": 0.1324901580810547,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 1737
+    },
+    {
+      "epoch": 2.356290672451193,
+      "grad_norm": 0.12291241437196732,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1738
+    },
+    {
+      "epoch": 2.357646420824295,
+      "grad_norm": 0.13602417707443237,
+      "learning_rate": 0.0001,
+      "loss": 0.1735,
+      "step": 1739
+    },
+    {
+      "epoch": 2.359002169197397,
+      "grad_norm": 0.12108414620161057,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 1740
+    },
+    {
+      "epoch": 2.3603579175704987,
+      "grad_norm": 0.2806040048599243,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1741
+    },
+    {
+      "epoch": 2.3617136659436007,
+      "grad_norm": 0.14715340733528137,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1742
+    },
+    {
+      "epoch": 2.363069414316703,
+      "grad_norm": 0.12058188766241074,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 1743
+    },
+    {
+      "epoch": 2.364425162689805,
+      "grad_norm": 0.13248476386070251,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1744
+    },
+    {
+      "epoch": 2.365780911062907,
+      "grad_norm": 0.17047809064388275,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1745
+    },
+    {
+      "epoch": 2.367136659436009,
+      "grad_norm": 0.12245938926935196,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 1746
+    },
+    {
+      "epoch": 2.3684924078091107,
+      "grad_norm": 0.1338166743516922,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 1747
+    },
+    {
+      "epoch": 2.3698481561822127,
+      "grad_norm": 0.13590995967388153,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 1748
+    },
+    {
+      "epoch": 2.3712039045553146,
+      "grad_norm": 0.17029538750648499,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 1749
+    },
+    {
+      "epoch": 2.3725596529284165,
+      "grad_norm": 0.15089908242225647,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 1750
+    },
+    {
+      "epoch": 2.3739154013015185,
+      "grad_norm": 0.14287139475345612,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 1751
+    },
+    {
+      "epoch": 2.3752711496746204,
+      "grad_norm": 0.14319908618927002,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1752
+    },
+    {
+      "epoch": 2.3766268980477223,
+      "grad_norm": 0.1444883942604065,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "step": 1753
+    },
+    {
+      "epoch": 2.3779826464208242,
+      "grad_norm": 0.13091912865638733,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1754
+    },
+    {
+      "epoch": 2.379338394793926,
+      "grad_norm": 0.15321703255176544,
+      "learning_rate": 0.0001,
+      "loss": 0.1921,
+      "step": 1755
+    },
+    {
+      "epoch": 2.380694143167028,
+      "grad_norm": 0.1223989874124527,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 1756
+    },
+    {
+      "epoch": 2.38204989154013,
+      "grad_norm": 0.12739157676696777,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 1757
+    },
+    {
+      "epoch": 2.383405639913232,
+      "grad_norm": 0.11810073256492615,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 1758
+    },
+    {
+      "epoch": 2.384761388286334,
+      "grad_norm": 0.1277090609073639,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1759
+    },
+    {
+      "epoch": 2.3861171366594363,
+      "grad_norm": 0.12391582131385803,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 1760
+    },
+    {
+      "epoch": 2.3874728850325377,
+      "grad_norm": 0.1423174887895584,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 1761
+    },
+    {
+      "epoch": 2.38882863340564,
+      "grad_norm": 0.11707404255867004,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 1762
+    },
+    {
+      "epoch": 2.390184381778742,
+      "grad_norm": 0.1519981026649475,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1763
+    },
+    {
+      "epoch": 2.391540130151844,
+      "grad_norm": 0.14756429195404053,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1764
+    },
+    {
+      "epoch": 2.392895878524946,
+      "grad_norm": 0.14617183804512024,
+      "learning_rate": 0.0001,
+      "loss": 0.1934,
+      "step": 1765
+    },
+    {
+      "epoch": 2.394251626898048,
+      "grad_norm": 0.11868690699338913,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 1766
+    },
+    {
+      "epoch": 2.3956073752711498,
+      "grad_norm": 0.12079311907291412,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 1767
+    },
+    {
+      "epoch": 2.3969631236442517,
+      "grad_norm": 0.18320432305335999,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 1768
+    },
+    {
+      "epoch": 2.3983188720173536,
+      "grad_norm": 0.13834840059280396,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1769
+    },
+    {
+      "epoch": 2.3996746203904555,
+      "grad_norm": 0.1619447022676468,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 1770
+    },
+    {
+      "epoch": 2.4010303687635575,
+      "grad_norm": 0.1345987617969513,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1771
+    },
+    {
+      "epoch": 2.4023861171366594,
+      "grad_norm": 0.15047743916511536,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 1772
+    },
+    {
+      "epoch": 2.4037418655097613,
+      "grad_norm": 0.1442045122385025,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 1773
+    },
+    {
+      "epoch": 2.4050976138828633,
+      "grad_norm": 0.18145479261875153,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1774
+    },
+    {
+      "epoch": 2.406453362255965,
+      "grad_norm": 0.12682834267616272,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1775
+    },
+    {
+      "epoch": 2.407809110629067,
+      "grad_norm": 0.15174347162246704,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1776
+    },
+    {
+      "epoch": 2.409164859002169,
+      "grad_norm": 0.14735017716884613,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 1777
+    },
+    {
+      "epoch": 2.410520607375271,
+      "grad_norm": 0.1435907632112503,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1778
+    },
+    {
+      "epoch": 2.4118763557483733,
+      "grad_norm": 0.13979308307170868,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 1779
+    },
+    {
+      "epoch": 2.413232104121475,
+      "grad_norm": 0.17801782488822937,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 1780
+    },
+    {
+      "epoch": 2.414587852494577,
+      "grad_norm": 0.12973207235336304,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 1781
+    },
+    {
+      "epoch": 2.415943600867679,
+      "grad_norm": 0.1462128460407257,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1782
+    },
+    {
+      "epoch": 2.417299349240781,
+      "grad_norm": 0.12530887126922607,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 1783
+    },
+    {
+      "epoch": 2.418655097613883,
+      "grad_norm": 0.13171686232089996,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1784
+    },
+    {
+      "epoch": 2.420010845986985,
+      "grad_norm": 0.13668584823608398,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1785
+    },
+    {
+      "epoch": 2.421366594360087,
+      "grad_norm": 0.14053654670715332,
+      "learning_rate": 0.0001,
+      "loss": 0.1914,
+      "step": 1786
+    },
+    {
+      "epoch": 2.4227223427331888,
+      "grad_norm": 0.1259441375732422,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 1787
+    },
+    {
+      "epoch": 2.4240780911062907,
+      "grad_norm": 0.13038687407970428,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 1788
+    },
+    {
+      "epoch": 2.4254338394793926,
+      "grad_norm": 0.14212144911289215,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1789
+    },
+    {
+      "epoch": 2.4267895878524945,
+      "grad_norm": 0.13585209846496582,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 1790
+    },
+    {
+      "epoch": 2.4281453362255965,
+      "grad_norm": 0.13463978469371796,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 1791
+    },
+    {
+      "epoch": 2.4295010845986984,
+      "grad_norm": 0.12228163331747055,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 1792
+    },
+    {
+      "epoch": 2.4308568329718003,
+      "grad_norm": 0.15495437383651733,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 1793
+    },
+    {
+      "epoch": 2.4322125813449023,
+      "grad_norm": 0.1255296915769577,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 1794
+    },
+    {
+      "epoch": 2.433568329718004,
+      "grad_norm": 0.14157532155513763,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 1795
+    },
+    {
+      "epoch": 2.434924078091106,
+      "grad_norm": 0.13550399243831635,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 1796
+    },
+    {
+      "epoch": 2.436279826464208,
+      "grad_norm": 0.24305929243564606,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1797
+    },
+    {
+      "epoch": 2.4376355748373104,
+      "grad_norm": 0.13397979736328125,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 1798
+    },
+    {
+      "epoch": 2.438991323210412,
+      "grad_norm": 0.12909966707229614,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1799
+    },
+    {
+      "epoch": 2.4403470715835143,
+      "grad_norm": 0.1501222848892212,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1800
+    },
+    {
+      "epoch": 2.441702819956616,
+      "grad_norm": 0.18210765719413757,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 1801
+    },
+    {
+      "epoch": 2.443058568329718,
+      "grad_norm": 0.1371433287858963,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1802
+    },
+    {
+      "epoch": 2.44441431670282,
+      "grad_norm": 0.13496483862400055,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1803
+    },
+    {
+      "epoch": 2.445770065075922,
+      "grad_norm": 0.13801205158233643,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 1804
+    },
+    {
+      "epoch": 2.447125813449024,
+      "grad_norm": 0.15438641607761383,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 1805
+    },
+    {
+      "epoch": 2.448481561822126,
+      "grad_norm": 0.14702734351158142,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 1806
+    },
+    {
+      "epoch": 2.4498373101952278,
+      "grad_norm": 0.13744059205055237,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 1807
+    },
+    {
+      "epoch": 2.4511930585683297,
+      "grad_norm": 0.1277231127023697,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 1808
+    },
+    {
+      "epoch": 2.4525488069414316,
+      "grad_norm": 0.13715390861034393,
+      "learning_rate": 0.0001,
+      "loss": 0.1964,
+      "step": 1809
+    },
+    {
+      "epoch": 2.4539045553145336,
+      "grad_norm": 0.13407765328884125,
+      "learning_rate": 0.0001,
+      "loss": 0.1735,
+      "step": 1810
+    },
+    {
+      "epoch": 2.4552603036876355,
+      "grad_norm": 0.14057017862796783,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 1811
+    },
+    {
+      "epoch": 2.4566160520607374,
+      "grad_norm": 0.18376190960407257,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1812
+    },
+    {
+      "epoch": 2.4579718004338393,
+      "grad_norm": 0.13984379172325134,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 1813
+    },
+    {
+      "epoch": 2.4593275488069413,
+      "grad_norm": 0.13656407594680786,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1814
+    },
+    {
+      "epoch": 2.460683297180043,
+      "grad_norm": 0.12705717980861664,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 1815
+    },
+    {
+      "epoch": 2.462039045553145,
+      "grad_norm": 0.143668070435524,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1816
+    },
+    {
+      "epoch": 2.4633947939262475,
+      "grad_norm": 0.15458276867866516,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 1817
+    },
+    {
+      "epoch": 2.4647505422993494,
+      "grad_norm": 0.15107545256614685,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1818
+    },
+    {
+      "epoch": 2.4661062906724514,
+      "grad_norm": 0.12110920250415802,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 1819
+    },
+    {
+      "epoch": 2.4674620390455533,
+      "grad_norm": 0.13763803243637085,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 1820
+    },
+    {
+      "epoch": 2.468817787418655,
+      "grad_norm": 0.13304542005062103,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 1821
+    },
+    {
+      "epoch": 2.470173535791757,
+      "grad_norm": 0.12976214289665222,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 1822
+    },
+    {
+      "epoch": 2.471529284164859,
+      "grad_norm": 0.13440373539924622,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 1823
+    },
+    {
+      "epoch": 2.472885032537961,
+      "grad_norm": 0.13109742105007172,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 1824
+    },
+    {
+      "epoch": 2.474240780911063,
+      "grad_norm": 0.17504096031188965,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 1825
+    },
+    {
+      "epoch": 2.475596529284165,
+      "grad_norm": 0.1382952779531479,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 1826
+    },
+    {
+      "epoch": 2.4769522776572668,
+      "grad_norm": 0.1462082862854004,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 1827
+    },
+    {
+      "epoch": 2.4783080260303687,
+      "grad_norm": 0.17059199512004852,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 1828
+    },
+    {
+      "epoch": 2.4796637744034706,
+      "grad_norm": 0.1578904539346695,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1829
+    },
+    {
+      "epoch": 2.4810195227765726,
+      "grad_norm": 0.14130403101444244,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 1830
+    },
+    {
+      "epoch": 2.4823752711496745,
+      "grad_norm": 0.12075085937976837,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 1831
+    },
+    {
+      "epoch": 2.4837310195227764,
+      "grad_norm": 0.137804314494133,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 1832
+    },
+    {
+      "epoch": 2.4850867678958783,
+      "grad_norm": 0.1434846967458725,
+      "learning_rate": 0.0001,
+      "loss": 0.1828,
+      "step": 1833
+    },
+    {
+      "epoch": 2.4864425162689807,
+      "grad_norm": 0.1402842253446579,
+      "learning_rate": 0.0001,
+      "loss": 0.183,
+      "step": 1834
+    },
+    {
+      "epoch": 2.487798264642082,
+      "grad_norm": 0.11893846094608307,
+      "learning_rate": 0.0001,
+      "loss": 0.1908,
+      "step": 1835
+    },
+    {
+      "epoch": 2.4891540130151846,
+      "grad_norm": 0.14502321183681488,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 1836
+    },
+    {
+      "epoch": 2.4905097613882865,
+      "grad_norm": 0.13424380123615265,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 1837
+    },
+    {
+      "epoch": 2.4918655097613884,
+      "grad_norm": 0.1576715111732483,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1838
+    },
+    {
+      "epoch": 2.4932212581344904,
+      "grad_norm": 0.12319088727235794,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 1839
+    },
+    {
+      "epoch": 2.4945770065075923,
+      "grad_norm": 0.1391392946243286,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1840
+    },
+    {
+      "epoch": 2.495932754880694,
+      "grad_norm": 0.13067327439785004,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 1841
+    },
+    {
+      "epoch": 2.497288503253796,
+      "grad_norm": 0.14210474491119385,
+      "learning_rate": 0.0001,
+      "loss": 0.1919,
+      "step": 1842
+    },
+    {
+      "epoch": 2.498644251626898,
+      "grad_norm": 0.13928142189979553,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 1843
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.13748638331890106,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1844
+    },
+    {
+      "epoch": 2.501355748373102,
+      "grad_norm": 0.13826265931129456,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 1845
+    },
+    {
+      "epoch": 2.502711496746204,
+      "grad_norm": 0.1676761656999588,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 1846
+    },
+    {
+      "epoch": 2.504067245119306,
+      "grad_norm": 0.12582869827747345,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 1847
+    },
+    {
+      "epoch": 2.5054229934924077,
+      "grad_norm": 0.12027784436941147,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 1848
+    },
+    {
+      "epoch": 2.5067787418655096,
+      "grad_norm": 0.14522966742515564,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 1849
+    },
+    {
+      "epoch": 2.5081344902386116,
+      "grad_norm": 0.12168299406766891,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 1850
+    },
+    {
+      "epoch": 2.509490238611714,
+      "grad_norm": 0.13633276522159576,
+      "learning_rate": 0.0001,
+      "loss": 0.1882,
+      "step": 1851
+    },
+    {
+      "epoch": 2.5108459869848154,
+      "grad_norm": 0.14344076812267303,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 1852
+    },
+    {
+      "epoch": 2.512201735357918,
+      "grad_norm": 0.12768501043319702,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 1853
+    },
+    {
+      "epoch": 2.5135574837310193,
+      "grad_norm": 0.13997620344161987,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 1854
+    },
+    {
+      "epoch": 2.5149132321041217,
+      "grad_norm": 0.12250923365354538,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 1855
+    },
+    {
+      "epoch": 2.5162689804772236,
+      "grad_norm": 0.14338773488998413,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 1856
+    },
+    {
+      "epoch": 2.5176247288503255,
+      "grad_norm": 0.1400040090084076,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 1857
+    },
+    {
+      "epoch": 2.5189804772234274,
+      "grad_norm": 0.14169816672801971,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 1858
+    },
+    {
+      "epoch": 2.5203362255965294,
+      "grad_norm": 0.12995946407318115,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 1859
+    },
+    {
+      "epoch": 2.5216919739696313,
+      "grad_norm": 0.13280880451202393,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1860
+    },
+    {
+      "epoch": 2.5230477223427332,
+      "grad_norm": 0.13232649862766266,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 1861
+    },
+    {
+      "epoch": 2.524403470715835,
+      "grad_norm": 0.1372680962085724,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 1862
+    },
+    {
+      "epoch": 2.525759219088937,
+      "grad_norm": 0.13191598653793335,
+      "learning_rate": 0.0001,
+      "loss": 0.1778,
+      "step": 1863
+    },
+    {
+      "epoch": 2.527114967462039,
+      "grad_norm": 0.13343602418899536,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1864
+    },
+    {
+      "epoch": 2.528470715835141,
+      "grad_norm": 0.12199801206588745,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1865
+    },
+    {
+      "epoch": 2.529826464208243,
+      "grad_norm": 0.12493172287940979,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1866
+    },
+    {
+      "epoch": 2.531182212581345,
+      "grad_norm": 0.12890738248825073,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 1867
+    },
+    {
+      "epoch": 2.5325379609544467,
+      "grad_norm": 0.12689892947673798,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 1868
+    },
+    {
+      "epoch": 2.5338937093275486,
+      "grad_norm": 0.1452014148235321,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1869
+    },
+    {
+      "epoch": 2.535249457700651,
+      "grad_norm": 0.1474027782678604,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 1870
+    },
+    {
+      "epoch": 2.5366052060737525,
+      "grad_norm": 0.1485602855682373,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1871
+    },
+    {
+      "epoch": 2.537960954446855,
+      "grad_norm": 0.14245092868804932,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 1872
+    },
+    {
+      "epoch": 2.5393167028199564,
+      "grad_norm": 0.14742210507392883,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 1873
+    },
+    {
+      "epoch": 2.5406724511930587,
+      "grad_norm": 0.1464325189590454,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 1874
+    },
+    {
+      "epoch": 2.5420281995661607,
+      "grad_norm": 0.14377877116203308,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 1875
+    },
+    {
+      "epoch": 2.5433839479392626,
+      "grad_norm": 0.1271800696849823,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 1876
+    },
+    {
+      "epoch": 2.5447396963123645,
+      "grad_norm": 0.1481257975101471,
+      "learning_rate": 0.0001,
+      "loss": 0.1844,
+      "step": 1877
+    },
+    {
+      "epoch": 2.5460954446854664,
+      "grad_norm": 0.14053773880004883,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1878
+    },
+    {
+      "epoch": 2.5474511930585684,
+      "grad_norm": 0.1297697126865387,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1879
+    },
+    {
+      "epoch": 2.5488069414316703,
+      "grad_norm": 0.12199372798204422,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 1880
+    },
+    {
+      "epoch": 2.5501626898047722,
+      "grad_norm": 0.10428114235401154,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 1881
+    },
+    {
+      "epoch": 2.551518438177874,
+      "grad_norm": 0.12577450275421143,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 1882
+    },
+    {
+      "epoch": 2.552874186550976,
+      "grad_norm": 0.12370701134204865,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 1883
+    },
+    {
+      "epoch": 2.554229934924078,
+      "grad_norm": 0.12553389370441437,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1884
+    },
+    {
+      "epoch": 2.55558568329718,
+      "grad_norm": 0.13276353478431702,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 1885
+    },
+    {
+      "epoch": 2.556941431670282,
+      "grad_norm": 0.14791543781757355,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 1886
+    },
+    {
+      "epoch": 2.558297180043384,
+      "grad_norm": 0.12053053826093674,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 1887
+    },
+    {
+      "epoch": 2.5596529284164857,
+      "grad_norm": 0.13749542832374573,
+      "learning_rate": 0.0001,
+      "loss": 0.1832,
+      "step": 1888
+    },
+    {
+      "epoch": 2.561008676789588,
+      "grad_norm": 0.12509578466415405,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 1889
+    },
+    {
+      "epoch": 2.5623644251626896,
+      "grad_norm": 0.11939408630132675,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 1890
+    },
+    {
+      "epoch": 2.563720173535792,
+      "grad_norm": 0.14279554784297943,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 1891
+    },
+    {
+      "epoch": 2.5650759219088934,
+      "grad_norm": 0.14142704010009766,
+      "learning_rate": 0.0001,
+      "loss": 0.1898,
+      "step": 1892
+    },
+    {
+      "epoch": 2.566431670281996,
+      "grad_norm": 0.12455581873655319,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 1893
+    },
+    {
+      "epoch": 2.5677874186550977,
+      "grad_norm": 0.13160939514636993,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1894
+    },
+    {
+      "epoch": 2.5691431670281997,
+      "grad_norm": 0.15024451911449432,
+      "learning_rate": 0.0001,
+      "loss": 0.1856,
+      "step": 1895
+    },
+    {
+      "epoch": 2.5704989154013016,
+      "grad_norm": 0.1242373138666153,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1896
+    },
+    {
+      "epoch": 2.5718546637744035,
+      "grad_norm": 0.12569019198417664,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 1897
+    },
+    {
+      "epoch": 2.5732104121475055,
+      "grad_norm": 0.133148655295372,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 1898
+    },
+    {
+      "epoch": 2.5745661605206074,
+      "grad_norm": 0.11414492130279541,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 1899
+    },
+    {
+      "epoch": 2.5759219088937093,
+      "grad_norm": 0.12560661137104034,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 1900
+    },
+    {
+      "epoch": 2.5772776572668112,
+      "grad_norm": 0.13206538558006287,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 1901
+    },
+    {
+      "epoch": 2.578633405639913,
+      "grad_norm": 0.14923961460590363,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 1902
+    },
+    {
+      "epoch": 2.579989154013015,
+      "grad_norm": 0.15315191447734833,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 1903
+    },
+    {
+      "epoch": 2.581344902386117,
+      "grad_norm": 0.14751382172107697,
+      "learning_rate": 0.0001,
+      "loss": 0.1802,
+      "step": 1904
+    },
+    {
+      "epoch": 2.582700650759219,
+      "grad_norm": 0.12548348307609558,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 1905
+    },
+    {
+      "epoch": 2.584056399132321,
+      "grad_norm": 0.16742448508739471,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "step": 1906
+    },
+    {
+      "epoch": 2.585412147505423,
+      "grad_norm": 0.12434748560190201,
+      "learning_rate": 0.0001,
+      "loss": 0.1564,
+      "step": 1907
+    },
+    {
+      "epoch": 2.586767895878525,
+      "grad_norm": 0.13706189393997192,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 1908
+    },
+    {
+      "epoch": 2.5881236442516267,
+      "grad_norm": 0.1266116350889206,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 1909
+    },
+    {
+      "epoch": 2.589479392624729,
+      "grad_norm": 0.15433481335639954,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 1910
+    },
+    {
+      "epoch": 2.590835140997831,
+      "grad_norm": 0.1440829038619995,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 1911
+    },
+    {
+      "epoch": 2.592190889370933,
+      "grad_norm": 0.1689286082983017,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 1912
+    },
+    {
+      "epoch": 2.593546637744035,
+      "grad_norm": 0.11618445068597794,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1913
+    },
+    {
+      "epoch": 2.5949023861171367,
+      "grad_norm": 0.11921945959329605,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 1914
+    },
+    {
+      "epoch": 2.5962581344902387,
+      "grad_norm": 0.12561652064323425,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1915
+    },
+    {
+      "epoch": 2.5976138828633406,
+      "grad_norm": 0.1339619755744934,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 1916
+    },
+    {
+      "epoch": 2.5989696312364425,
+      "grad_norm": 0.13795486092567444,
+      "learning_rate": 0.0001,
+      "loss": 0.1706,
+      "step": 1917
+    },
+    {
+      "epoch": 2.6003253796095445,
+      "grad_norm": 0.15075692534446716,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1918
+    },
+    {
+      "epoch": 2.6016811279826464,
+      "grad_norm": 0.14152175188064575,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 1919
+    },
+    {
+      "epoch": 2.6030368763557483,
+      "grad_norm": 0.12410273402929306,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 1920
+    },
+    {
+      "epoch": 2.6043926247288502,
+      "grad_norm": 0.13495725393295288,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 1921
+    },
+    {
+      "epoch": 2.605748373101952,
+      "grad_norm": 0.19644947350025177,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1922
+    },
+    {
+      "epoch": 2.607104121475054,
+      "grad_norm": 0.12734754383563995,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 1923
+    },
+    {
+      "epoch": 2.608459869848156,
+      "grad_norm": 0.1315966248512268,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 1924
+    },
+    {
+      "epoch": 2.6098156182212584,
+      "grad_norm": 0.1388804316520691,
+      "learning_rate": 0.0001,
+      "loss": 0.1991,
+      "step": 1925
+    },
+    {
+      "epoch": 2.61117136659436,
+      "grad_norm": 0.15680627524852753,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 1926
+    },
+    {
+      "epoch": 2.6125271149674623,
+      "grad_norm": 0.1269816756248474,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 1927
+    },
+    {
+      "epoch": 2.6138828633405637,
+      "grad_norm": 0.13718153536319733,
+      "learning_rate": 0.0001,
+      "loss": 0.1949,
+      "step": 1928
+    },
+    {
+      "epoch": 2.615238611713666,
+      "grad_norm": 0.12914173305034637,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 1929
+    },
+    {
+      "epoch": 2.616594360086768,
+      "grad_norm": 0.14232942461967468,
+      "learning_rate": 0.0001,
+      "loss": 0.1673,
+      "step": 1930
+    },
+    {
+      "epoch": 2.61795010845987,
+      "grad_norm": 0.129470556974411,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 1931
+    },
+    {
+      "epoch": 2.619305856832972,
+      "grad_norm": 0.14031817018985748,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 1932
+    },
+    {
+      "epoch": 2.620661605206074,
+      "grad_norm": 0.12831702828407288,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 1933
+    },
+    {
+      "epoch": 2.6220173535791758,
+      "grad_norm": 0.15820452570915222,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 1934
+    },
+    {
+      "epoch": 2.6233731019522777,
+      "grad_norm": 0.15755008161067963,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 1935
+    },
+    {
+      "epoch": 2.6247288503253796,
+      "grad_norm": 0.14792533218860626,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 1936
+    },
+    {
+      "epoch": 2.6260845986984815,
+      "grad_norm": 0.12934459745883942,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 1937
+    },
+    {
+      "epoch": 2.6274403470715835,
+      "grad_norm": 0.12777739763259888,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 1938
+    },
+    {
+      "epoch": 2.6287960954446854,
+      "grad_norm": 0.11808548122644424,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 1939
+    },
+    {
+      "epoch": 2.6301518438177873,
+      "grad_norm": 0.14801612496376038,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 1940
+    },
+    {
+      "epoch": 2.6315075921908893,
+      "grad_norm": 0.15699250996112823,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 1941
+    },
+    {
+      "epoch": 2.632863340563991,
+      "grad_norm": 0.13633574545383453,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 1942
+    },
+    {
+      "epoch": 2.634219088937093,
+      "grad_norm": 0.15441839396953583,
+      "learning_rate": 0.0001,
+      "loss": 0.1825,
+      "step": 1943
+    },
+    {
+      "epoch": 2.6355748373101955,
+      "grad_norm": 0.18334195017814636,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 1944
+    },
+    {
+      "epoch": 2.636930585683297,
+      "grad_norm": 0.14635561406612396,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 1945
+    },
+    {
+      "epoch": 2.6382863340563993,
+      "grad_norm": 0.1391373723745346,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1946
+    },
+    {
+      "epoch": 2.639642082429501,
+      "grad_norm": 0.1198771521449089,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 1947
+    },
+    {
+      "epoch": 2.640997830802603,
+      "grad_norm": 0.14645600318908691,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 1948
+    },
+    {
+      "epoch": 2.642353579175705,
+      "grad_norm": 0.165008544921875,
+      "learning_rate": 0.0001,
+      "loss": 0.1984,
+      "step": 1949
+    },
+    {
+      "epoch": 2.643709327548807,
+      "grad_norm": 0.1475944221019745,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 1950
+    },
+    {
+      "epoch": 2.645065075921909,
+      "grad_norm": 0.1383599042892456,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 1951
+    },
+    {
+      "epoch": 2.646420824295011,
+      "grad_norm": 0.1427065134048462,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 1952
+    },
+    {
+      "epoch": 2.647776572668113,
+      "grad_norm": 0.12833432853221893,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 1953
+    },
+    {
+      "epoch": 2.6491323210412148,
+      "grad_norm": 0.14999869465827942,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 1954
+    },
+    {
+      "epoch": 2.6504880694143167,
+      "grad_norm": 0.1497984677553177,
+      "learning_rate": 0.0001,
+      "loss": 0.1815,
+      "step": 1955
+    },
+    {
+      "epoch": 2.6518438177874186,
+      "grad_norm": 0.12127870321273804,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 1956
+    },
+    {
+      "epoch": 2.6531995661605206,
+      "grad_norm": 0.09923010319471359,
+      "learning_rate": 0.0001,
+      "loss": 0.1489,
+      "step": 1957
+    },
+    {
+      "epoch": 2.6545553145336225,
+      "grad_norm": 0.13323865830898285,
+      "learning_rate": 0.0001,
+      "loss": 0.1759,
+      "step": 1958
+    },
+    {
+      "epoch": 2.6559110629067244,
+      "grad_norm": 0.22315238416194916,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 1959
+    },
+    {
+      "epoch": 2.6572668112798263,
+      "grad_norm": 0.1184796392917633,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 1960
+    },
+    {
+      "epoch": 2.6586225596529283,
+      "grad_norm": 0.15067408978939056,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1961
+    },
+    {
+      "epoch": 2.65997830802603,
+      "grad_norm": 0.14283666014671326,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 1962
+    },
+    {
+      "epoch": 2.6613340563991326,
+      "grad_norm": 0.1382881999015808,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 1963
+    },
+    {
+      "epoch": 2.662689804772234,
+      "grad_norm": 0.13599041104316711,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1964
+    },
+    {
+      "epoch": 2.6640455531453364,
+      "grad_norm": 0.13776415586471558,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 1965
+    },
+    {
+      "epoch": 2.665401301518438,
+      "grad_norm": 0.12191472202539444,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 1966
+    },
+    {
+      "epoch": 2.6667570498915403,
+      "grad_norm": 0.13184627890586853,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 1967
+    },
+    {
+      "epoch": 2.668112798264642,
+      "grad_norm": 0.14143764972686768,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 1968
+    },
+    {
+      "epoch": 2.669468546637744,
+      "grad_norm": 0.1333450824022293,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 1969
+    },
+    {
+      "epoch": 2.670824295010846,
+      "grad_norm": 0.1790412962436676,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1970
+    },
+    {
+      "epoch": 2.672180043383948,
+      "grad_norm": 0.18152832984924316,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 1971
+    },
+    {
+      "epoch": 2.67353579175705,
+      "grad_norm": 0.12695561349391937,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 1972
+    },
+    {
+      "epoch": 2.674891540130152,
+      "grad_norm": 0.17696592211723328,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 1973
+    },
+    {
+      "epoch": 2.6762472885032538,
+      "grad_norm": 0.15102648735046387,
+      "learning_rate": 0.0001,
+      "loss": 0.1861,
+      "step": 1974
+    },
+    {
+      "epoch": 2.6776030368763557,
+      "grad_norm": 0.15162773430347443,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1975
+    },
+    {
+      "epoch": 2.6789587852494576,
+      "grad_norm": 0.1267489641904831,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 1976
+    },
+    {
+      "epoch": 2.6803145336225596,
+      "grad_norm": 0.12590762972831726,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 1977
+    },
+    {
+      "epoch": 2.6816702819956615,
+      "grad_norm": 0.14625167846679688,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 1978
+    },
+    {
+      "epoch": 2.6830260303687634,
+      "grad_norm": 0.14681249856948853,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 1979
+    },
+    {
+      "epoch": 2.684381778741866,
+      "grad_norm": 0.13122019171714783,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 1980
+    },
+    {
+      "epoch": 2.6857375271149673,
+      "grad_norm": 0.13073109090328217,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 1981
+    },
+    {
+      "epoch": 2.6870932754880696,
+      "grad_norm": 0.1315685361623764,
+      "learning_rate": 0.0001,
+      "loss": 0.1868,
+      "step": 1982
+    },
+    {
+      "epoch": 2.688449023861171,
+      "grad_norm": 0.14419212937355042,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 1983
+    },
+    {
+      "epoch": 2.6898047722342735,
+      "grad_norm": 0.1433769166469574,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 1984
+    },
+    {
+      "epoch": 2.6911605206073754,
+      "grad_norm": 0.12288834154605865,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 1985
+    },
+    {
+      "epoch": 2.6925162689804774,
+      "grad_norm": 0.1356910914182663,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1986
+    },
+    {
+      "epoch": 2.6938720173535793,
+      "grad_norm": 0.147144615650177,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 1987
+    },
+    {
+      "epoch": 2.695227765726681,
+      "grad_norm": 0.13220946490764618,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 1988
+    },
+    {
+      "epoch": 2.696583514099783,
+      "grad_norm": 0.173974871635437,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 1989
+    },
+    {
+      "epoch": 2.697939262472885,
+      "grad_norm": 0.13259093463420868,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 1990
+    },
+    {
+      "epoch": 2.699295010845987,
+      "grad_norm": 0.16548143327236176,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 1991
+    },
+    {
+      "epoch": 2.700650759219089,
+      "grad_norm": 0.14933434128761292,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 1992
+    },
+    {
+      "epoch": 2.702006507592191,
+      "grad_norm": 0.2362046092748642,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 1993
+    },
+    {
+      "epoch": 2.703362255965293,
+      "grad_norm": 0.13495609164237976,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 1994
+    },
+    {
+      "epoch": 2.7047180043383947,
+      "grad_norm": 0.12468313425779343,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 1995
+    },
+    {
+      "epoch": 2.7060737527114966,
+      "grad_norm": 0.15858210623264313,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 1996
+    },
+    {
+      "epoch": 2.7074295010845986,
+      "grad_norm": 0.12299206852912903,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 1997
+    },
+    {
+      "epoch": 2.7087852494577005,
+      "grad_norm": 0.14146758615970612,
+      "learning_rate": 0.0001,
+      "loss": 0.1913,
+      "step": 1998
+    },
+    {
+      "epoch": 2.710140997830803,
+      "grad_norm": 0.12425468116998672,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 1999
+    },
+    {
+      "epoch": 2.7114967462039044,
+      "grad_norm": 0.11883139610290527,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 2000
+    },
+    {
+      "epoch": 2.7114967462039044,
+      "eval_loss": 0.1897897720336914,
+      "eval_runtime": 595.6562,
+      "eval_samples_per_second": 4.644,
+      "eval_steps_per_second": 1.162,
+      "step": 2000
+    },
+    {
+      "epoch": 2.7128524945770067,
+      "grad_norm": 0.12397783249616623,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 2001
+    },
+    {
+      "epoch": 2.714208242950108,
+      "grad_norm": 0.12275779247283936,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2002
+    },
+    {
+      "epoch": 2.7155639913232106,
+      "grad_norm": 0.14315107464790344,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 2003
+    },
+    {
+      "epoch": 2.7169197396963125,
+      "grad_norm": 0.11354929208755493,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 2004
+    },
+    {
+      "epoch": 2.7182754880694144,
+      "grad_norm": 0.1379585862159729,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 2005
+    },
+    {
+      "epoch": 2.7196312364425164,
+      "grad_norm": 0.13700850307941437,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 2006
+    },
+    {
+      "epoch": 2.7209869848156183,
+      "grad_norm": 0.12351281195878983,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 2007
+    },
+    {
+      "epoch": 2.72234273318872,
+      "grad_norm": 0.1405455619096756,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 2008
+    },
+    {
+      "epoch": 2.723698481561822,
+      "grad_norm": 0.15236125886440277,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 2009
+    },
+    {
+      "epoch": 2.725054229934924,
+      "grad_norm": 0.14302493631839752,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 2010
+    },
+    {
+      "epoch": 2.726409978308026,
+      "grad_norm": 0.13301464915275574,
+      "learning_rate": 0.0001,
+      "loss": 0.1846,
+      "step": 2011
+    },
+    {
+      "epoch": 2.727765726681128,
+      "grad_norm": 0.38629502058029175,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 2012
+    },
+    {
+      "epoch": 2.72912147505423,
+      "grad_norm": 0.1269562840461731,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 2013
+    },
+    {
+      "epoch": 2.730477223427332,
+      "grad_norm": 0.13090528547763824,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 2014
+    },
+    {
+      "epoch": 2.7318329718004337,
+      "grad_norm": 0.24724939465522766,
+      "learning_rate": 0.0001,
+      "loss": 0.1858,
+      "step": 2015
+    },
+    {
+      "epoch": 2.7331887201735356,
+      "grad_norm": 0.14747942984104156,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 2016
+    },
+    {
+      "epoch": 2.7345444685466376,
+      "grad_norm": 0.12814967334270477,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2017
+    },
+    {
+      "epoch": 2.73590021691974,
+      "grad_norm": 0.1294751912355423,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2018
+    },
+    {
+      "epoch": 2.7372559652928414,
+      "grad_norm": 0.12861749529838562,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 2019
+    },
+    {
+      "epoch": 2.738611713665944,
+      "grad_norm": 0.1127142533659935,
+      "learning_rate": 0.0001,
+      "loss": 0.1753,
+      "step": 2020
+    },
+    {
+      "epoch": 2.7399674620390453,
+      "grad_norm": 0.13834327459335327,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 2021
+    },
+    {
+      "epoch": 2.7413232104121477,
+      "grad_norm": 0.14146527647972107,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 2022
+    },
+    {
+      "epoch": 2.7426789587852496,
+      "grad_norm": 0.2168230414390564,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 2023
+    },
+    {
+      "epoch": 2.7440347071583515,
+      "grad_norm": 0.13438177108764648,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 2024
+    },
+    {
+      "epoch": 2.7453904555314534,
+      "grad_norm": 0.1466539055109024,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 2025
+    },
+    {
+      "epoch": 2.7467462039045554,
+      "grad_norm": 0.13190555572509766,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 2026
+    },
+    {
+      "epoch": 2.7481019522776573,
+      "grad_norm": 0.1323634684085846,
+      "learning_rate": 0.0001,
+      "loss": 0.1767,
+      "step": 2027
+    },
+    {
+      "epoch": 2.7494577006507592,
+      "grad_norm": 0.15097901225090027,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 2028
+    },
+    {
+      "epoch": 2.750813449023861,
+      "grad_norm": 0.14419881999492645,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 2029
+    },
+    {
+      "epoch": 2.752169197396963,
+      "grad_norm": 0.12083493173122406,
+      "learning_rate": 0.0001,
+      "loss": 0.1696,
+      "step": 2030
+    },
+    {
+      "epoch": 2.753524945770065,
+      "grad_norm": 0.14197097718715668,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 2031
+    },
+    {
+      "epoch": 2.754880694143167,
+      "grad_norm": 0.1430617719888687,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 2032
+    },
+    {
+      "epoch": 2.756236442516269,
+      "grad_norm": 0.1246759295463562,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2033
+    },
+    {
+      "epoch": 2.757592190889371,
+      "grad_norm": 0.1293947398662567,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 2034
+    },
+    {
+      "epoch": 2.7589479392624727,
+      "grad_norm": 0.17829981446266174,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 2035
+    },
+    {
+      "epoch": 2.7603036876355747,
+      "grad_norm": 0.14870043098926544,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 2036
+    },
+    {
+      "epoch": 2.761659436008677,
+      "grad_norm": 0.15229080617427826,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 2037
+    },
+    {
+      "epoch": 2.7630151843817785,
+      "grad_norm": 0.15877912938594818,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 2038
+    },
+    {
+      "epoch": 2.764370932754881,
+      "grad_norm": 0.1829831600189209,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 2039
+    },
+    {
+      "epoch": 2.765726681127983,
+      "grad_norm": 0.1173824742436409,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2040
+    },
+    {
+      "epoch": 2.7670824295010847,
+      "grad_norm": 0.1389331817626953,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2041
+    },
+    {
+      "epoch": 2.7684381778741867,
+      "grad_norm": 0.13311904668807983,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2042
+    },
+    {
+      "epoch": 2.7697939262472886,
+      "grad_norm": 0.14681003987789154,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2043
+    },
+    {
+      "epoch": 2.7711496746203905,
+      "grad_norm": 0.14954963326454163,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 2044
+    },
+    {
+      "epoch": 2.7725054229934925,
+      "grad_norm": 0.11881741136312485,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2045
+    },
+    {
+      "epoch": 2.7738611713665944,
+      "grad_norm": 0.12055434286594391,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2046
+    },
+    {
+      "epoch": 2.7752169197396963,
+      "grad_norm": 0.13618500530719757,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2047
+    },
+    {
+      "epoch": 2.7765726681127982,
+      "grad_norm": 0.12559756636619568,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 2048
+    },
+    {
+      "epoch": 2.7779284164859,
+      "grad_norm": 0.14541500806808472,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 2049
+    },
+    {
+      "epoch": 2.779284164859002,
+      "grad_norm": 0.13511964678764343,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2050
+    },
+    {
+      "epoch": 2.780639913232104,
+      "grad_norm": 0.12749966979026794,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2051
+    },
+    {
+      "epoch": 2.781995661605206,
+      "grad_norm": 0.11417023092508316,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2052
+    },
+    {
+      "epoch": 2.783351409978308,
+      "grad_norm": 0.12989024817943573,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 2053
+    },
+    {
+      "epoch": 2.7847071583514102,
+      "grad_norm": 0.13588270545005798,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 2054
+    },
+    {
+      "epoch": 2.7860629067245117,
+      "grad_norm": 0.14273977279663086,
+      "learning_rate": 0.0001,
+      "loss": 0.1814,
+      "step": 2055
+    },
+    {
+      "epoch": 2.787418655097614,
+      "grad_norm": 0.15713432431221008,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2056
+    },
+    {
+      "epoch": 2.7887744034707156,
+      "grad_norm": 0.1456521451473236,
+      "learning_rate": 0.0001,
+      "loss": 0.181,
+      "step": 2057
+    },
+    {
+      "epoch": 2.790130151843818,
+      "grad_norm": 0.1191895604133606,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2058
+    },
+    {
+      "epoch": 2.79148590021692,
+      "grad_norm": 0.14499974250793457,
+      "learning_rate": 0.0001,
+      "loss": 0.1835,
+      "step": 2059
+    },
+    {
+      "epoch": 2.792841648590022,
+      "grad_norm": 0.12618711590766907,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 2060
+    },
+    {
+      "epoch": 2.7941973969631237,
+      "grad_norm": 0.1511884182691574,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2061
+    },
+    {
+      "epoch": 2.7955531453362257,
+      "grad_norm": 0.13503320515155792,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 2062
+    },
+    {
+      "epoch": 2.7969088937093276,
+      "grad_norm": 0.1364138126373291,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 2063
+    },
+    {
+      "epoch": 2.7982646420824295,
+      "grad_norm": 0.14437627792358398,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2064
+    },
+    {
+      "epoch": 2.7996203904555315,
+      "grad_norm": 0.12529070675373077,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2065
+    },
+    {
+      "epoch": 2.8009761388286334,
+      "grad_norm": 0.14226627349853516,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 2066
+    },
+    {
+      "epoch": 2.8023318872017353,
+      "grad_norm": 0.1430516541004181,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 2067
+    },
+    {
+      "epoch": 2.8036876355748372,
+      "grad_norm": 0.11484866589307785,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2068
+    },
+    {
+      "epoch": 2.805043383947939,
+      "grad_norm": 0.1336757391691208,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 2069
+    },
+    {
+      "epoch": 2.806399132321041,
+      "grad_norm": 0.1148701086640358,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 2070
+    },
+    {
+      "epoch": 2.807754880694143,
+      "grad_norm": 0.14043770730495453,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 2071
+    },
+    {
+      "epoch": 2.809110629067245,
+      "grad_norm": 0.13801445066928864,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 2072
+    },
+    {
+      "epoch": 2.8104663774403473,
+      "grad_norm": 0.16923271119594574,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 2073
+    },
+    {
+      "epoch": 2.811822125813449,
+      "grad_norm": 0.14196190237998962,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 2074
+    },
+    {
+      "epoch": 2.813177874186551,
+      "grad_norm": 0.15726996958255768,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 2075
+    },
+    {
+      "epoch": 2.8145336225596527,
+      "grad_norm": 0.2199648767709732,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 2076
+    },
+    {
+      "epoch": 2.815889370932755,
+      "grad_norm": 0.17038697004318237,
+      "learning_rate": 0.0001,
+      "loss": 0.18,
+      "step": 2077
+    },
+    {
+      "epoch": 2.817245119305857,
+      "grad_norm": 0.13924957811832428,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 2078
+    },
+    {
+      "epoch": 2.818600867678959,
+      "grad_norm": 0.14736485481262207,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2079
+    },
+    {
+      "epoch": 2.819956616052061,
+      "grad_norm": 0.13877762854099274,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 2080
+    },
+    {
+      "epoch": 2.8213123644251628,
+      "grad_norm": 0.15414167940616608,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 2081
+    },
+    {
+      "epoch": 2.8226681127982647,
+      "grad_norm": 0.13077403604984283,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 2082
+    },
+    {
+      "epoch": 2.8240238611713666,
+      "grad_norm": 0.12939341366291046,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2083
+    },
+    {
+      "epoch": 2.8253796095444685,
+      "grad_norm": 0.13884030282497406,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2084
+    },
+    {
+      "epoch": 2.8267353579175705,
+      "grad_norm": 0.1523071676492691,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 2085
+    },
+    {
+      "epoch": 2.8280911062906724,
+      "grad_norm": 0.1512306034564972,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 2086
+    },
+    {
+      "epoch": 2.8294468546637743,
+      "grad_norm": 0.12778601050376892,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 2087
+    },
+    {
+      "epoch": 2.8308026030368763,
+      "grad_norm": 0.11462337523698807,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 2088
+    },
+    {
+      "epoch": 2.832158351409978,
+      "grad_norm": 0.153813436627388,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 2089
+    },
+    {
+      "epoch": 2.83351409978308,
+      "grad_norm": 0.14251075685024261,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 2090
+    },
+    {
+      "epoch": 2.834869848156182,
+      "grad_norm": 0.14991649985313416,
+      "learning_rate": 0.0001,
+      "loss": 0.1847,
+      "step": 2091
+    },
+    {
+      "epoch": 2.8362255965292844,
+      "grad_norm": 0.14339275658130646,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2092
+    },
+    {
+      "epoch": 2.837581344902386,
+      "grad_norm": 0.13376349210739136,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2093
+    },
+    {
+      "epoch": 2.8389370932754883,
+      "grad_norm": 0.15242373943328857,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 2094
+    },
+    {
+      "epoch": 2.8402928416485898,
+      "grad_norm": 0.11569581180810928,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 2095
+    },
+    {
+      "epoch": 2.841648590021692,
+      "grad_norm": 0.13502709567546844,
+      "learning_rate": 0.0001,
+      "loss": 0.1799,
+      "step": 2096
+    },
+    {
+      "epoch": 2.843004338394794,
+      "grad_norm": 0.14046362042427063,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 2097
+    },
+    {
+      "epoch": 2.844360086767896,
+      "grad_norm": 0.12773831188678741,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 2098
+    },
+    {
+      "epoch": 2.845715835140998,
+      "grad_norm": 0.1311541646718979,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 2099
+    },
+    {
+      "epoch": 2.8470715835141,
+      "grad_norm": 0.11310502886772156,
+      "learning_rate": 0.0001,
+      "loss": 0.1811,
+      "step": 2100
+    },
+    {
+      "epoch": 2.8484273318872018,
+      "grad_norm": 0.1597481220960617,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 2101
+    },
+    {
+      "epoch": 2.8497830802603037,
+      "grad_norm": 0.160044327378273,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 2102
+    },
+    {
+      "epoch": 2.8511388286334056,
+      "grad_norm": 0.1390226185321808,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 2103
+    },
+    {
+      "epoch": 2.8524945770065075,
+      "grad_norm": 0.13010533154010773,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 2104
+    },
+    {
+      "epoch": 2.8538503253796095,
+      "grad_norm": 0.15699219703674316,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 2105
+    },
+    {
+      "epoch": 2.8552060737527114,
+      "grad_norm": 0.12529852986335754,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 2106
+    },
+    {
+      "epoch": 2.8565618221258133,
+      "grad_norm": 0.12192243337631226,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 2107
+    },
+    {
+      "epoch": 2.8579175704989153,
+      "grad_norm": 0.14108632504940033,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 2108
+    },
+    {
+      "epoch": 2.859273318872017,
+      "grad_norm": 0.13616618514060974,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 2109
+    },
+    {
+      "epoch": 2.860629067245119,
+      "grad_norm": 0.13566507399082184,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 2110
+    },
+    {
+      "epoch": 2.8619848156182215,
+      "grad_norm": 0.11854743212461472,
+      "learning_rate": 0.0001,
+      "loss": 0.1712,
+      "step": 2111
+    },
+    {
+      "epoch": 2.863340563991323,
+      "grad_norm": 0.12565210461616516,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2112
+    },
+    {
+      "epoch": 2.8646963123644253,
+      "grad_norm": 0.1160890981554985,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2113
+    },
+    {
+      "epoch": 2.8660520607375273,
+      "grad_norm": 0.13395078480243683,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 2114
+    },
+    {
+      "epoch": 2.867407809110629,
+      "grad_norm": 0.12654687464237213,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2115
+    },
+    {
+      "epoch": 2.868763557483731,
+      "grad_norm": 0.152633398771286,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 2116
+    },
+    {
+      "epoch": 2.870119305856833,
+      "grad_norm": 0.13074658811092377,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 2117
+    },
+    {
+      "epoch": 2.871475054229935,
+      "grad_norm": 0.14301712810993195,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2118
+    },
+    {
+      "epoch": 2.872830802603037,
+      "grad_norm": 0.1383468210697174,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 2119
+    },
+    {
+      "epoch": 2.874186550976139,
+      "grad_norm": 0.1486644446849823,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2120
+    },
+    {
+      "epoch": 2.8755422993492408,
+      "grad_norm": 0.1446397602558136,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2121
+    },
+    {
+      "epoch": 2.8768980477223427,
+      "grad_norm": 0.1732262521982193,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 2122
+    },
+    {
+      "epoch": 2.8782537960954446,
+      "grad_norm": 0.15687234699726105,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 2123
+    },
+    {
+      "epoch": 2.8796095444685466,
+      "grad_norm": 0.14339923858642578,
+      "learning_rate": 0.0001,
+      "loss": 0.1812,
+      "step": 2124
+    },
+    {
+      "epoch": 2.8809652928416485,
+      "grad_norm": 0.15801295638084412,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 2125
+    },
+    {
+      "epoch": 2.8823210412147504,
+      "grad_norm": 0.14594346284866333,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 2126
+    },
+    {
+      "epoch": 2.8836767895878523,
+      "grad_norm": 0.15615153312683105,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 2127
+    },
+    {
+      "epoch": 2.8850325379609547,
+      "grad_norm": 0.1376756876707077,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 2128
+    },
+    {
+      "epoch": 2.886388286334056,
+      "grad_norm": 0.12823255360126495,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 2129
+    },
+    {
+      "epoch": 2.8877440347071586,
+      "grad_norm": 0.15170139074325562,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 2130
+    },
+    {
+      "epoch": 2.88909978308026,
+      "grad_norm": 0.15942566096782684,
+      "learning_rate": 0.0001,
+      "loss": 0.1818,
+      "step": 2131
+    },
+    {
+      "epoch": 2.8904555314533624,
+      "grad_norm": 0.12604333460330963,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2132
+    },
+    {
+      "epoch": 2.8918112798264644,
+      "grad_norm": 0.14116209745407104,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 2133
+    },
+    {
+      "epoch": 2.8931670281995663,
+      "grad_norm": 0.11690447479486465,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2134
+    },
+    {
+      "epoch": 2.894522776572668,
+      "grad_norm": 0.11743026226758957,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2135
+    },
+    {
+      "epoch": 2.89587852494577,
+      "grad_norm": 0.15033484995365143,
+      "learning_rate": 0.0001,
+      "loss": 0.1897,
+      "step": 2136
+    },
+    {
+      "epoch": 2.897234273318872,
+      "grad_norm": 0.11322303861379623,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2137
+    },
+    {
+      "epoch": 2.898590021691974,
+      "grad_norm": 0.13592827320098877,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 2138
+    },
+    {
+      "epoch": 2.899945770065076,
+      "grad_norm": 0.1211232990026474,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2139
+    },
+    {
+      "epoch": 2.901301518438178,
+      "grad_norm": 0.11984856426715851,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 2140
+    },
+    {
+      "epoch": 2.90265726681128,
+      "grad_norm": 0.13312803208827972,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 2141
+    },
+    {
+      "epoch": 2.9040130151843817,
+      "grad_norm": 0.13648386299610138,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 2142
+    },
+    {
+      "epoch": 2.9053687635574836,
+      "grad_norm": 0.13040630519390106,
+      "learning_rate": 0.0001,
+      "loss": 0.1731,
+      "step": 2143
+    },
+    {
+      "epoch": 2.9067245119305856,
+      "grad_norm": 0.1259726881980896,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 2144
+    },
+    {
+      "epoch": 2.9080802603036875,
+      "grad_norm": 0.14458923041820526,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2145
+    },
+    {
+      "epoch": 2.9094360086767894,
+      "grad_norm": 0.12854906916618347,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 2146
+    },
+    {
+      "epoch": 2.910791757049892,
+      "grad_norm": 0.15396557748317719,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 2147
+    },
+    {
+      "epoch": 2.9121475054229933,
+      "grad_norm": 0.13944202661514282,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 2148
+    },
+    {
+      "epoch": 2.9135032537960956,
+      "grad_norm": 0.14258044958114624,
+      "learning_rate": 0.0001,
+      "loss": 0.1822,
+      "step": 2149
+    },
+    {
+      "epoch": 2.914859002169197,
+      "grad_norm": 0.139817476272583,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2150
+    },
+    {
+      "epoch": 2.9162147505422995,
+      "grad_norm": 0.13093924522399902,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 2151
+    },
+    {
+      "epoch": 2.9175704989154014,
+      "grad_norm": 0.13620473444461823,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 2152
+    },
+    {
+      "epoch": 2.9189262472885034,
+      "grad_norm": 0.13285833597183228,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2153
+    },
+    {
+      "epoch": 2.9202819956616053,
+      "grad_norm": 0.1278178095817566,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 2154
+    },
+    {
+      "epoch": 2.921637744034707,
+      "grad_norm": 0.13010932505130768,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2155
+    },
+    {
+      "epoch": 2.922993492407809,
+      "grad_norm": 0.13045336306095123,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 2156
+    },
+    {
+      "epoch": 2.924349240780911,
+      "grad_norm": 0.11513189971446991,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2157
+    },
+    {
+      "epoch": 2.925704989154013,
+      "grad_norm": 0.1280595064163208,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 2158
+    },
+    {
+      "epoch": 2.927060737527115,
+      "grad_norm": 0.13762879371643066,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 2159
+    },
+    {
+      "epoch": 2.928416485900217,
+      "grad_norm": 0.1397835910320282,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 2160
+    },
+    {
+      "epoch": 2.929772234273319,
+      "grad_norm": 0.11842179298400879,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 2161
+    },
+    {
+      "epoch": 2.9311279826464207,
+      "grad_norm": 0.12166810035705566,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 2162
+    },
+    {
+      "epoch": 2.9324837310195226,
+      "grad_norm": 0.1664976328611374,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 2163
+    },
+    {
+      "epoch": 2.9338394793926246,
+      "grad_norm": 0.11895212531089783,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 2164
+    },
+    {
+      "epoch": 2.9351952277657265,
+      "grad_norm": 0.13561517000198364,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2165
+    },
+    {
+      "epoch": 2.936550976138829,
+      "grad_norm": 0.15283958613872528,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 2166
+    },
+    {
+      "epoch": 2.9379067245119304,
+      "grad_norm": 0.13208049535751343,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2167
+    },
+    {
+      "epoch": 2.9392624728850327,
+      "grad_norm": 0.15181857347488403,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 2168
+    },
+    {
+      "epoch": 2.940618221258134,
+      "grad_norm": 0.13352788984775543,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2169
+    },
+    {
+      "epoch": 2.9419739696312366,
+      "grad_norm": 0.15127092599868774,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 2170
+    },
+    {
+      "epoch": 2.9433297180043385,
+      "grad_norm": 0.14183396100997925,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 2171
+    },
+    {
+      "epoch": 2.9446854663774404,
+      "grad_norm": 0.14630752801895142,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 2172
+    },
+    {
+      "epoch": 2.9460412147505424,
+      "grad_norm": 0.13918223977088928,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 2173
+    },
+    {
+      "epoch": 2.9473969631236443,
+      "grad_norm": 0.14010651409626007,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 2174
+    },
+    {
+      "epoch": 2.9487527114967462,
+      "grad_norm": 0.1406862735748291,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2175
+    },
+    {
+      "epoch": 2.950108459869848,
+      "grad_norm": 0.14357620477676392,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 2176
+    },
+    {
+      "epoch": 2.95146420824295,
+      "grad_norm": 0.140491783618927,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 2177
+    },
+    {
+      "epoch": 2.952819956616052,
+      "grad_norm": 0.13593845069408417,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2178
+    },
+    {
+      "epoch": 2.954175704989154,
+      "grad_norm": 0.14203645288944244,
+      "learning_rate": 0.0001,
+      "loss": 0.1776,
+      "step": 2179
+    },
+    {
+      "epoch": 2.955531453362256,
+      "grad_norm": 0.14183808863162994,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2180
+    },
+    {
+      "epoch": 2.956887201735358,
+      "grad_norm": 0.11786481738090515,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2181
+    },
+    {
+      "epoch": 2.9582429501084597,
+      "grad_norm": 0.14809578657150269,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2182
+    },
+    {
+      "epoch": 2.959598698481562,
+      "grad_norm": 0.1340261548757553,
+      "learning_rate": 0.0001,
+      "loss": 0.1829,
+      "step": 2183
+    },
+    {
+      "epoch": 2.9609544468546636,
+      "grad_norm": 0.12443133443593979,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2184
+    },
+    {
+      "epoch": 2.962310195227766,
+      "grad_norm": 0.138412743806839,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 2185
+    },
+    {
+      "epoch": 2.9636659436008674,
+      "grad_norm": 0.13598176836967468,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 2186
+    },
+    {
+      "epoch": 2.96502169197397,
+      "grad_norm": 0.15635687112808228,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2187
+    },
+    {
+      "epoch": 2.9663774403470717,
+      "grad_norm": 0.16849760711193085,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 2188
+    },
+    {
+      "epoch": 2.9677331887201737,
+      "grad_norm": 0.14587105810642242,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 2189
+    },
+    {
+      "epoch": 2.9690889370932756,
+      "grad_norm": 0.15269461274147034,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 2190
+    },
+    {
+      "epoch": 2.9704446854663775,
+      "grad_norm": 0.16920822858810425,
+      "learning_rate": 0.0001,
+      "loss": 0.1784,
+      "step": 2191
+    },
+    {
+      "epoch": 2.9718004338394794,
+      "grad_norm": 0.12815198302268982,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 2192
+    },
+    {
+      "epoch": 2.9731561822125814,
+      "grad_norm": 0.1632385551929474,
+      "learning_rate": 0.0001,
+      "loss": 0.1869,
+      "step": 2193
+    },
+    {
+      "epoch": 2.9745119305856833,
+      "grad_norm": 0.13035379350185394,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 2194
+    },
+    {
+      "epoch": 2.9758676789587852,
+      "grad_norm": 0.13182269036769867,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 2195
+    },
+    {
+      "epoch": 2.977223427331887,
+      "grad_norm": 0.13356974720954895,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 2196
+    },
+    {
+      "epoch": 2.978579175704989,
+      "grad_norm": 0.13227279484272003,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2197
+    },
+    {
+      "epoch": 2.979934924078091,
+      "grad_norm": 0.16959238052368164,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2198
+    },
+    {
+      "epoch": 2.981290672451193,
+      "grad_norm": 0.16944217681884766,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 2199
+    },
+    {
+      "epoch": 2.982646420824295,
+      "grad_norm": 0.14828181266784668,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 2200
+    },
+    {
+      "epoch": 2.984002169197397,
+      "grad_norm": 0.14222432672977448,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 2201
+    },
+    {
+      "epoch": 2.985357917570499,
+      "grad_norm": 0.13746659457683563,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2202
+    },
+    {
+      "epoch": 2.9867136659436007,
+      "grad_norm": 0.13914121687412262,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2203
+    },
+    {
+      "epoch": 2.988069414316703,
+      "grad_norm": 0.16071587800979614,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2204
+    },
+    {
+      "epoch": 2.9894251626898045,
+      "grad_norm": 0.12272656708955765,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 2205
+    },
+    {
+      "epoch": 2.990780911062907,
+      "grad_norm": 0.14005914330482483,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2206
+    },
+    {
+      "epoch": 2.992136659436009,
+      "grad_norm": 0.13052700459957123,
+      "learning_rate": 0.0001,
+      "loss": 0.1772,
+      "step": 2207
+    },
+    {
+      "epoch": 2.9934924078091107,
+      "grad_norm": 0.13045227527618408,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2208
+    },
+    {
+      "epoch": 2.9948481561822127,
+      "grad_norm": 0.15388114750385284,
+      "learning_rate": 0.0001,
+      "loss": 0.1821,
+      "step": 2209
+    },
+    {
+      "epoch": 2.9962039045553146,
+      "grad_norm": 0.140912726521492,
+      "learning_rate": 0.0001,
+      "loss": 0.1791,
+      "step": 2210
+    },
+    {
+      "epoch": 2.9975596529284165,
+      "grad_norm": 0.14813022315502167,
+      "learning_rate": 0.0001,
+      "loss": 0.1782,
+      "step": 2211
+    },
+    {
+      "epoch": 2.9989154013015185,
+      "grad_norm": 0.1417742520570755,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 2212
+    },
+    {
+      "epoch": 3.0002711496746204,
+      "grad_norm": 0.13807924091815948,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 2213
+    },
+    {
+      "epoch": 3.0016268980477223,
+      "grad_norm": 0.14138858020305634,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2214
+    },
+    {
+      "epoch": 3.0029826464208242,
+      "grad_norm": 0.11364789307117462,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 2215
+    },
+    {
+      "epoch": 3.004338394793926,
+      "grad_norm": 0.14850568771362305,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2216
+    },
+    {
+      "epoch": 3.005694143167028,
+      "grad_norm": 0.13135933876037598,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2217
+    },
+    {
+      "epoch": 3.00704989154013,
+      "grad_norm": 0.136728897690773,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 2218
+    },
+    {
+      "epoch": 3.008405639913232,
+      "grad_norm": 0.1431213915348053,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 2219
+    },
+    {
+      "epoch": 3.009761388286334,
+      "grad_norm": 0.1464388370513916,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 2220
+    },
+    {
+      "epoch": 3.011117136659436,
+      "grad_norm": 0.13815464079380035,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2221
+    },
+    {
+      "epoch": 3.0124728850325377,
+      "grad_norm": 0.1383426934480667,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2222
+    },
+    {
+      "epoch": 3.01382863340564,
+      "grad_norm": 0.16520783305168152,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2223
+    },
+    {
+      "epoch": 3.015184381778742,
+      "grad_norm": 0.15953969955444336,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2224
+    },
+    {
+      "epoch": 3.016540130151844,
+      "grad_norm": 0.17421478033065796,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 2225
+    },
+    {
+      "epoch": 3.017895878524946,
+      "grad_norm": 0.1434691846370697,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 2226
+    },
+    {
+      "epoch": 3.019251626898048,
+      "grad_norm": 0.15614356100559235,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2227
+    },
+    {
+      "epoch": 3.0206073752711498,
+      "grad_norm": 0.1709403544664383,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 2228
+    },
+    {
+      "epoch": 3.0219631236442517,
+      "grad_norm": 0.15666942298412323,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 2229
+    },
+    {
+      "epoch": 3.0233188720173536,
+      "grad_norm": 0.1859796941280365,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2230
+    },
+    {
+      "epoch": 3.0246746203904555,
+      "grad_norm": 0.1576000601053238,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2231
+    },
+    {
+      "epoch": 3.0260303687635575,
+      "grad_norm": 0.1873348355293274,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 2232
+    },
+    {
+      "epoch": 3.0273861171366594,
+      "grad_norm": 0.18812987208366394,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2233
+    },
+    {
+      "epoch": 3.0287418655097613,
+      "grad_norm": 0.17330804467201233,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2234
+    },
+    {
+      "epoch": 3.0300976138828633,
+      "grad_norm": 0.14229650795459747,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 2235
+    },
+    {
+      "epoch": 3.031453362255965,
+      "grad_norm": 0.14434930682182312,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2236
+    },
+    {
+      "epoch": 3.032809110629067,
+      "grad_norm": 0.15087705850601196,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 2237
+    },
+    {
+      "epoch": 3.034164859002169,
+      "grad_norm": 0.23194921016693115,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 2238
+    },
+    {
+      "epoch": 3.035520607375271,
+      "grad_norm": 0.14184103906154633,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2239
+    },
+    {
+      "epoch": 3.036876355748373,
+      "grad_norm": 0.1501649171113968,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 2240
+    },
+    {
+      "epoch": 3.0382321041214753,
+      "grad_norm": 0.15698102116584778,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2241
+    },
+    {
+      "epoch": 3.039587852494577,
+      "grad_norm": 0.13590776920318604,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2242
+    },
+    {
+      "epoch": 3.040943600867679,
+      "grad_norm": 0.14524251222610474,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 2243
+    },
+    {
+      "epoch": 3.042299349240781,
+      "grad_norm": 0.15004342794418335,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2244
+    },
+    {
+      "epoch": 3.043655097613883,
+      "grad_norm": 0.14076711237430573,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 2245
+    },
+    {
+      "epoch": 3.045010845986985,
+      "grad_norm": 0.12258098274469376,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 2246
+    },
+    {
+      "epoch": 3.046366594360087,
+      "grad_norm": 0.15866003930568695,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2247
+    },
+    {
+      "epoch": 3.0477223427331888,
+      "grad_norm": 0.16526712477207184,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 2248
+    },
+    {
+      "epoch": 3.0490780911062907,
+      "grad_norm": 0.14540021121501923,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2249
+    },
+    {
+      "epoch": 3.0504338394793926,
+      "grad_norm": 0.13921208679676056,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 2250
+    },
+    {
+      "epoch": 3.0517895878524945,
+      "grad_norm": 0.15207825601100922,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 2251
+    },
+    {
+      "epoch": 3.0531453362255965,
+      "grad_norm": 0.13675197958946228,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2252
+    },
+    {
+      "epoch": 3.0545010845986984,
+      "grad_norm": 0.13035574555397034,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 2253
+    },
+    {
+      "epoch": 3.0558568329718003,
+      "grad_norm": 0.14229965209960938,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 2254
+    },
+    {
+      "epoch": 3.0572125813449023,
+      "grad_norm": 0.1502005159854889,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2255
+    },
+    {
+      "epoch": 3.058568329718004,
+      "grad_norm": 0.16858530044555664,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2256
+    },
+    {
+      "epoch": 3.059924078091106,
+      "grad_norm": 0.16003169119358063,
+      "learning_rate": 0.0001,
+      "loss": 0.1539,
+      "step": 2257
+    },
+    {
+      "epoch": 3.061279826464208,
+      "grad_norm": 0.16369187831878662,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 2258
+    },
+    {
+      "epoch": 3.0626355748373104,
+      "grad_norm": 0.15883556008338928,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 2259
+    },
+    {
+      "epoch": 3.0639913232104123,
+      "grad_norm": 0.14656442403793335,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 2260
+    },
+    {
+      "epoch": 3.0653470715835143,
+      "grad_norm": 0.33001768589019775,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2261
+    },
+    {
+      "epoch": 3.066702819956616,
+      "grad_norm": 0.153219074010849,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 2262
+    },
+    {
+      "epoch": 3.068058568329718,
+      "grad_norm": 0.1779169738292694,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 2263
+    },
+    {
+      "epoch": 3.06941431670282,
+      "grad_norm": 0.14143891632556915,
+      "learning_rate": 0.0001,
+      "loss": 0.1508,
+      "step": 2264
+    },
+    {
+      "epoch": 3.070770065075922,
+      "grad_norm": 0.1325828731060028,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 2265
+    },
+    {
+      "epoch": 3.072125813449024,
+      "grad_norm": 0.1314278542995453,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 2266
+    },
+    {
+      "epoch": 3.073481561822126,
+      "grad_norm": 0.13844600319862366,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 2267
+    },
+    {
+      "epoch": 3.0748373101952278,
+      "grad_norm": 0.1527375727891922,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2268
+    },
+    {
+      "epoch": 3.0761930585683297,
+      "grad_norm": 0.14922308921813965,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2269
+    },
+    {
+      "epoch": 3.0775488069414316,
+      "grad_norm": 0.1705380380153656,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 2270
+    },
+    {
+      "epoch": 3.0789045553145336,
+      "grad_norm": 0.1337367743253708,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 2271
+    },
+    {
+      "epoch": 3.0802603036876355,
+      "grad_norm": 0.1621292233467102,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2272
+    },
+    {
+      "epoch": 3.0816160520607374,
+      "grad_norm": 0.13324204087257385,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2273
+    },
+    {
+      "epoch": 3.0829718004338393,
+      "grad_norm": 0.16292999684810638,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 2274
+    },
+    {
+      "epoch": 3.0843275488069413,
+      "grad_norm": 0.17439806461334229,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2275
+    },
+    {
+      "epoch": 3.085683297180043,
+      "grad_norm": 0.13746024668216705,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 2276
+    },
+    {
+      "epoch": 3.087039045553145,
+      "grad_norm": 0.17941106855869293,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 2277
+    },
+    {
+      "epoch": 3.0883947939262475,
+      "grad_norm": 0.14422409236431122,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2278
+    },
+    {
+      "epoch": 3.0897505422993494,
+      "grad_norm": 0.13735170662403107,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 2279
+    },
+    {
+      "epoch": 3.0911062906724514,
+      "grad_norm": 0.1612592488527298,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 2280
+    },
+    {
+      "epoch": 3.0924620390455533,
+      "grad_norm": 0.16273127496242523,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2281
+    },
+    {
+      "epoch": 3.093817787418655,
+      "grad_norm": 0.1495095044374466,
+      "learning_rate": 0.0001,
+      "loss": 0.1619,
+      "step": 2282
+    },
+    {
+      "epoch": 3.095173535791757,
+      "grad_norm": 0.1490403264760971,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 2283
+    },
+    {
+      "epoch": 3.096529284164859,
+      "grad_norm": 0.14860506355762482,
+      "learning_rate": 0.0001,
+      "loss": 0.1756,
+      "step": 2284
+    },
+    {
+      "epoch": 3.097885032537961,
+      "grad_norm": 0.17510156333446503,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 2285
+    },
+    {
+      "epoch": 3.099240780911063,
+      "grad_norm": 0.15011431276798248,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 2286
+    },
+    {
+      "epoch": 3.100596529284165,
+      "grad_norm": 0.1781640201807022,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 2287
+    },
+    {
+      "epoch": 3.1019522776572668,
+      "grad_norm": 0.17372199892997742,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 2288
+    },
+    {
+      "epoch": 3.1033080260303687,
+      "grad_norm": 0.15176717936992645,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 2289
+    },
+    {
+      "epoch": 3.1046637744034706,
+      "grad_norm": 0.16659778356552124,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 2290
+    },
+    {
+      "epoch": 3.1060195227765726,
+      "grad_norm": 0.14768049120903015,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 2291
+    },
+    {
+      "epoch": 3.1073752711496745,
+      "grad_norm": 0.1401548832654953,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2292
+    },
+    {
+      "epoch": 3.1087310195227764,
+      "grad_norm": 0.1575610637664795,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 2293
+    },
+    {
+      "epoch": 3.1100867678958783,
+      "grad_norm": 0.16102854907512665,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 2294
+    },
+    {
+      "epoch": 3.1114425162689803,
+      "grad_norm": 0.1624155044555664,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 2295
+    },
+    {
+      "epoch": 3.112798264642082,
+      "grad_norm": 0.13114473223686218,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 2296
+    },
+    {
+      "epoch": 3.1141540130151846,
+      "grad_norm": 0.1467607617378235,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2297
+    },
+    {
+      "epoch": 3.1155097613882865,
+      "grad_norm": 0.13350450992584229,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2298
+    },
+    {
+      "epoch": 3.1168655097613884,
+      "grad_norm": 0.1487187296152115,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2299
+    },
+    {
+      "epoch": 3.1182212581344904,
+      "grad_norm": 0.16139210760593414,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2300
+    },
+    {
+      "epoch": 3.1195770065075923,
+      "grad_norm": 0.15119358897209167,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2301
+    },
+    {
+      "epoch": 3.120932754880694,
+      "grad_norm": 0.13180941343307495,
+      "learning_rate": 0.0001,
+      "loss": 0.1406,
+      "step": 2302
+    },
+    {
+      "epoch": 3.122288503253796,
+      "grad_norm": 0.1811341494321823,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 2303
+    },
+    {
+      "epoch": 3.123644251626898,
+      "grad_norm": 0.1877623349428177,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 2304
+    },
+    {
+      "epoch": 3.125,
+      "grad_norm": 0.16834235191345215,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2305
+    },
+    {
+      "epoch": 3.126355748373102,
+      "grad_norm": 0.15660707652568817,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 2306
+    },
+    {
+      "epoch": 3.127711496746204,
+      "grad_norm": 0.15572404861450195,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2307
+    },
+    {
+      "epoch": 3.129067245119306,
+      "grad_norm": 0.16924771666526794,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2308
+    },
+    {
+      "epoch": 3.1304229934924077,
+      "grad_norm": 0.1799774318933487,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 2309
+    },
+    {
+      "epoch": 3.1317787418655096,
+      "grad_norm": 0.1967582255601883,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2310
+    },
+    {
+      "epoch": 3.1331344902386116,
+      "grad_norm": 0.1828840970993042,
+      "learning_rate": 0.0001,
+      "loss": 0.1539,
+      "step": 2311
+    },
+    {
+      "epoch": 3.1344902386117135,
+      "grad_norm": 0.16360065340995789,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 2312
+    },
+    {
+      "epoch": 3.1358459869848154,
+      "grad_norm": 0.18070866167545319,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2313
+    },
+    {
+      "epoch": 3.137201735357918,
+      "grad_norm": 0.1830625832080841,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2314
+    },
+    {
+      "epoch": 3.1385574837310197,
+      "grad_norm": 0.18410149216651917,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 2315
+    },
+    {
+      "epoch": 3.1399132321041217,
+      "grad_norm": 0.17564988136291504,
+      "learning_rate": 0.0001,
+      "loss": 0.1754,
+      "step": 2316
+    },
+    {
+      "epoch": 3.1412689804772236,
+      "grad_norm": 0.1398431360721588,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2317
+    },
+    {
+      "epoch": 3.1426247288503255,
+      "grad_norm": 0.14292146265506744,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 2318
+    },
+    {
+      "epoch": 3.1439804772234274,
+      "grad_norm": 0.15934522449970245,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 2319
+    },
+    {
+      "epoch": 3.1453362255965294,
+      "grad_norm": 0.1471049189567566,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 2320
+    },
+    {
+      "epoch": 3.1466919739696313,
+      "grad_norm": 0.14916302263736725,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2321
+    },
+    {
+      "epoch": 3.1480477223427332,
+      "grad_norm": 0.13538584113121033,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2322
+    },
+    {
+      "epoch": 3.149403470715835,
+      "grad_norm": 0.15153051912784576,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2323
+    },
+    {
+      "epoch": 3.150759219088937,
+      "grad_norm": 0.1522514373064041,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2324
+    },
+    {
+      "epoch": 3.152114967462039,
+      "grad_norm": 0.13817016780376434,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2325
+    },
+    {
+      "epoch": 3.153470715835141,
+      "grad_norm": 0.12548638880252838,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 2326
+    },
+    {
+      "epoch": 3.154826464208243,
+      "grad_norm": 0.15126220881938934,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2327
+    },
+    {
+      "epoch": 3.156182212581345,
+      "grad_norm": 0.19650490581989288,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2328
+    },
+    {
+      "epoch": 3.1575379609544467,
+      "grad_norm": 0.17447184026241302,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2329
+    },
+    {
+      "epoch": 3.1588937093275486,
+      "grad_norm": 0.15806710720062256,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2330
+    },
+    {
+      "epoch": 3.1602494577006506,
+      "grad_norm": 0.17301368713378906,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 2331
+    },
+    {
+      "epoch": 3.1616052060737525,
+      "grad_norm": 0.15702709555625916,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 2332
+    },
+    {
+      "epoch": 3.162960954446855,
+      "grad_norm": 0.17072759568691254,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2333
+    },
+    {
+      "epoch": 3.164316702819957,
+      "grad_norm": 0.15865343809127808,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 2334
+    },
+    {
+      "epoch": 3.1656724511930587,
+      "grad_norm": 0.18712405860424042,
+      "learning_rate": 0.0001,
+      "loss": 0.1727,
+      "step": 2335
+    },
+    {
+      "epoch": 3.1670281995661607,
+      "grad_norm": 0.18091611564159393,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2336
+    },
+    {
+      "epoch": 3.1683839479392626,
+      "grad_norm": 0.15953943133354187,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 2337
+    },
+    {
+      "epoch": 3.1697396963123645,
+      "grad_norm": 0.14962907135486603,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 2338
+    },
+    {
+      "epoch": 3.1710954446854664,
+      "grad_norm": 0.154321551322937,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2339
+    },
+    {
+      "epoch": 3.1724511930585684,
+      "grad_norm": 0.17167094349861145,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2340
+    },
+    {
+      "epoch": 3.1738069414316703,
+      "grad_norm": 0.21884886920452118,
+      "learning_rate": 0.0001,
+      "loss": 0.1804,
+      "step": 2341
+    },
+    {
+      "epoch": 3.1751626898047722,
+      "grad_norm": 0.14249640703201294,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 2342
+    },
+    {
+      "epoch": 3.176518438177874,
+      "grad_norm": 0.15161462128162384,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 2343
+    },
+    {
+      "epoch": 3.177874186550976,
+      "grad_norm": 0.13675270974636078,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2344
+    },
+    {
+      "epoch": 3.179229934924078,
+      "grad_norm": 0.1574469357728958,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 2345
+    },
+    {
+      "epoch": 3.18058568329718,
+      "grad_norm": 0.16912750899791718,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 2346
+    },
+    {
+      "epoch": 3.181941431670282,
+      "grad_norm": 0.15086965262889862,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2347
+    },
+    {
+      "epoch": 3.183297180043384,
+      "grad_norm": 0.15471553802490234,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2348
+    },
+    {
+      "epoch": 3.1846529284164857,
+      "grad_norm": 0.25240352749824524,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 2349
+    },
+    {
+      "epoch": 3.1860086767895877,
+      "grad_norm": 0.1704157590866089,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 2350
+    },
+    {
+      "epoch": 3.1873644251626896,
+      "grad_norm": 0.1510051190853119,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2351
+    },
+    {
+      "epoch": 3.188720173535792,
+      "grad_norm": 0.15401935577392578,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 2352
+    },
+    {
+      "epoch": 3.190075921908894,
+      "grad_norm": 0.15235289931297302,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 2353
+    },
+    {
+      "epoch": 3.191431670281996,
+      "grad_norm": 0.14300242066383362,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2354
+    },
+    {
+      "epoch": 3.1927874186550977,
+      "grad_norm": 0.1607629954814911,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2355
+    },
+    {
+      "epoch": 3.1941431670281997,
+      "grad_norm": 0.1579870581626892,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 2356
+    },
+    {
+      "epoch": 3.1954989154013016,
+      "grad_norm": 0.16326583921909332,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2357
+    },
+    {
+      "epoch": 3.1968546637744035,
+      "grad_norm": 0.16185395419597626,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 2358
+    },
+    {
+      "epoch": 3.1982104121475055,
+      "grad_norm": 0.15090146660804749,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2359
+    },
+    {
+      "epoch": 3.1995661605206074,
+      "grad_norm": 0.409793496131897,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2360
+    },
+    {
+      "epoch": 3.2009219088937093,
+      "grad_norm": 0.14878787100315094,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 2361
+    },
+    {
+      "epoch": 3.2022776572668112,
+      "grad_norm": 0.14488345384597778,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2362
+    },
+    {
+      "epoch": 3.203633405639913,
+      "grad_norm": 0.1631544679403305,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2363
+    },
+    {
+      "epoch": 3.204989154013015,
+      "grad_norm": 0.15509828925132751,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2364
+    },
+    {
+      "epoch": 3.206344902386117,
+      "grad_norm": 0.18135444819927216,
+      "learning_rate": 0.0001,
+      "loss": 0.1777,
+      "step": 2365
+    },
+    {
+      "epoch": 3.207700650759219,
+      "grad_norm": 0.15392181277275085,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2366
+    },
+    {
+      "epoch": 3.209056399132321,
+      "grad_norm": 0.1661810576915741,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 2367
+    },
+    {
+      "epoch": 3.210412147505423,
+      "grad_norm": 0.13853557407855988,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 2368
+    },
+    {
+      "epoch": 3.211767895878525,
+      "grad_norm": 0.16404078900814056,
+      "learning_rate": 0.0001,
+      "loss": 0.1702,
+      "step": 2369
+    },
+    {
+      "epoch": 3.2131236442516267,
+      "grad_norm": 0.14383164048194885,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 2370
+    },
+    {
+      "epoch": 3.214479392624729,
+      "grad_norm": 0.1615705043077469,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 2371
+    },
+    {
+      "epoch": 3.215835140997831,
+      "grad_norm": 0.15717385709285736,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2372
+    },
+    {
+      "epoch": 3.217190889370933,
+      "grad_norm": 0.1619398295879364,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2373
+    },
+    {
+      "epoch": 3.218546637744035,
+      "grad_norm": 0.17338089644908905,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2374
+    },
+    {
+      "epoch": 3.2199023861171367,
+      "grad_norm": 0.1745430827140808,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2375
+    },
+    {
+      "epoch": 3.2212581344902387,
+      "grad_norm": 0.15895505249500275,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2376
+    },
+    {
+      "epoch": 3.2226138828633406,
+      "grad_norm": 0.16125641763210297,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 2377
+    },
+    {
+      "epoch": 3.2239696312364425,
+      "grad_norm": 0.15767817199230194,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2378
+    },
+    {
+      "epoch": 3.2253253796095445,
+      "grad_norm": 0.14919808506965637,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 2379
+    },
+    {
+      "epoch": 3.2266811279826464,
+      "grad_norm": 0.1733301430940628,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 2380
+    },
+    {
+      "epoch": 3.2280368763557483,
+      "grad_norm": 0.14477920532226562,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 2381
+    },
+    {
+      "epoch": 3.2293926247288502,
+      "grad_norm": 0.14425316452980042,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 2382
+    },
+    {
+      "epoch": 3.230748373101952,
+      "grad_norm": 0.17209988832473755,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 2383
+    },
+    {
+      "epoch": 3.232104121475054,
+      "grad_norm": 0.13570253551006317,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2384
+    },
+    {
+      "epoch": 3.233459869848156,
+      "grad_norm": 0.14842700958251953,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2385
+    },
+    {
+      "epoch": 3.234815618221258,
+      "grad_norm": 0.15187129378318787,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2386
+    },
+    {
+      "epoch": 3.23617136659436,
+      "grad_norm": 0.15740656852722168,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2387
+    },
+    {
+      "epoch": 3.2375271149674623,
+      "grad_norm": 0.1541413962841034,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2388
+    },
+    {
+      "epoch": 3.238882863340564,
+      "grad_norm": 0.13286563754081726,
+      "learning_rate": 0.0001,
+      "loss": 0.1271,
+      "step": 2389
+    },
+    {
+      "epoch": 3.240238611713666,
+      "grad_norm": 0.17290298640727997,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 2390
+    },
+    {
+      "epoch": 3.241594360086768,
+      "grad_norm": 0.1403786838054657,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 2391
+    },
+    {
+      "epoch": 3.24295010845987,
+      "grad_norm": 0.2124680131673813,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2392
+    },
+    {
+      "epoch": 3.244305856832972,
+      "grad_norm": 0.25377726554870605,
+      "learning_rate": 0.0001,
+      "loss": 0.1666,
+      "step": 2393
+    },
+    {
+      "epoch": 3.245661605206074,
+      "grad_norm": 0.14522013068199158,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 2394
+    },
+    {
+      "epoch": 3.2470173535791758,
+      "grad_norm": 0.22826969623565674,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 2395
+    },
+    {
+      "epoch": 3.2483731019522777,
+      "grad_norm": 0.14808537065982819,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 2396
+    },
+    {
+      "epoch": 3.2497288503253796,
+      "grad_norm": 0.46450135111808777,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2397
+    },
+    {
+      "epoch": 3.2510845986984815,
+      "grad_norm": 0.17384158074855804,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2398
+    },
+    {
+      "epoch": 3.2524403470715835,
+      "grad_norm": 0.17674694955348969,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2399
+    },
+    {
+      "epoch": 3.2537960954446854,
+      "grad_norm": 0.18835771083831787,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2400
+    },
+    {
+      "epoch": 3.2551518438177873,
+      "grad_norm": 0.15509071946144104,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2401
+    },
+    {
+      "epoch": 3.2565075921908893,
+      "grad_norm": 0.1910165250301361,
+      "learning_rate": 0.0001,
+      "loss": 0.1723,
+      "step": 2402
+    },
+    {
+      "epoch": 3.257863340563991,
+      "grad_norm": 0.15176598727703094,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 2403
+    },
+    {
+      "epoch": 3.259219088937093,
+      "grad_norm": 0.15014053881168365,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 2404
+    },
+    {
+      "epoch": 3.260574837310195,
+      "grad_norm": 0.1603916436433792,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2405
+    },
+    {
+      "epoch": 3.261930585683297,
+      "grad_norm": 0.1491696834564209,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2406
+    },
+    {
+      "epoch": 3.2632863340563993,
+      "grad_norm": 0.18440701067447662,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2407
+    },
+    {
+      "epoch": 3.2646420824295013,
+      "grad_norm": 0.18081986904144287,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2408
+    },
+    {
+      "epoch": 3.265997830802603,
+      "grad_norm": 0.15875808894634247,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 2409
+    },
+    {
+      "epoch": 3.267353579175705,
+      "grad_norm": 0.17781615257263184,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2410
+    },
+    {
+      "epoch": 3.268709327548807,
+      "grad_norm": 0.18011820316314697,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 2411
+    },
+    {
+      "epoch": 3.270065075921909,
+      "grad_norm": 0.1495649814605713,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 2412
+    },
+    {
+      "epoch": 3.271420824295011,
+      "grad_norm": 0.20535027980804443,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2413
+    },
+    {
+      "epoch": 3.272776572668113,
+      "grad_norm": 0.15221257507801056,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 2414
+    },
+    {
+      "epoch": 3.2741323210412148,
+      "grad_norm": 0.20074611902236938,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2415
+    },
+    {
+      "epoch": 3.2754880694143167,
+      "grad_norm": 0.1791308969259262,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 2416
+    },
+    {
+      "epoch": 3.2768438177874186,
+      "grad_norm": 0.1700924187898636,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2417
+    },
+    {
+      "epoch": 3.2781995661605206,
+      "grad_norm": 0.1564846634864807,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2418
+    },
+    {
+      "epoch": 3.2795553145336225,
+      "grad_norm": 0.15070368349552155,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 2419
+    },
+    {
+      "epoch": 3.2809110629067244,
+      "grad_norm": 0.13476112484931946,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 2420
+    },
+    {
+      "epoch": 3.2822668112798263,
+      "grad_norm": 0.19369345903396606,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2421
+    },
+    {
+      "epoch": 3.2836225596529283,
+      "grad_norm": 0.1484072059392929,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 2422
+    },
+    {
+      "epoch": 3.28497830802603,
+      "grad_norm": 0.13568224012851715,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 2423
+    },
+    {
+      "epoch": 3.2863340563991326,
+      "grad_norm": 0.1549876183271408,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 2424
+    },
+    {
+      "epoch": 3.287689804772234,
+      "grad_norm": 0.19247488677501678,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2425
+    },
+    {
+      "epoch": 3.2890455531453364,
+      "grad_norm": 0.17537818849086761,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 2426
+    },
+    {
+      "epoch": 3.2904013015184383,
+      "grad_norm": 0.16504104435443878,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2427
+    },
+    {
+      "epoch": 3.2917570498915403,
+      "grad_norm": 0.15638776123523712,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2428
+    },
+    {
+      "epoch": 3.293112798264642,
+      "grad_norm": 0.20485970377922058,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 2429
+    },
+    {
+      "epoch": 3.294468546637744,
+      "grad_norm": 0.15873217582702637,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2430
+    },
+    {
+      "epoch": 3.295824295010846,
+      "grad_norm": 0.1497223973274231,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2431
+    },
+    {
+      "epoch": 3.297180043383948,
+      "grad_norm": 0.14203524589538574,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2432
+    },
+    {
+      "epoch": 3.29853579175705,
+      "grad_norm": 0.17679710686206818,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2433
+    },
+    {
+      "epoch": 3.299891540130152,
+      "grad_norm": 0.14183694124221802,
+      "learning_rate": 0.0001,
+      "loss": 0.1511,
+      "step": 2434
+    },
+    {
+      "epoch": 3.3012472885032538,
+      "grad_norm": 0.18190963566303253,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 2435
+    },
+    {
+      "epoch": 3.3026030368763557,
+      "grad_norm": 0.17313319444656372,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 2436
+    },
+    {
+      "epoch": 3.3039587852494576,
+      "grad_norm": 0.16535137593746185,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2437
+    },
+    {
+      "epoch": 3.3053145336225596,
+      "grad_norm": 0.16098295152187347,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 2438
+    },
+    {
+      "epoch": 3.3066702819956615,
+      "grad_norm": 0.1638297140598297,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2439
+    },
+    {
+      "epoch": 3.3080260303687634,
+      "grad_norm": 0.15255194902420044,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 2440
+    },
+    {
+      "epoch": 3.3093817787418653,
+      "grad_norm": 0.18055832386016846,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 2441
+    },
+    {
+      "epoch": 3.3107375271149673,
+      "grad_norm": 0.16359764337539673,
+      "learning_rate": 0.0001,
+      "loss": 0.1539,
+      "step": 2442
+    },
+    {
+      "epoch": 3.3120932754880696,
+      "grad_norm": 0.1597631275653839,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2443
+    },
+    {
+      "epoch": 3.313449023861171,
+      "grad_norm": 0.15246689319610596,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 2444
+    },
+    {
+      "epoch": 3.3148047722342735,
+      "grad_norm": 0.14146846532821655,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2445
+    },
+    {
+      "epoch": 3.3161605206073754,
+      "grad_norm": 0.19698724150657654,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2446
+    },
+    {
+      "epoch": 3.3175162689804774,
+      "grad_norm": 0.16221840679645538,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2447
+    },
+    {
+      "epoch": 3.3188720173535793,
+      "grad_norm": 0.21154597401618958,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 2448
+    },
+    {
+      "epoch": 3.320227765726681,
+      "grad_norm": 0.19013451039791107,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2449
+    },
+    {
+      "epoch": 3.321583514099783,
+      "grad_norm": 0.4075399935245514,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2450
+    },
+    {
+      "epoch": 3.322939262472885,
+      "grad_norm": 0.1591944396495819,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 2451
+    },
+    {
+      "epoch": 3.324295010845987,
+      "grad_norm": 0.1645313799381256,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 2452
+    },
+    {
+      "epoch": 3.325650759219089,
+      "grad_norm": 0.1796121746301651,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2453
+    },
+    {
+      "epoch": 3.327006507592191,
+      "grad_norm": 0.1702168583869934,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2454
+    },
+    {
+      "epoch": 3.328362255965293,
+      "grad_norm": 0.15014436841011047,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2455
+    },
+    {
+      "epoch": 3.3297180043383947,
+      "grad_norm": 0.15040822327136993,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2456
+    },
+    {
+      "epoch": 3.3310737527114966,
+      "grad_norm": 0.13505277037620544,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 2457
+    },
+    {
+      "epoch": 3.3324295010845986,
+      "grad_norm": 0.13957437872886658,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 2458
+    },
+    {
+      "epoch": 3.3337852494577005,
+      "grad_norm": 0.190876767039299,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 2459
+    },
+    {
+      "epoch": 3.3351409978308024,
+      "grad_norm": 0.15771274268627167,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2460
+    },
+    {
+      "epoch": 3.3364967462039044,
+      "grad_norm": 0.15260708332061768,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 2461
+    },
+    {
+      "epoch": 3.3378524945770067,
+      "grad_norm": 0.20701749622821808,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 2462
+    },
+    {
+      "epoch": 3.3392082429501087,
+      "grad_norm": 0.1716563105583191,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 2463
+    },
+    {
+      "epoch": 3.3405639913232106,
+      "grad_norm": 0.13960693776607513,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2464
+    },
+    {
+      "epoch": 3.3419197396963125,
+      "grad_norm": 0.15983158349990845,
+      "learning_rate": 0.0001,
+      "loss": 0.1702,
+      "step": 2465
+    },
+    {
+      "epoch": 3.3432754880694144,
+      "grad_norm": 0.1560283601284027,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2466
+    },
+    {
+      "epoch": 3.3446312364425164,
+      "grad_norm": 0.1569933444261551,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 2467
+    },
+    {
+      "epoch": 3.3459869848156183,
+      "grad_norm": 0.1704384982585907,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 2468
+    },
+    {
+      "epoch": 3.34734273318872,
+      "grad_norm": 0.1795862764120102,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 2469
+    },
+    {
+      "epoch": 3.348698481561822,
+      "grad_norm": 0.1574523150920868,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 2470
+    },
+    {
+      "epoch": 3.350054229934924,
+      "grad_norm": 0.16271772980690002,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 2471
+    },
+    {
+      "epoch": 3.351409978308026,
+      "grad_norm": 0.13909746706485748,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 2472
+    },
+    {
+      "epoch": 3.352765726681128,
+      "grad_norm": 0.14185744524002075,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2473
+    },
+    {
+      "epoch": 3.35412147505423,
+      "grad_norm": 0.13352836668491364,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 2474
+    },
+    {
+      "epoch": 3.355477223427332,
+      "grad_norm": 0.14704184234142303,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2475
+    },
+    {
+      "epoch": 3.3568329718004337,
+      "grad_norm": 0.25432902574539185,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 2476
+    },
+    {
+      "epoch": 3.3581887201735356,
+      "grad_norm": 0.19570669531822205,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2477
+    },
+    {
+      "epoch": 3.3595444685466376,
+      "grad_norm": 0.13247574865818024,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 2478
+    },
+    {
+      "epoch": 3.3609002169197395,
+      "grad_norm": 0.14239732921123505,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2479
+    },
+    {
+      "epoch": 3.3622559652928414,
+      "grad_norm": 0.15854567289352417,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 2480
+    },
+    {
+      "epoch": 3.363611713665944,
+      "grad_norm": 0.13693656027317047,
+      "learning_rate": 0.0001,
+      "loss": 0.1422,
+      "step": 2481
+    },
+    {
+      "epoch": 3.3649674620390457,
+      "grad_norm": 0.159483402967453,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 2482
+    },
+    {
+      "epoch": 3.3663232104121477,
+      "grad_norm": 0.1692899912595749,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 2483
+    },
+    {
+      "epoch": 3.3676789587852496,
+      "grad_norm": 0.25198855996131897,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 2484
+    },
+    {
+      "epoch": 3.3690347071583515,
+      "grad_norm": 0.1502055823802948,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 2485
+    },
+    {
+      "epoch": 3.3703904555314534,
+      "grad_norm": 0.18828855454921722,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 2486
+    },
+    {
+      "epoch": 3.3717462039045554,
+      "grad_norm": 0.13937613368034363,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 2487
+    },
+    {
+      "epoch": 3.3731019522776573,
+      "grad_norm": 0.16886910796165466,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2488
+    },
+    {
+      "epoch": 3.3744577006507592,
+      "grad_norm": 0.17923012375831604,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2489
+    },
+    {
+      "epoch": 3.375813449023861,
+      "grad_norm": 0.1627826988697052,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2490
+    },
+    {
+      "epoch": 3.377169197396963,
+      "grad_norm": 0.19154857099056244,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 2491
+    },
+    {
+      "epoch": 3.378524945770065,
+      "grad_norm": 0.16060292720794678,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 2492
+    },
+    {
+      "epoch": 3.379880694143167,
+      "grad_norm": 0.1754320114850998,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2493
+    },
+    {
+      "epoch": 3.381236442516269,
+      "grad_norm": 0.1911039650440216,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 2494
+    },
+    {
+      "epoch": 3.382592190889371,
+      "grad_norm": 0.13971376419067383,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2495
+    },
+    {
+      "epoch": 3.3839479392624727,
+      "grad_norm": 0.15873578190803528,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2496
+    },
+    {
+      "epoch": 3.3853036876355747,
+      "grad_norm": 0.16023893654346466,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 2497
+    },
+    {
+      "epoch": 3.386659436008677,
+      "grad_norm": 0.13998129963874817,
+      "learning_rate": 0.0001,
+      "loss": 0.1415,
+      "step": 2498
+    },
+    {
+      "epoch": 3.3880151843817785,
+      "grad_norm": 0.15550246834754944,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 2499
+    },
+    {
+      "epoch": 3.389370932754881,
+      "grad_norm": 0.14183057844638824,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 2500
+    },
+    {
+      "epoch": 3.389370932754881,
+      "eval_loss": 0.19053812325000763,
+      "eval_runtime": 596.7279,
+      "eval_samples_per_second": 4.635,
+      "eval_steps_per_second": 1.16,
+      "step": 2500
+    },
+    {
+      "epoch": 3.390726681127983,
+      "grad_norm": 0.16997988522052765,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2501
+    },
+    {
+      "epoch": 3.3920824295010847,
+      "grad_norm": 0.15445274114608765,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 2502
+    },
+    {
+      "epoch": 3.3934381778741867,
+      "grad_norm": 0.15618814527988434,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2503
+    },
+    {
+      "epoch": 3.3947939262472886,
+      "grad_norm": 0.1604243516921997,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 2504
+    },
+    {
+      "epoch": 3.3961496746203905,
+      "grad_norm": 0.17166003584861755,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2505
+    },
+    {
+      "epoch": 3.3975054229934925,
+      "grad_norm": 0.16772575676441193,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 2506
+    },
+    {
+      "epoch": 3.3988611713665944,
+      "grad_norm": 0.18077148497104645,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2507
+    },
+    {
+      "epoch": 3.4002169197396963,
+      "grad_norm": 0.23562654852867126,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2508
+    },
+    {
+      "epoch": 3.4015726681127982,
+      "grad_norm": 0.1569528728723526,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 2509
+    },
+    {
+      "epoch": 3.4029284164859,
+      "grad_norm": 0.15726135671138763,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 2510
+    },
+    {
+      "epoch": 3.404284164859002,
+      "grad_norm": 0.15072834491729736,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2511
+    },
+    {
+      "epoch": 3.405639913232104,
+      "grad_norm": 0.14606283605098724,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2512
+    },
+    {
+      "epoch": 3.406995661605206,
+      "grad_norm": 0.1499405801296234,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2513
+    },
+    {
+      "epoch": 3.408351409978308,
+      "grad_norm": 0.15820340812206268,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2514
+    },
+    {
+      "epoch": 3.40970715835141,
+      "grad_norm": 0.1538952738046646,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 2515
+    },
+    {
+      "epoch": 3.4110629067245117,
+      "grad_norm": 0.16778220236301422,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 2516
+    },
+    {
+      "epoch": 3.412418655097614,
+      "grad_norm": 0.14528389275074005,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 2517
+    },
+    {
+      "epoch": 3.4137744034707156,
+      "grad_norm": 0.14293047785758972,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 2518
+    },
+    {
+      "epoch": 3.415130151843818,
+      "grad_norm": 0.13009153306484222,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 2519
+    },
+    {
+      "epoch": 3.41648590021692,
+      "grad_norm": 0.14934027194976807,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2520
+    },
+    {
+      "epoch": 3.417841648590022,
+      "grad_norm": 0.1441921591758728,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 2521
+    },
+    {
+      "epoch": 3.4191973969631237,
+      "grad_norm": 0.17225897312164307,
+      "learning_rate": 0.0001,
+      "loss": 0.1794,
+      "step": 2522
+    },
+    {
+      "epoch": 3.4205531453362257,
+      "grad_norm": 0.20791494846343994,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2523
+    },
+    {
+      "epoch": 3.4219088937093276,
+      "grad_norm": 0.15273058414459229,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 2524
+    },
+    {
+      "epoch": 3.4232646420824295,
+      "grad_norm": 0.14015185832977295,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2525
+    },
+    {
+      "epoch": 3.4246203904555315,
+      "grad_norm": 0.14834067225456238,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 2526
+    },
+    {
+      "epoch": 3.4259761388286334,
+      "grad_norm": 0.16980774700641632,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2527
+    },
+    {
+      "epoch": 3.4273318872017353,
+      "grad_norm": 0.13769009709358215,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2528
+    },
+    {
+      "epoch": 3.4286876355748372,
+      "grad_norm": 0.15673941373825073,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2529
+    },
+    {
+      "epoch": 3.430043383947939,
+      "grad_norm": 0.13200503587722778,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 2530
+    },
+    {
+      "epoch": 3.431399132321041,
+      "grad_norm": 0.1564064770936966,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2531
+    },
+    {
+      "epoch": 3.432754880694143,
+      "grad_norm": 0.1718449592590332,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 2532
+    },
+    {
+      "epoch": 3.434110629067245,
+      "grad_norm": 0.16252994537353516,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2533
+    },
+    {
+      "epoch": 3.435466377440347,
+      "grad_norm": 0.145872101187706,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2534
+    },
+    {
+      "epoch": 3.436822125813449,
+      "grad_norm": 0.16268296539783478,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2535
+    },
+    {
+      "epoch": 3.438177874186551,
+      "grad_norm": 0.15465375781059265,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2536
+    },
+    {
+      "epoch": 3.439533622559653,
+      "grad_norm": 0.17358919978141785,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2537
+    },
+    {
+      "epoch": 3.440889370932755,
+      "grad_norm": 0.1394934058189392,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 2538
+    },
+    {
+      "epoch": 3.442245119305857,
+      "grad_norm": 0.14238563179969788,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2539
+    },
+    {
+      "epoch": 3.443600867678959,
+      "grad_norm": 0.16594667732715607,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 2540
+    },
+    {
+      "epoch": 3.444956616052061,
+      "grad_norm": 0.15875719487667084,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 2541
+    },
+    {
+      "epoch": 3.4463123644251628,
+      "grad_norm": 0.14355012774467468,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2542
+    },
+    {
+      "epoch": 3.4476681127982647,
+      "grad_norm": 0.16719385981559753,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2543
+    },
+    {
+      "epoch": 3.4490238611713666,
+      "grad_norm": 0.1574428379535675,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 2544
+    },
+    {
+      "epoch": 3.4503796095444685,
+      "grad_norm": 0.16792340576648712,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 2545
+    },
+    {
+      "epoch": 3.4517353579175705,
+      "grad_norm": 0.1877073049545288,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 2546
+    },
+    {
+      "epoch": 3.4530911062906724,
+      "grad_norm": 0.153830423951149,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 2547
+    },
+    {
+      "epoch": 3.4544468546637743,
+      "grad_norm": 0.1471448391675949,
+      "learning_rate": 0.0001,
+      "loss": 0.1556,
+      "step": 2548
+    },
+    {
+      "epoch": 3.4558026030368763,
+      "grad_norm": 0.1630573868751526,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 2549
+    },
+    {
+      "epoch": 3.457158351409978,
+      "grad_norm": 0.14236333966255188,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2550
+    },
+    {
+      "epoch": 3.45851409978308,
+      "grad_norm": 0.15863585472106934,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2551
+    },
+    {
+      "epoch": 3.459869848156182,
+      "grad_norm": 0.18011000752449036,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 2552
+    },
+    {
+      "epoch": 3.4612255965292844,
+      "grad_norm": 0.15175767242908478,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 2553
+    },
+    {
+      "epoch": 3.462581344902386,
+      "grad_norm": 0.13171948492527008,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 2554
+    },
+    {
+      "epoch": 3.4639370932754883,
+      "grad_norm": 0.19765448570251465,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 2555
+    },
+    {
+      "epoch": 3.46529284164859,
+      "grad_norm": 0.17993134260177612,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 2556
+    },
+    {
+      "epoch": 3.466648590021692,
+      "grad_norm": 0.15176431834697723,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2557
+    },
+    {
+      "epoch": 3.468004338394794,
+      "grad_norm": 0.14828142523765564,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2558
+    },
+    {
+      "epoch": 3.469360086767896,
+      "grad_norm": 0.12828227877616882,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 2559
+    },
+    {
+      "epoch": 3.470715835140998,
+      "grad_norm": 0.14546357095241547,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2560
+    },
+    {
+      "epoch": 3.4720715835141,
+      "grad_norm": 0.15510666370391846,
+      "learning_rate": 0.0001,
+      "loss": 0.1658,
+      "step": 2561
+    },
+    {
+      "epoch": 3.4734273318872018,
+      "grad_norm": 0.1496156007051468,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 2562
+    },
+    {
+      "epoch": 3.4747830802603037,
+      "grad_norm": 0.1395443081855774,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2563
+    },
+    {
+      "epoch": 3.4761388286334056,
+      "grad_norm": 0.16556783020496368,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 2564
+    },
+    {
+      "epoch": 3.4774945770065075,
+      "grad_norm": 0.16682906448841095,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2565
+    },
+    {
+      "epoch": 3.4788503253796095,
+      "grad_norm": 0.15924140810966492,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 2566
+    },
+    {
+      "epoch": 3.4802060737527114,
+      "grad_norm": 0.14105217158794403,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 2567
+    },
+    {
+      "epoch": 3.4815618221258133,
+      "grad_norm": 0.14939287304878235,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2568
+    },
+    {
+      "epoch": 3.4829175704989153,
+      "grad_norm": 0.1450740545988083,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 2569
+    },
+    {
+      "epoch": 3.484273318872017,
+      "grad_norm": 0.16792826354503632,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2570
+    },
+    {
+      "epoch": 3.485629067245119,
+      "grad_norm": 0.17464768886566162,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 2571
+    },
+    {
+      "epoch": 3.4869848156182215,
+      "grad_norm": 0.17932334542274475,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2572
+    },
+    {
+      "epoch": 3.488340563991323,
+      "grad_norm": 0.17748095095157623,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2573
+    },
+    {
+      "epoch": 3.4896963123644253,
+      "grad_norm": 0.1734902411699295,
+      "learning_rate": 0.0001,
+      "loss": 0.1745,
+      "step": 2574
+    },
+    {
+      "epoch": 3.4910520607375273,
+      "grad_norm": 0.18096892535686493,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 2575
+    },
+    {
+      "epoch": 3.492407809110629,
+      "grad_norm": 0.1502802073955536,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 2576
+    },
+    {
+      "epoch": 3.493763557483731,
+      "grad_norm": 0.17010998725891113,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 2577
+    },
+    {
+      "epoch": 3.495119305856833,
+      "grad_norm": 0.16089653968811035,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 2578
+    },
+    {
+      "epoch": 3.496475054229935,
+      "grad_norm": 0.16266506910324097,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2579
+    },
+    {
+      "epoch": 3.497830802603037,
+      "grad_norm": 0.1488586813211441,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2580
+    },
+    {
+      "epoch": 3.499186550976139,
+      "grad_norm": 0.15447132289409637,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 2581
+    },
+    {
+      "epoch": 3.5005422993492408,
+      "grad_norm": 0.13703694939613342,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 2582
+    },
+    {
+      "epoch": 3.5018980477223427,
+      "grad_norm": 0.1550346314907074,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 2583
+    },
+    {
+      "epoch": 3.5032537960954446,
+      "grad_norm": 0.15209990739822388,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2584
+    },
+    {
+      "epoch": 3.5046095444685466,
+      "grad_norm": 0.1457691490650177,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 2585
+    },
+    {
+      "epoch": 3.5059652928416485,
+      "grad_norm": 0.17237336933612823,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2586
+    },
+    {
+      "epoch": 3.5073210412147504,
+      "grad_norm": 0.1388378143310547,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2587
+    },
+    {
+      "epoch": 3.5086767895878523,
+      "grad_norm": 0.1496792733669281,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 2588
+    },
+    {
+      "epoch": 3.5100325379609547,
+      "grad_norm": 0.1541091948747635,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 2589
+    },
+    {
+      "epoch": 3.511388286334056,
+      "grad_norm": 0.15943634510040283,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2590
+    },
+    {
+      "epoch": 3.5127440347071586,
+      "grad_norm": 0.17224274575710297,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 2591
+    },
+    {
+      "epoch": 3.51409978308026,
+      "grad_norm": 0.15529750287532806,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2592
+    },
+    {
+      "epoch": 3.5154555314533624,
+      "grad_norm": 0.12557969987392426,
+      "learning_rate": 0.0001,
+      "loss": 0.1365,
+      "step": 2593
+    },
+    {
+      "epoch": 3.5168112798264644,
+      "grad_norm": 0.14643815159797668,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2594
+    },
+    {
+      "epoch": 3.5181670281995663,
+      "grad_norm": 0.15257121622562408,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 2595
+    },
+    {
+      "epoch": 3.519522776572668,
+      "grad_norm": 0.13983963429927826,
+      "learning_rate": 0.0001,
+      "loss": 0.154,
+      "step": 2596
+    },
+    {
+      "epoch": 3.52087852494577,
+      "grad_norm": 0.15229947865009308,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 2597
+    },
+    {
+      "epoch": 3.522234273318872,
+      "grad_norm": 0.1477479785680771,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 2598
+    },
+    {
+      "epoch": 3.523590021691974,
+      "grad_norm": 0.14167387783527374,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 2599
+    },
+    {
+      "epoch": 3.524945770065076,
+      "grad_norm": 0.14755629003047943,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 2600
+    },
+    {
+      "epoch": 3.526301518438178,
+      "grad_norm": 0.14123700559139252,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 2601
+    },
+    {
+      "epoch": 3.52765726681128,
+      "grad_norm": 0.14378969371318817,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 2602
+    },
+    {
+      "epoch": 3.5290130151843817,
+      "grad_norm": 0.1532048135995865,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2603
+    },
+    {
+      "epoch": 3.5303687635574836,
+      "grad_norm": 0.1400327980518341,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 2604
+    },
+    {
+      "epoch": 3.5317245119305856,
+      "grad_norm": 0.1576610803604126,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 2605
+    },
+    {
+      "epoch": 3.5330802603036875,
+      "grad_norm": 0.13320432603359222,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 2606
+    },
+    {
+      "epoch": 3.5344360086767894,
+      "grad_norm": 0.16769415140151978,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2607
+    },
+    {
+      "epoch": 3.535791757049892,
+      "grad_norm": 0.15157988667488098,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2608
+    },
+    {
+      "epoch": 3.5371475054229933,
+      "grad_norm": 0.16438204050064087,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 2609
+    },
+    {
+      "epoch": 3.5385032537960956,
+      "grad_norm": 0.15150579810142517,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 2610
+    },
+    {
+      "epoch": 3.539859002169197,
+      "grad_norm": 0.17023158073425293,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 2611
+    },
+    {
+      "epoch": 3.5412147505422995,
+      "grad_norm": 0.16784429550170898,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 2612
+    },
+    {
+      "epoch": 3.5425704989154014,
+      "grad_norm": 0.15418335795402527,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 2613
+    },
+    {
+      "epoch": 3.5439262472885034,
+      "grad_norm": 0.16598984599113464,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 2614
+    },
+    {
+      "epoch": 3.5452819956616053,
+      "grad_norm": 0.16992273926734924,
+      "learning_rate": 0.0001,
+      "loss": 0.1739,
+      "step": 2615
+    },
+    {
+      "epoch": 3.546637744034707,
+      "grad_norm": 0.14226000010967255,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 2616
+    },
+    {
+      "epoch": 3.547993492407809,
+      "grad_norm": 0.14399456977844238,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 2617
+    },
+    {
+      "epoch": 3.549349240780911,
+      "grad_norm": 0.1911216527223587,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2618
+    },
+    {
+      "epoch": 3.550704989154013,
+      "grad_norm": 0.17496278882026672,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 2619
+    },
+    {
+      "epoch": 3.552060737527115,
+      "grad_norm": 0.14863373339176178,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 2620
+    },
+    {
+      "epoch": 3.553416485900217,
+      "grad_norm": 0.14962869882583618,
+      "learning_rate": 0.0001,
+      "loss": 0.1725,
+      "step": 2621
+    },
+    {
+      "epoch": 3.554772234273319,
+      "grad_norm": 0.17029985785484314,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 2622
+    },
+    {
+      "epoch": 3.5561279826464207,
+      "grad_norm": 0.14021581411361694,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 2623
+    },
+    {
+      "epoch": 3.5574837310195226,
+      "grad_norm": 0.14801941812038422,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 2624
+    },
+    {
+      "epoch": 3.5588394793926246,
+      "grad_norm": 0.15773849189281464,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 2625
+    },
+    {
+      "epoch": 3.5601952277657265,
+      "grad_norm": 0.14386820793151855,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 2626
+    },
+    {
+      "epoch": 3.561550976138829,
+      "grad_norm": 0.15504400432109833,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2627
+    },
+    {
+      "epoch": 3.5629067245119304,
+      "grad_norm": 0.15898136794567108,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 2628
+    },
+    {
+      "epoch": 3.5642624728850327,
+      "grad_norm": 0.2847587466239929,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 2629
+    },
+    {
+      "epoch": 3.565618221258134,
+      "grad_norm": 0.15627999603748322,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2630
+    },
+    {
+      "epoch": 3.5669739696312366,
+      "grad_norm": 0.17560088634490967,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 2631
+    },
+    {
+      "epoch": 3.5683297180043385,
+      "grad_norm": 0.16823215782642365,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2632
+    },
+    {
+      "epoch": 3.5696854663774404,
+      "grad_norm": 0.1530900001525879,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2633
+    },
+    {
+      "epoch": 3.5710412147505424,
+      "grad_norm": 0.1596616953611374,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 2634
+    },
+    {
+      "epoch": 3.5723969631236443,
+      "grad_norm": 0.16543450951576233,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2635
+    },
+    {
+      "epoch": 3.5737527114967462,
+      "grad_norm": 0.1256786286830902,
+      "learning_rate": 0.0001,
+      "loss": 0.1388,
+      "step": 2636
+    },
+    {
+      "epoch": 3.575108459869848,
+      "grad_norm": 0.19122563302516937,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2637
+    },
+    {
+      "epoch": 3.57646420824295,
+      "grad_norm": 0.1845654994249344,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2638
+    },
+    {
+      "epoch": 3.577819956616052,
+      "grad_norm": 0.17347964644432068,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 2639
+    },
+    {
+      "epoch": 3.579175704989154,
+      "grad_norm": 0.16678568720817566,
+      "learning_rate": 0.0001,
+      "loss": 0.1697,
+      "step": 2640
+    },
+    {
+      "epoch": 3.580531453362256,
+      "grad_norm": 0.15969890356063843,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 2641
+    },
+    {
+      "epoch": 3.581887201735358,
+      "grad_norm": 0.16116023063659668,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 2642
+    },
+    {
+      "epoch": 3.5832429501084597,
+      "grad_norm": 0.13768270611763,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 2643
+    },
+    {
+      "epoch": 3.584598698481562,
+      "grad_norm": 0.13503140211105347,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2644
+    },
+    {
+      "epoch": 3.5859544468546636,
+      "grad_norm": 0.14652325212955475,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2645
+    },
+    {
+      "epoch": 3.587310195227766,
+      "grad_norm": 0.12742452323436737,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2646
+    },
+    {
+      "epoch": 3.5886659436008674,
+      "grad_norm": 0.13989649713039398,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 2647
+    },
+    {
+      "epoch": 3.59002169197397,
+      "grad_norm": 0.14834266901016235,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2648
+    },
+    {
+      "epoch": 3.5913774403470717,
+      "grad_norm": 0.1464710384607315,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2649
+    },
+    {
+      "epoch": 3.5927331887201737,
+      "grad_norm": 0.14421992003917694,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 2650
+    },
+    {
+      "epoch": 3.5940889370932756,
+      "grad_norm": 0.1573207527399063,
+      "learning_rate": 0.0001,
+      "loss": 0.1703,
+      "step": 2651
+    },
+    {
+      "epoch": 3.5954446854663775,
+      "grad_norm": 0.1377679854631424,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 2652
+    },
+    {
+      "epoch": 3.5968004338394794,
+      "grad_norm": 0.15891823172569275,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2653
+    },
+    {
+      "epoch": 3.5981561822125814,
+      "grad_norm": 0.1421186923980713,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 2654
+    },
+    {
+      "epoch": 3.5995119305856833,
+      "grad_norm": 0.1599660962820053,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 2655
+    },
+    {
+      "epoch": 3.6008676789587852,
+      "grad_norm": 0.14331567287445068,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 2656
+    },
+    {
+      "epoch": 3.602223427331887,
+      "grad_norm": 0.16436892747879028,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 2657
+    },
+    {
+      "epoch": 3.603579175704989,
+      "grad_norm": 0.18394041061401367,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 2658
+    },
+    {
+      "epoch": 3.604934924078091,
+      "grad_norm": 0.15727116167545319,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 2659
+    },
+    {
+      "epoch": 3.606290672451193,
+      "grad_norm": 0.1752326488494873,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 2660
+    },
+    {
+      "epoch": 3.607646420824295,
+      "grad_norm": 0.15931715071201324,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 2661
+    },
+    {
+      "epoch": 3.609002169197397,
+      "grad_norm": 0.1857602298259735,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 2662
+    },
+    {
+      "epoch": 3.610357917570499,
+      "grad_norm": 0.15587493777275085,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 2663
+    },
+    {
+      "epoch": 3.6117136659436007,
+      "grad_norm": 0.14881742000579834,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 2664
+    },
+    {
+      "epoch": 3.613069414316703,
+      "grad_norm": 0.1456703394651413,
+      "learning_rate": 0.0001,
+      "loss": 0.1392,
+      "step": 2665
+    },
+    {
+      "epoch": 3.6144251626898045,
+      "grad_norm": 0.1480480134487152,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2666
+    },
+    {
+      "epoch": 3.615780911062907,
+      "grad_norm": 0.36961251497268677,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 2667
+    },
+    {
+      "epoch": 3.617136659436009,
+      "grad_norm": 0.21676255762577057,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 2668
+    },
+    {
+      "epoch": 3.6184924078091107,
+      "grad_norm": 0.16135169565677643,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 2669
+    },
+    {
+      "epoch": 3.6198481561822127,
+      "grad_norm": 0.17254024744033813,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 2670
+    },
+    {
+      "epoch": 3.6212039045553146,
+      "grad_norm": 0.14174917340278625,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 2671
+    },
+    {
+      "epoch": 3.6225596529284165,
+      "grad_norm": 0.16573701798915863,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 2672
+    },
+    {
+      "epoch": 3.6239154013015185,
+      "grad_norm": 0.15386857092380524,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 2673
+    },
+    {
+      "epoch": 3.6252711496746204,
+      "grad_norm": 0.14717094600200653,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2674
+    },
+    {
+      "epoch": 3.6266268980477223,
+      "grad_norm": 0.16347670555114746,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 2675
+    },
+    {
+      "epoch": 3.6279826464208242,
+      "grad_norm": 0.15971679985523224,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 2676
+    },
+    {
+      "epoch": 3.629338394793926,
+      "grad_norm": 0.20520301163196564,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 2677
+    },
+    {
+      "epoch": 3.630694143167028,
+      "grad_norm": 0.1506335735321045,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2678
+    },
+    {
+      "epoch": 3.63204989154013,
+      "grad_norm": 0.23513415455818176,
+      "learning_rate": 0.0001,
+      "loss": 0.1637,
+      "step": 2679
+    },
+    {
+      "epoch": 3.633405639913232,
+      "grad_norm": 0.12930750846862793,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 2680
+    },
+    {
+      "epoch": 3.634761388286334,
+      "grad_norm": 0.1499703824520111,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2681
+    },
+    {
+      "epoch": 3.6361171366594363,
+      "grad_norm": 0.15872111916542053,
+      "learning_rate": 0.0001,
+      "loss": 0.1676,
+      "step": 2682
+    },
+    {
+      "epoch": 3.6374728850325377,
+      "grad_norm": 0.132711261510849,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2683
+    },
+    {
+      "epoch": 3.63882863340564,
+      "grad_norm": 0.14700354635715485,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 2684
+    },
+    {
+      "epoch": 3.6401843817787416,
+      "grad_norm": 0.13801933825016022,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2685
+    },
+    {
+      "epoch": 3.641540130151844,
+      "grad_norm": 0.17367765307426453,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2686
+    },
+    {
+      "epoch": 3.642895878524946,
+      "grad_norm": 0.16489700973033905,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2687
+    },
+    {
+      "epoch": 3.644251626898048,
+      "grad_norm": 0.15636318922042847,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 2688
+    },
+    {
+      "epoch": 3.6456073752711498,
+      "grad_norm": 0.16537873446941376,
+      "learning_rate": 0.0001,
+      "loss": 0.1711,
+      "step": 2689
+    },
+    {
+      "epoch": 3.6469631236442517,
+      "grad_norm": 0.1409836709499359,
+      "learning_rate": 0.0001,
+      "loss": 0.1373,
+      "step": 2690
+    },
+    {
+      "epoch": 3.6483188720173536,
+      "grad_norm": 0.16423116624355316,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 2691
+    },
+    {
+      "epoch": 3.6496746203904555,
+      "grad_norm": 0.16167095303535461,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 2692
+    },
+    {
+      "epoch": 3.6510303687635575,
+      "grad_norm": 0.15226684510707855,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2693
+    },
+    {
+      "epoch": 3.6523861171366594,
+      "grad_norm": 0.16376663744449615,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 2694
+    },
+    {
+      "epoch": 3.6537418655097613,
+      "grad_norm": 0.1734432727098465,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2695
+    },
+    {
+      "epoch": 3.6550976138828633,
+      "grad_norm": 0.15235179662704468,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 2696
+    },
+    {
+      "epoch": 3.656453362255965,
+      "grad_norm": 0.14821048080921173,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 2697
+    },
+    {
+      "epoch": 3.657809110629067,
+      "grad_norm": 0.1788949817419052,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 2698
+    },
+    {
+      "epoch": 3.659164859002169,
+      "grad_norm": 0.1847229301929474,
+      "learning_rate": 0.0001,
+      "loss": 0.1801,
+      "step": 2699
+    },
+    {
+      "epoch": 3.660520607375271,
+      "grad_norm": 0.17677125334739685,
+      "learning_rate": 0.0001,
+      "loss": 0.1773,
+      "step": 2700
+    },
+    {
+      "epoch": 3.6618763557483733,
+      "grad_norm": 0.1632291078567505,
+      "learning_rate": 0.0001,
+      "loss": 0.1634,
+      "step": 2701
+    },
+    {
+      "epoch": 3.663232104121475,
+      "grad_norm": 0.17263372242450714,
+      "learning_rate": 0.0001,
+      "loss": 0.1688,
+      "step": 2702
+    },
+    {
+      "epoch": 3.664587852494577,
+      "grad_norm": 0.16380535066127777,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 2703
+    },
+    {
+      "epoch": 3.665943600867679,
+      "grad_norm": 0.17518150806427002,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2704
+    },
+    {
+      "epoch": 3.667299349240781,
+      "grad_norm": 0.15553908050060272,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2705
+    },
+    {
+      "epoch": 3.668655097613883,
+      "grad_norm": 0.1682671755552292,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 2706
+    },
+    {
+      "epoch": 3.670010845986985,
+      "grad_norm": 0.15094614028930664,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2707
+    },
+    {
+      "epoch": 3.671366594360087,
+      "grad_norm": 0.19669011235237122,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2708
+    },
+    {
+      "epoch": 3.6727223427331888,
+      "grad_norm": 0.15402093529701233,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 2709
+    },
+    {
+      "epoch": 3.6740780911062907,
+      "grad_norm": 0.17541085183620453,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 2710
+    },
+    {
+      "epoch": 3.6754338394793926,
+      "grad_norm": 0.15710343420505524,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 2711
+    },
+    {
+      "epoch": 3.6767895878524945,
+      "grad_norm": 0.174346461892128,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 2712
+    },
+    {
+      "epoch": 3.6781453362255965,
+      "grad_norm": 0.14234352111816406,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 2713
+    },
+    {
+      "epoch": 3.6795010845986984,
+      "grad_norm": 0.14463748037815094,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 2714
+    },
+    {
+      "epoch": 3.6808568329718003,
+      "grad_norm": 0.14920274913311005,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2715
+    },
+    {
+      "epoch": 3.6822125813449023,
+      "grad_norm": 0.14372390508651733,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2716
+    },
+    {
+      "epoch": 3.683568329718004,
+      "grad_norm": 0.13746927678585052,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 2717
+    },
+    {
+      "epoch": 3.6849240780911066,
+      "grad_norm": 0.1479881852865219,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 2718
+    },
+    {
+      "epoch": 3.686279826464208,
+      "grad_norm": 0.14492109417915344,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2719
+    },
+    {
+      "epoch": 3.6876355748373104,
+      "grad_norm": 0.17420607805252075,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2720
+    },
+    {
+      "epoch": 3.688991323210412,
+      "grad_norm": 0.1540064811706543,
+      "learning_rate": 0.0001,
+      "loss": 0.1668,
+      "step": 2721
+    },
+    {
+      "epoch": 3.6903470715835143,
+      "grad_norm": 0.17386126518249512,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 2722
+    },
+    {
+      "epoch": 3.691702819956616,
+      "grad_norm": 0.16175483167171478,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 2723
+    },
+    {
+      "epoch": 3.693058568329718,
+      "grad_norm": 0.15551897883415222,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2724
+    },
+    {
+      "epoch": 3.69441431670282,
+      "grad_norm": 0.1443144679069519,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 2725
+    },
+    {
+      "epoch": 3.695770065075922,
+      "grad_norm": 0.13969336450099945,
+      "learning_rate": 0.0001,
+      "loss": 0.146,
+      "step": 2726
+    },
+    {
+      "epoch": 3.697125813449024,
+      "grad_norm": 0.15192480385303497,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 2727
+    },
+    {
+      "epoch": 3.698481561822126,
+      "grad_norm": 0.15215624868869781,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 2728
+    },
+    {
+      "epoch": 3.6998373101952278,
+      "grad_norm": 0.13872087001800537,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2729
+    },
+    {
+      "epoch": 3.7011930585683297,
+      "grad_norm": 0.652064323425293,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 2730
+    },
+    {
+      "epoch": 3.7025488069414316,
+      "grad_norm": 0.14527101814746857,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 2731
+    },
+    {
+      "epoch": 3.7039045553145336,
+      "grad_norm": 0.15561053156852722,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 2732
+    },
+    {
+      "epoch": 3.7052603036876355,
+      "grad_norm": 0.16167162358760834,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 2733
+    },
+    {
+      "epoch": 3.7066160520607374,
+      "grad_norm": 0.1698143035173416,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 2734
+    },
+    {
+      "epoch": 3.7079718004338393,
+      "grad_norm": 0.16764485836029053,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 2735
+    },
+    {
+      "epoch": 3.7093275488069413,
+      "grad_norm": 0.149230495095253,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2736
+    },
+    {
+      "epoch": 3.7106832971800436,
+      "grad_norm": 0.14793086051940918,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 2737
+    },
+    {
+      "epoch": 3.712039045553145,
+      "grad_norm": 0.17704516649246216,
+      "learning_rate": 0.0001,
+      "loss": 0.1719,
+      "step": 2738
+    },
+    {
+      "epoch": 3.7133947939262475,
+      "grad_norm": 0.1529775857925415,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2739
+    },
+    {
+      "epoch": 3.714750542299349,
+      "grad_norm": 0.16555173695087433,
+      "learning_rate": 0.0001,
+      "loss": 0.1797,
+      "step": 2740
+    },
+    {
+      "epoch": 3.7161062906724514,
+      "grad_norm": 0.1634128987789154,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2741
+    },
+    {
+      "epoch": 3.7174620390455533,
+      "grad_norm": 0.13754227757453918,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 2742
+    },
+    {
+      "epoch": 3.718817787418655,
+      "grad_norm": 0.15394237637519836,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2743
+    },
+    {
+      "epoch": 3.720173535791757,
+      "grad_norm": 0.13416503369808197,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 2744
+    },
+    {
+      "epoch": 3.721529284164859,
+      "grad_norm": 0.14506463706493378,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2745
+    },
+    {
+      "epoch": 3.722885032537961,
+      "grad_norm": 0.14290563762187958,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2746
+    },
+    {
+      "epoch": 3.724240780911063,
+      "grad_norm": 0.15483912825584412,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 2747
+    },
+    {
+      "epoch": 3.725596529284165,
+      "grad_norm": 0.14464138448238373,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 2748
+    },
+    {
+      "epoch": 3.7269522776572668,
+      "grad_norm": 0.17056520283222198,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 2749
+    },
+    {
+      "epoch": 3.7283080260303687,
+      "grad_norm": 0.18358400464057922,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 2750
+    },
+    {
+      "epoch": 3.7296637744034706,
+      "grad_norm": 0.144338920712471,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 2751
+    },
+    {
+      "epoch": 3.7310195227765726,
+      "grad_norm": 0.147384375333786,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 2752
+    },
+    {
+      "epoch": 3.7323752711496745,
+      "grad_norm": 0.20390057563781738,
+      "learning_rate": 0.0001,
+      "loss": 0.1729,
+      "step": 2753
+    },
+    {
+      "epoch": 3.7337310195227764,
+      "grad_norm": 0.15162310004234314,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2754
+    },
+    {
+      "epoch": 3.7350867678958783,
+      "grad_norm": 0.17859813570976257,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 2755
+    },
+    {
+      "epoch": 3.7364425162689807,
+      "grad_norm": 0.16477078199386597,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2756
+    },
+    {
+      "epoch": 3.737798264642082,
+      "grad_norm": 0.14724603295326233,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2757
+    },
+    {
+      "epoch": 3.7391540130151846,
+      "grad_norm": 0.15618321299552917,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2758
+    },
+    {
+      "epoch": 3.740509761388286,
+      "grad_norm": 0.19311751425266266,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2759
+    },
+    {
+      "epoch": 3.7418655097613884,
+      "grad_norm": 0.15470702946186066,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 2760
+    },
+    {
+      "epoch": 3.7432212581344904,
+      "grad_norm": 0.1534377932548523,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2761
+    },
+    {
+      "epoch": 3.7445770065075923,
+      "grad_norm": 0.1786029189825058,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 2762
+    },
+    {
+      "epoch": 3.745932754880694,
+      "grad_norm": 0.13603319227695465,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 2763
+    },
+    {
+      "epoch": 3.747288503253796,
+      "grad_norm": 0.14858290553092957,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2764
+    },
+    {
+      "epoch": 3.748644251626898,
+      "grad_norm": 0.14512623846530914,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2765
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 0.14246520400047302,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2766
+    },
+    {
+      "epoch": 3.751355748373102,
+      "grad_norm": 0.1363978534936905,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 2767
+    },
+    {
+      "epoch": 3.752711496746204,
+      "grad_norm": 0.1423819363117218,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 2768
+    },
+    {
+      "epoch": 3.754067245119306,
+      "grad_norm": 0.16204799711704254,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 2769
+    },
+    {
+      "epoch": 3.7554229934924077,
+      "grad_norm": 0.1760062575340271,
+      "learning_rate": 0.0001,
+      "loss": 0.1657,
+      "step": 2770
+    },
+    {
+      "epoch": 3.7567787418655096,
+      "grad_norm": 0.1922033280134201,
+      "learning_rate": 0.0001,
+      "loss": 0.1686,
+      "step": 2771
+    },
+    {
+      "epoch": 3.7581344902386116,
+      "grad_norm": 0.17930467426776886,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 2772
+    },
+    {
+      "epoch": 3.759490238611714,
+      "grad_norm": 0.15225033462047577,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 2773
+    },
+    {
+      "epoch": 3.7608459869848154,
+      "grad_norm": 0.1512169986963272,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 2774
+    },
+    {
+      "epoch": 3.762201735357918,
+      "grad_norm": 0.17175115644931793,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 2775
+    },
+    {
+      "epoch": 3.7635574837310193,
+      "grad_norm": 0.167411670088768,
+      "learning_rate": 0.0001,
+      "loss": 0.1677,
+      "step": 2776
+    },
+    {
+      "epoch": 3.7649132321041217,
+      "grad_norm": 0.1543254554271698,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 2777
+    },
+    {
+      "epoch": 3.7662689804772236,
+      "grad_norm": 0.1378801465034485,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2778
+    },
+    {
+      "epoch": 3.7676247288503255,
+      "grad_norm": 0.14081145823001862,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 2779
+    },
+    {
+      "epoch": 3.7689804772234274,
+      "grad_norm": 0.11823280900716782,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 2780
+    },
+    {
+      "epoch": 3.7703362255965294,
+      "grad_norm": 0.1597810834646225,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 2781
+    },
+    {
+      "epoch": 3.7716919739696313,
+      "grad_norm": 0.1381738930940628,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 2782
+    },
+    {
+      "epoch": 3.7730477223427332,
+      "grad_norm": 0.1568753719329834,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 2783
+    },
+    {
+      "epoch": 3.774403470715835,
+      "grad_norm": 0.17981283366680145,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2784
+    },
+    {
+      "epoch": 3.775759219088937,
+      "grad_norm": 0.17404842376708984,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 2785
+    },
+    {
+      "epoch": 3.777114967462039,
+      "grad_norm": 0.1349598467350006,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 2786
+    },
+    {
+      "epoch": 3.778470715835141,
+      "grad_norm": 0.1523754745721817,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 2787
+    },
+    {
+      "epoch": 3.779826464208243,
+      "grad_norm": 0.1363971084356308,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 2788
+    },
+    {
+      "epoch": 3.781182212581345,
+      "grad_norm": 0.1705772429704666,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 2789
+    },
+    {
+      "epoch": 3.7825379609544467,
+      "grad_norm": 0.16787302494049072,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 2790
+    },
+    {
+      "epoch": 3.7838937093275486,
+      "grad_norm": 0.14577355980873108,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 2791
+    },
+    {
+      "epoch": 3.785249457700651,
+      "grad_norm": 0.15121209621429443,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 2792
+    },
+    {
+      "epoch": 3.7866052060737525,
+      "grad_norm": 0.16669277846813202,
+      "learning_rate": 0.0001,
+      "loss": 0.1766,
+      "step": 2793
+    },
+    {
+      "epoch": 3.787960954446855,
+      "grad_norm": 0.14357241988182068,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 2794
+    },
+    {
+      "epoch": 3.7893167028199564,
+      "grad_norm": 0.16519425809383392,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2795
+    },
+    {
+      "epoch": 3.7906724511930587,
+      "grad_norm": 0.1451537162065506,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2796
+    },
+    {
+      "epoch": 3.7920281995661607,
+      "grad_norm": 0.16696877777576447,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 2797
+    },
+    {
+      "epoch": 3.7933839479392626,
+      "grad_norm": 0.2212374061346054,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2798
+    },
+    {
+      "epoch": 3.7947396963123645,
+      "grad_norm": 0.2377454936504364,
+      "learning_rate": 0.0001,
+      "loss": 0.1652,
+      "step": 2799
+    },
+    {
+      "epoch": 3.7960954446854664,
+      "grad_norm": 0.15816669166088104,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 2800
+    },
+    {
+      "epoch": 3.7974511930585684,
+      "grad_norm": 0.139626145362854,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 2801
+    },
+    {
+      "epoch": 3.7988069414316703,
+      "grad_norm": 0.13801495730876923,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 2802
+    },
+    {
+      "epoch": 3.8001626898047722,
+      "grad_norm": 0.1485288143157959,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 2803
+    },
+    {
+      "epoch": 3.801518438177874,
+      "grad_norm": 0.15326790511608124,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2804
+    },
+    {
+      "epoch": 3.802874186550976,
+      "grad_norm": 0.14995482563972473,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 2805
+    },
+    {
+      "epoch": 3.804229934924078,
+      "grad_norm": 0.15299053490161896,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 2806
+    },
+    {
+      "epoch": 3.80558568329718,
+      "grad_norm": 0.13578273355960846,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 2807
+    },
+    {
+      "epoch": 3.806941431670282,
+      "grad_norm": 0.14390116930007935,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2808
+    },
+    {
+      "epoch": 3.808297180043384,
+      "grad_norm": 0.14130103588104248,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 2809
+    },
+    {
+      "epoch": 3.8096529284164857,
+      "grad_norm": 0.1663191318511963,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 2810
+    },
+    {
+      "epoch": 3.811008676789588,
+      "grad_norm": 0.1893029510974884,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2811
+    },
+    {
+      "epoch": 3.8123644251626896,
+      "grad_norm": 0.14949651062488556,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 2812
+    },
+    {
+      "epoch": 3.813720173535792,
+      "grad_norm": 0.14349301159381866,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 2813
+    },
+    {
+      "epoch": 3.8150759219088934,
+      "grad_norm": 0.17193593084812164,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2814
+    },
+    {
+      "epoch": 3.816431670281996,
+      "grad_norm": 0.16820582747459412,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 2815
+    },
+    {
+      "epoch": 3.8177874186550977,
+      "grad_norm": 0.15749230980873108,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 2816
+    },
+    {
+      "epoch": 3.8191431670281997,
+      "grad_norm": 0.17658893764019012,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 2817
+    },
+    {
+      "epoch": 3.8204989154013016,
+      "grad_norm": 0.14029137790203094,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2818
+    },
+    {
+      "epoch": 3.8218546637744035,
+      "grad_norm": 0.1439065933227539,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2819
+    },
+    {
+      "epoch": 3.8232104121475055,
+      "grad_norm": 0.16036003828048706,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 2820
+    },
+    {
+      "epoch": 3.8245661605206074,
+      "grad_norm": 0.178243488073349,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2821
+    },
+    {
+      "epoch": 3.8259219088937093,
+      "grad_norm": 0.1941789984703064,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 2822
+    },
+    {
+      "epoch": 3.8272776572668112,
+      "grad_norm": 0.13744157552719116,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 2823
+    },
+    {
+      "epoch": 3.828633405639913,
+      "grad_norm": 0.16209127008914948,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 2824
+    },
+    {
+      "epoch": 3.829989154013015,
+      "grad_norm": 0.12773777544498444,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2825
+    },
+    {
+      "epoch": 3.831344902386117,
+      "grad_norm": 0.16617202758789062,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 2826
+    },
+    {
+      "epoch": 3.832700650759219,
+      "grad_norm": 0.17002686858177185,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2827
+    },
+    {
+      "epoch": 3.834056399132321,
+      "grad_norm": 0.15777309238910675,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 2828
+    },
+    {
+      "epoch": 3.835412147505423,
+      "grad_norm": 0.14134517312049866,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 2829
+    },
+    {
+      "epoch": 3.836767895878525,
+      "grad_norm": 0.1615971028804779,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2830
+    },
+    {
+      "epoch": 3.8381236442516267,
+      "grad_norm": 0.15932130813598633,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 2831
+    },
+    {
+      "epoch": 3.839479392624729,
+      "grad_norm": 0.15152499079704285,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 2832
+    },
+    {
+      "epoch": 3.840835140997831,
+      "grad_norm": 0.16078177094459534,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 2833
+    },
+    {
+      "epoch": 3.842190889370933,
+      "grad_norm": 0.15189415216445923,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 2834
+    },
+    {
+      "epoch": 3.843546637744035,
+      "grad_norm": 0.15980514883995056,
+      "learning_rate": 0.0001,
+      "loss": 0.1636,
+      "step": 2835
+    },
+    {
+      "epoch": 3.8449023861171367,
+      "grad_norm": 0.13054947555065155,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 2836
+    },
+    {
+      "epoch": 3.8462581344902387,
+      "grad_norm": 0.1309269219636917,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 2837
+    },
+    {
+      "epoch": 3.8476138828633406,
+      "grad_norm": 0.1666618287563324,
+      "learning_rate": 0.0001,
+      "loss": 0.176,
+      "step": 2838
+    },
+    {
+      "epoch": 3.8489696312364425,
+      "grad_norm": 0.1420322060585022,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 2839
+    },
+    {
+      "epoch": 3.8503253796095445,
+      "grad_norm": 0.1437780112028122,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 2840
+    },
+    {
+      "epoch": 3.8516811279826464,
+      "grad_norm": 0.14303426444530487,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2841
+    },
+    {
+      "epoch": 3.8530368763557483,
+      "grad_norm": 0.1704157590866089,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2842
+    },
+    {
+      "epoch": 3.8543926247288502,
+      "grad_norm": 0.3279649615287781,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 2843
+    },
+    {
+      "epoch": 3.855748373101952,
+      "grad_norm": 0.19302241504192352,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 2844
+    },
+    {
+      "epoch": 3.857104121475054,
+      "grad_norm": 0.15895956754684448,
+      "learning_rate": 0.0001,
+      "loss": 0.167,
+      "step": 2845
+    },
+    {
+      "epoch": 3.858459869848156,
+      "grad_norm": 0.21440915763378143,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 2846
+    },
+    {
+      "epoch": 3.8598156182212584,
+      "grad_norm": 0.15005962550640106,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 2847
+    },
+    {
+      "epoch": 3.86117136659436,
+      "grad_norm": 0.16833405196666718,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 2848
+    },
+    {
+      "epoch": 3.8625271149674623,
+      "grad_norm": 0.14465893805027008,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 2849
+    },
+    {
+      "epoch": 3.8638828633405637,
+      "grad_norm": 0.25789856910705566,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2850
+    },
+    {
+      "epoch": 3.865238611713666,
+      "grad_norm": 0.14764338731765747,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 2851
+    },
+    {
+      "epoch": 3.866594360086768,
+      "grad_norm": 0.1372634321451187,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 2852
+    },
+    {
+      "epoch": 3.86795010845987,
+      "grad_norm": 0.15925072133541107,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 2853
+    },
+    {
+      "epoch": 3.869305856832972,
+      "grad_norm": 0.1467796266078949,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2854
+    },
+    {
+      "epoch": 3.870661605206074,
+      "grad_norm": 0.137185201048851,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2855
+    },
+    {
+      "epoch": 3.8720173535791758,
+      "grad_norm": 0.16390427947044373,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2856
+    },
+    {
+      "epoch": 3.8733731019522777,
+      "grad_norm": 0.13747979700565338,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 2857
+    },
+    {
+      "epoch": 3.8747288503253796,
+      "grad_norm": 0.1651531606912613,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2858
+    },
+    {
+      "epoch": 3.8760845986984815,
+      "grad_norm": 0.15669108927249908,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 2859
+    },
+    {
+      "epoch": 3.8774403470715835,
+      "grad_norm": 0.18074403703212738,
+      "learning_rate": 0.0001,
+      "loss": 0.1699,
+      "step": 2860
+    },
+    {
+      "epoch": 3.8787960954446854,
+      "grad_norm": 0.1874494105577469,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 2861
+    },
+    {
+      "epoch": 3.8801518438177873,
+      "grad_norm": 0.16740724444389343,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 2862
+    },
+    {
+      "epoch": 3.8815075921908893,
+      "grad_norm": 0.16185173392295837,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 2863
+    },
+    {
+      "epoch": 3.882863340563991,
+      "grad_norm": 0.17150770127773285,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2864
+    },
+    {
+      "epoch": 3.884219088937093,
+      "grad_norm": 0.16465029120445251,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 2865
+    },
+    {
+      "epoch": 3.8855748373101955,
+      "grad_norm": 0.15753479301929474,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 2866
+    },
+    {
+      "epoch": 3.886930585683297,
+      "grad_norm": 0.23851017653942108,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 2867
+    },
+    {
+      "epoch": 3.8882863340563993,
+      "grad_norm": 0.14542925357818604,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 2868
+    },
+    {
+      "epoch": 3.889642082429501,
+      "grad_norm": 0.1427689790725708,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2869
+    },
+    {
+      "epoch": 3.890997830802603,
+      "grad_norm": 0.15112115442752838,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 2870
+    },
+    {
+      "epoch": 3.892353579175705,
+      "grad_norm": 0.5378669500350952,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2871
+    },
+    {
+      "epoch": 3.893709327548807,
+      "grad_norm": 0.1717974990606308,
+      "learning_rate": 0.0001,
+      "loss": 0.1623,
+      "step": 2872
+    },
+    {
+      "epoch": 3.895065075921909,
+      "grad_norm": 0.1533907800912857,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 2873
+    },
+    {
+      "epoch": 3.896420824295011,
+      "grad_norm": 0.16258637607097626,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2874
+    },
+    {
+      "epoch": 3.897776572668113,
+      "grad_norm": 0.16248074173927307,
+      "learning_rate": 0.0001,
+      "loss": 0.1696,
+      "step": 2875
+    },
+    {
+      "epoch": 3.8991323210412148,
+      "grad_norm": 0.155836284160614,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2876
+    },
+    {
+      "epoch": 3.9004880694143167,
+      "grad_norm": 0.17854665219783783,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 2877
+    },
+    {
+      "epoch": 3.9018438177874186,
+      "grad_norm": 0.1741473376750946,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 2878
+    },
+    {
+      "epoch": 3.9031995661605206,
+      "grad_norm": 0.1607498973608017,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 2879
+    },
+    {
+      "epoch": 3.9045553145336225,
+      "grad_norm": 0.1566384881734848,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 2880
+    },
+    {
+      "epoch": 3.9059110629067244,
+      "grad_norm": 0.15827764570713043,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 2881
+    },
+    {
+      "epoch": 3.9072668112798263,
+      "grad_norm": 0.1698606014251709,
+      "learning_rate": 0.0001,
+      "loss": 0.174,
+      "step": 2882
+    },
+    {
+      "epoch": 3.9086225596529283,
+      "grad_norm": 0.16148601472377777,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 2883
+    },
+    {
+      "epoch": 3.90997830802603,
+      "grad_norm": 0.1354098618030548,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 2884
+    },
+    {
+      "epoch": 3.9113340563991326,
+      "grad_norm": 0.17961689829826355,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 2885
+    },
+    {
+      "epoch": 3.912689804772234,
+      "grad_norm": 0.20635277032852173,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2886
+    },
+    {
+      "epoch": 3.9140455531453364,
+      "grad_norm": 0.15820805728435516,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2887
+    },
+    {
+      "epoch": 3.915401301518438,
+      "grad_norm": 0.15462972223758698,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 2888
+    },
+    {
+      "epoch": 3.9167570498915403,
+      "grad_norm": 0.15324915945529938,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2889
+    },
+    {
+      "epoch": 3.918112798264642,
+      "grad_norm": 0.14915446937084198,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 2890
+    },
+    {
+      "epoch": 3.919468546637744,
+      "grad_norm": 0.13327746093273163,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 2891
+    },
+    {
+      "epoch": 3.920824295010846,
+      "grad_norm": 0.15090790390968323,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2892
+    },
+    {
+      "epoch": 3.922180043383948,
+      "grad_norm": 0.17159931361675262,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 2893
+    },
+    {
+      "epoch": 3.92353579175705,
+      "grad_norm": 0.19333592057228088,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2894
+    },
+    {
+      "epoch": 3.924891540130152,
+      "grad_norm": 0.16537462174892426,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 2895
+    },
+    {
+      "epoch": 3.9262472885032538,
+      "grad_norm": 0.14215967059135437,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 2896
+    },
+    {
+      "epoch": 3.9276030368763557,
+      "grad_norm": 0.14604297280311584,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 2897
+    },
+    {
+      "epoch": 3.9289587852494576,
+      "grad_norm": 0.20058350265026093,
+      "learning_rate": 0.0001,
+      "loss": 0.1665,
+      "step": 2898
+    },
+    {
+      "epoch": 3.9303145336225596,
+      "grad_norm": 0.1720837503671646,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 2899
+    },
+    {
+      "epoch": 3.9316702819956615,
+      "grad_norm": 0.15031293034553528,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 2900
+    },
+    {
+      "epoch": 3.9330260303687634,
+      "grad_norm": 0.1626656949520111,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 2901
+    },
+    {
+      "epoch": 3.934381778741866,
+      "grad_norm": 0.15931637585163116,
+      "learning_rate": 0.0001,
+      "loss": 0.169,
+      "step": 2902
+    },
+    {
+      "epoch": 3.9357375271149673,
+      "grad_norm": 0.1602136790752411,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 2903
+    },
+    {
+      "epoch": 3.9370932754880696,
+      "grad_norm": 0.1575673669576645,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2904
+    },
+    {
+      "epoch": 3.938449023861171,
+      "grad_norm": 0.17590171098709106,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 2905
+    },
+    {
+      "epoch": 3.9398047722342735,
+      "grad_norm": 0.1493382602930069,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2906
+    },
+    {
+      "epoch": 3.9411605206073754,
+      "grad_norm": 0.1589704155921936,
+      "learning_rate": 0.0001,
+      "loss": 0.1618,
+      "step": 2907
+    },
+    {
+      "epoch": 3.9425162689804774,
+      "grad_norm": 0.17592303454875946,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 2908
+    },
+    {
+      "epoch": 3.9438720173535793,
+      "grad_norm": 0.16739055514335632,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2909
+    },
+    {
+      "epoch": 3.945227765726681,
+      "grad_norm": 0.16977554559707642,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 2910
+    },
+    {
+      "epoch": 3.946583514099783,
+      "grad_norm": 0.16012561321258545,
+      "learning_rate": 0.0001,
+      "loss": 0.1619,
+      "step": 2911
+    },
+    {
+      "epoch": 3.947939262472885,
+      "grad_norm": 0.1516956388950348,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 2912
+    },
+    {
+      "epoch": 3.949295010845987,
+      "grad_norm": 0.12917734682559967,
+      "learning_rate": 0.0001,
+      "loss": 0.1489,
+      "step": 2913
+    },
+    {
+      "epoch": 3.950650759219089,
+      "grad_norm": 0.15132172405719757,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 2914
+    },
+    {
+      "epoch": 3.952006507592191,
+      "grad_norm": 0.13989965617656708,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 2915
+    },
+    {
+      "epoch": 3.953362255965293,
+      "grad_norm": 0.1809939593076706,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 2916
+    },
+    {
+      "epoch": 3.9547180043383947,
+      "grad_norm": 0.14832746982574463,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 2917
+    },
+    {
+      "epoch": 3.9560737527114966,
+      "grad_norm": 0.1659536361694336,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 2918
+    },
+    {
+      "epoch": 3.9574295010845986,
+      "grad_norm": 0.1782654970884323,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 2919
+    },
+    {
+      "epoch": 3.9587852494577005,
+      "grad_norm": 0.1429128795862198,
+      "learning_rate": 0.0001,
+      "loss": 0.1632,
+      "step": 2920
+    },
+    {
+      "epoch": 3.960140997830803,
+      "grad_norm": 0.1613098382949829,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 2921
+    },
+    {
+      "epoch": 3.9614967462039044,
+      "grad_norm": 0.24199821054935455,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 2922
+    },
+    {
+      "epoch": 3.9628524945770067,
+      "grad_norm": 0.1532556265592575,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2923
+    },
+    {
+      "epoch": 3.964208242950108,
+      "grad_norm": 0.15800383687019348,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 2924
+    },
+    {
+      "epoch": 3.9655639913232106,
+      "grad_norm": 0.15722841024398804,
+      "learning_rate": 0.0001,
+      "loss": 0.1681,
+      "step": 2925
+    },
+    {
+      "epoch": 3.9669197396963125,
+      "grad_norm": 0.14384755492210388,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 2926
+    },
+    {
+      "epoch": 3.9682754880694144,
+      "grad_norm": 0.14615441858768463,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 2927
+    },
+    {
+      "epoch": 3.9696312364425164,
+      "grad_norm": 0.14744308590888977,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 2928
+    },
+    {
+      "epoch": 3.9709869848156183,
+      "grad_norm": 0.14238373935222626,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 2929
+    },
+    {
+      "epoch": 3.97234273318872,
+      "grad_norm": 0.14669066667556763,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 2930
+    },
+    {
+      "epoch": 3.973698481561822,
+      "grad_norm": 0.13884855806827545,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 2931
+    },
+    {
+      "epoch": 3.975054229934924,
+      "grad_norm": 0.179173082113266,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 2932
+    },
+    {
+      "epoch": 3.976409978308026,
+      "grad_norm": 0.14306093752384186,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 2933
+    },
+    {
+      "epoch": 3.977765726681128,
+      "grad_norm": 0.1840147078037262,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 2934
+    },
+    {
+      "epoch": 3.97912147505423,
+      "grad_norm": 0.1536615639925003,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 2935
+    },
+    {
+      "epoch": 3.980477223427332,
+      "grad_norm": 0.1712505966424942,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 2936
+    },
+    {
+      "epoch": 3.9818329718004337,
+      "grad_norm": 0.13559499382972717,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 2937
+    },
+    {
+      "epoch": 3.9831887201735356,
+      "grad_norm": 0.13713407516479492,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 2938
+    },
+    {
+      "epoch": 3.9845444685466376,
+      "grad_norm": 0.1530856192111969,
+      "learning_rate": 0.0001,
+      "loss": 0.1646,
+      "step": 2939
+    },
+    {
+      "epoch": 3.98590021691974,
+      "grad_norm": 0.14117655158042908,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 2940
+    },
+    {
+      "epoch": 3.9872559652928414,
+      "grad_norm": 0.14709687232971191,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 2941
+    },
+    {
+      "epoch": 3.988611713665944,
+      "grad_norm": 0.1643400937318802,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2942
+    },
+    {
+      "epoch": 3.9899674620390453,
+      "grad_norm": 0.13236233592033386,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 2943
+    },
+    {
+      "epoch": 3.9913232104121477,
+      "grad_norm": 0.1504112184047699,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 2944
+    },
+    {
+      "epoch": 3.9926789587852496,
+      "grad_norm": 0.15114104747772217,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 2945
+    },
+    {
+      "epoch": 3.9940347071583515,
+      "grad_norm": 0.15687744319438934,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 2946
+    },
+    {
+      "epoch": 3.9953904555314534,
+      "grad_norm": 0.1594519466161728,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2947
+    },
+    {
+      "epoch": 3.9967462039045554,
+      "grad_norm": 0.12185032665729523,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 2948
+    },
+    {
+      "epoch": 3.9981019522776573,
+      "grad_norm": 0.12994550168514252,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 2949
+    },
+    {
+      "epoch": 3.9994577006507592,
+      "grad_norm": 0.16108247637748718,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 2950
+    },
+    {
+      "epoch": 4.000813449023862,
+      "grad_norm": 0.15835429728031158,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 2951
+    },
+    {
+      "epoch": 4.002169197396963,
+      "grad_norm": 0.15806114673614502,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 2952
+    },
+    {
+      "epoch": 4.0035249457700655,
+      "grad_norm": 0.13707749545574188,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 2953
+    },
+    {
+      "epoch": 4.004880694143167,
+      "grad_norm": 0.1522742211818695,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 2954
+    },
+    {
+      "epoch": 4.006236442516269,
+      "grad_norm": 0.15703125298023224,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 2955
+    },
+    {
+      "epoch": 4.007592190889371,
+      "grad_norm": 0.20340880751609802,
+      "learning_rate": 0.0001,
+      "loss": 0.1653,
+      "step": 2956
+    },
+    {
+      "epoch": 4.008947939262473,
+      "grad_norm": 0.16579876840114594,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 2957
+    },
+    {
+      "epoch": 4.010303687635575,
+      "grad_norm": 0.19139209389686584,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 2958
+    },
+    {
+      "epoch": 4.011659436008677,
+      "grad_norm": 0.19883209466934204,
+      "learning_rate": 0.0001,
+      "loss": 0.1635,
+      "step": 2959
+    },
+    {
+      "epoch": 4.0130151843817785,
+      "grad_norm": 0.1661502569913864,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 2960
+    },
+    {
+      "epoch": 4.014370932754881,
+      "grad_norm": 0.18733543157577515,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 2961
+    },
+    {
+      "epoch": 4.015726681127982,
+      "grad_norm": 0.26604872941970825,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 2962
+    },
+    {
+      "epoch": 4.017082429501085,
+      "grad_norm": 0.2066808044910431,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 2963
+    },
+    {
+      "epoch": 4.018438177874186,
+      "grad_norm": 0.17991168797016144,
+      "learning_rate": 0.0001,
+      "loss": 0.1343,
+      "step": 2964
+    },
+    {
+      "epoch": 4.019793926247289,
+      "grad_norm": 0.19640707969665527,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 2965
+    },
+    {
+      "epoch": 4.02114967462039,
+      "grad_norm": 0.29766592383384705,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 2966
+    },
+    {
+      "epoch": 4.0225054229934925,
+      "grad_norm": 0.32087722420692444,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 2967
+    },
+    {
+      "epoch": 4.023861171366594,
+      "grad_norm": 0.1700161248445511,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 2968
+    },
+    {
+      "epoch": 4.025216919739696,
+      "grad_norm": 0.1960957795381546,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 2969
+    },
+    {
+      "epoch": 4.026572668112799,
+      "grad_norm": 0.31061866879463196,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 2970
+    },
+    {
+      "epoch": 4.0279284164859,
+      "grad_norm": 0.19869892299175262,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 2971
+    },
+    {
+      "epoch": 4.0292841648590025,
+      "grad_norm": 0.17028412222862244,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 2972
+    },
+    {
+      "epoch": 4.030639913232104,
+      "grad_norm": 0.16271734237670898,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 2973
+    },
+    {
+      "epoch": 4.031995661605206,
+      "grad_norm": 0.20057597756385803,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 2974
+    },
+    {
+      "epoch": 4.033351409978308,
+      "grad_norm": 0.18573065102100372,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 2975
+    },
+    {
+      "epoch": 4.03470715835141,
+      "grad_norm": 0.1508813053369522,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "step": 2976
+    },
+    {
+      "epoch": 4.036062906724512,
+      "grad_norm": 0.1559394896030426,
+      "learning_rate": 0.0001,
+      "loss": 0.1484,
+      "step": 2977
+    },
+    {
+      "epoch": 4.037418655097614,
+      "grad_norm": 0.1744004189968109,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 2978
+    },
+    {
+      "epoch": 4.038774403470716,
+      "grad_norm": 0.1584269255399704,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 2979
+    },
+    {
+      "epoch": 4.040130151843818,
+      "grad_norm": 0.16780586540699005,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 2980
+    },
+    {
+      "epoch": 4.0414859002169194,
+      "grad_norm": 0.15435896813869476,
+      "learning_rate": 0.0001,
+      "loss": 0.1278,
+      "step": 2981
+    },
+    {
+      "epoch": 4.042841648590022,
+      "grad_norm": 0.17540611326694489,
+      "learning_rate": 0.0001,
+      "loss": 0.1396,
+      "step": 2982
+    },
+    {
+      "epoch": 4.044197396963123,
+      "grad_norm": 0.16336849331855774,
+      "learning_rate": 0.0001,
+      "loss": 0.1406,
+      "step": 2983
+    },
+    {
+      "epoch": 4.045553145336226,
+      "grad_norm": 0.15283925831317902,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 2984
+    },
+    {
+      "epoch": 4.046908893709327,
+      "grad_norm": 0.1745975911617279,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 2985
+    },
+    {
+      "epoch": 4.0482646420824295,
+      "grad_norm": 0.16627509891986847,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 2986
+    },
+    {
+      "epoch": 4.049620390455531,
+      "grad_norm": 0.30267533659935,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 2987
+    },
+    {
+      "epoch": 4.050976138828633,
+      "grad_norm": 0.17126211524009705,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 2988
+    },
+    {
+      "epoch": 4.052331887201736,
+      "grad_norm": 0.16338945925235748,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 2989
+    },
+    {
+      "epoch": 4.053687635574837,
+      "grad_norm": 0.2238633781671524,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 2990
+    },
+    {
+      "epoch": 4.05504338394794,
+      "grad_norm": 0.16619086265563965,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "step": 2991
+    },
+    {
+      "epoch": 4.056399132321041,
+      "grad_norm": 0.16802458465099335,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 2992
+    },
+    {
+      "epoch": 4.0577548806941435,
+      "grad_norm": 0.2000352144241333,
+      "learning_rate": 0.0001,
+      "loss": 0.1315,
+      "step": 2993
+    },
+    {
+      "epoch": 4.059110629067245,
+      "grad_norm": 0.20343512296676636,
+      "learning_rate": 0.0001,
+      "loss": 0.1478,
+      "step": 2994
+    },
+    {
+      "epoch": 4.060466377440347,
+      "grad_norm": 0.18095913529396057,
+      "learning_rate": 0.0001,
+      "loss": 0.1305,
+      "step": 2995
+    },
+    {
+      "epoch": 4.061822125813449,
+      "grad_norm": 0.15246407687664032,
+      "learning_rate": 0.0001,
+      "loss": 0.1353,
+      "step": 2996
+    },
+    {
+      "epoch": 4.063177874186551,
+      "grad_norm": 0.17788852751255035,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 2997
+    },
+    {
+      "epoch": 4.064533622559653,
+      "grad_norm": 0.19316218793392181,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 2998
+    },
+    {
+      "epoch": 4.065889370932755,
+      "grad_norm": 0.17224259674549103,
+      "learning_rate": 0.0001,
+      "loss": 0.1366,
+      "step": 2999
+    },
+    {
+      "epoch": 4.0672451193058565,
+      "grad_norm": 0.21991267800331116,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 3000
+    },
+    {
+      "epoch": 4.0672451193058565,
+      "eval_loss": 0.19210749864578247,
+      "eval_runtime": 588.315,
+      "eval_samples_per_second": 4.702,
+      "eval_steps_per_second": 1.176,
+      "step": 3000
+    },
+    {
+      "epoch": 4.068600867678959,
+      "grad_norm": 0.19707490503787994,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 3001
+    },
+    {
+      "epoch": 4.06995661605206,
+      "grad_norm": 0.15894325077533722,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 3002
+    },
+    {
+      "epoch": 4.071312364425163,
+      "grad_norm": 0.1848834604024887,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 3003
+    },
+    {
+      "epoch": 4.072668112798264,
+      "grad_norm": 0.20216555893421173,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 3004
+    },
+    {
+      "epoch": 4.074023861171367,
+      "grad_norm": 0.19862866401672363,
+      "learning_rate": 0.0001,
+      "loss": 0.148,
+      "step": 3005
+    },
+    {
+      "epoch": 4.075379609544468,
+      "grad_norm": 0.2740095853805542,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 3006
+    },
+    {
+      "epoch": 4.0767353579175705,
+      "grad_norm": 0.20887617766857147,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 3007
+    },
+    {
+      "epoch": 4.078091106290673,
+      "grad_norm": 0.2118336409330368,
+      "learning_rate": 0.0001,
+      "loss": 0.1684,
+      "step": 3008
+    },
+    {
+      "epoch": 4.079446854663774,
+      "grad_norm": 0.17214307188987732,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 3009
+    },
+    {
+      "epoch": 4.080802603036877,
+      "grad_norm": 0.17742818593978882,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 3010
+    },
+    {
+      "epoch": 4.082158351409978,
+      "grad_norm": 0.20524069666862488,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 3011
+    },
+    {
+      "epoch": 4.0835140997830806,
+      "grad_norm": 0.16799645125865936,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3012
+    },
+    {
+      "epoch": 4.084869848156182,
+      "grad_norm": 0.19054214656352997,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 3013
+    },
+    {
+      "epoch": 4.086225596529284,
+      "grad_norm": 0.18611684441566467,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 3014
+    },
+    {
+      "epoch": 4.087581344902386,
+      "grad_norm": 0.20501847565174103,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 3015
+    },
+    {
+      "epoch": 4.088937093275488,
+      "grad_norm": 0.19660042226314545,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 3016
+    },
+    {
+      "epoch": 4.09029284164859,
+      "grad_norm": 0.16619573533535004,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 3017
+    },
+    {
+      "epoch": 4.091648590021692,
+      "grad_norm": 0.1531042754650116,
+      "learning_rate": 0.0001,
+      "loss": 0.1322,
+      "step": 3018
+    },
+    {
+      "epoch": 4.093004338394794,
+      "grad_norm": 0.13734383881092072,
+      "learning_rate": 0.0001,
+      "loss": 0.1339,
+      "step": 3019
+    },
+    {
+      "epoch": 4.094360086767896,
+      "grad_norm": 0.16135041415691376,
+      "learning_rate": 0.0001,
+      "loss": 0.1392,
+      "step": 3020
+    },
+    {
+      "epoch": 4.0957158351409975,
+      "grad_norm": 0.17987558245658875,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 3021
+    },
+    {
+      "epoch": 4.0970715835141,
+      "grad_norm": 0.20646414160728455,
+      "learning_rate": 0.0001,
+      "loss": 0.1704,
+      "step": 3022
+    },
+    {
+      "epoch": 4.098427331887201,
+      "grad_norm": 0.18541377782821655,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 3023
+    },
+    {
+      "epoch": 4.099783080260304,
+      "grad_norm": 0.17776483297348022,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 3024
+    },
+    {
+      "epoch": 4.101138828633406,
+      "grad_norm": 0.16828671097755432,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 3025
+    },
+    {
+      "epoch": 4.1024945770065075,
+      "grad_norm": 0.23097951710224152,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 3026
+    },
+    {
+      "epoch": 4.10385032537961,
+      "grad_norm": 0.23153112828731537,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 3027
+    },
+    {
+      "epoch": 4.105206073752711,
+      "grad_norm": 0.1904088258743286,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3028
+    },
+    {
+      "epoch": 4.106561822125814,
+      "grad_norm": 0.1716212034225464,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3029
+    },
+    {
+      "epoch": 4.107917570498915,
+      "grad_norm": 0.17687425017356873,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 3030
+    },
+    {
+      "epoch": 4.109273318872018,
+      "grad_norm": 0.18178464472293854,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 3031
+    },
+    {
+      "epoch": 4.110629067245119,
+      "grad_norm": 0.18645192682743073,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 3032
+    },
+    {
+      "epoch": 4.1119848156182215,
+      "grad_norm": 0.16422325372695923,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3033
+    },
+    {
+      "epoch": 4.113340563991323,
+      "grad_norm": 0.18043223023414612,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 3034
+    },
+    {
+      "epoch": 4.114696312364425,
+      "grad_norm": 0.17650586366653442,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3035
+    },
+    {
+      "epoch": 4.116052060737527,
+      "grad_norm": 0.2132258415222168,
+      "learning_rate": 0.0001,
+      "loss": 0.1648,
+      "step": 3036
+    },
+    {
+      "epoch": 4.117407809110629,
+      "grad_norm": 0.1852848380804062,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 3037
+    },
+    {
+      "epoch": 4.118763557483731,
+      "grad_norm": 0.16845722496509552,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 3038
+    },
+    {
+      "epoch": 4.120119305856833,
+      "grad_norm": 0.17701689898967743,
+      "learning_rate": 0.0001,
+      "loss": 0.1261,
+      "step": 3039
+    },
+    {
+      "epoch": 4.1214750542299345,
+      "grad_norm": 0.15773285925388336,
+      "learning_rate": 0.0001,
+      "loss": 0.1318,
+      "step": 3040
+    },
+    {
+      "epoch": 4.122830802603037,
+      "grad_norm": 0.2092285007238388,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3041
+    },
+    {
+      "epoch": 4.124186550976138,
+      "grad_norm": 0.2032230794429779,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 3042
+    },
+    {
+      "epoch": 4.125542299349241,
+      "grad_norm": 0.18238627910614014,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 3043
+    },
+    {
+      "epoch": 4.126898047722343,
+      "grad_norm": 0.18863163888454437,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 3044
+    },
+    {
+      "epoch": 4.128253796095445,
+      "grad_norm": 0.16953600943088531,
+      "learning_rate": 0.0001,
+      "loss": 0.1412,
+      "step": 3045
+    },
+    {
+      "epoch": 4.129609544468547,
+      "grad_norm": 0.2239847332239151,
+      "learning_rate": 0.0001,
+      "loss": 0.1683,
+      "step": 3046
+    },
+    {
+      "epoch": 4.1309652928416485,
+      "grad_norm": 0.22333739697933197,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 3047
+    },
+    {
+      "epoch": 4.132321041214751,
+      "grad_norm": 0.18489046394824982,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 3048
+    },
+    {
+      "epoch": 4.133676789587852,
+      "grad_norm": 0.20239531993865967,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 3049
+    },
+    {
+      "epoch": 4.135032537960955,
+      "grad_norm": 0.18180572986602783,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 3050
+    },
+    {
+      "epoch": 4.136388286334056,
+      "grad_norm": 0.16650179028511047,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3051
+    },
+    {
+      "epoch": 4.137744034707159,
+      "grad_norm": 0.18644386529922485,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 3052
+    },
+    {
+      "epoch": 4.13909978308026,
+      "grad_norm": 0.1759636104106903,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 3053
+    },
+    {
+      "epoch": 4.140455531453362,
+      "grad_norm": 0.2120070606470108,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 3054
+    },
+    {
+      "epoch": 4.141811279826464,
+      "grad_norm": 0.18065118789672852,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 3055
+    },
+    {
+      "epoch": 4.143167028199566,
+      "grad_norm": 0.17110495269298553,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3056
+    },
+    {
+      "epoch": 4.144522776572668,
+      "grad_norm": 0.1619340181350708,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 3057
+    },
+    {
+      "epoch": 4.14587852494577,
+      "grad_norm": 0.1736074984073639,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 3058
+    },
+    {
+      "epoch": 4.147234273318872,
+      "grad_norm": 0.15600594878196716,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 3059
+    },
+    {
+      "epoch": 4.148590021691974,
+      "grad_norm": 0.16228684782981873,
+      "learning_rate": 0.0001,
+      "loss": 0.1251,
+      "step": 3060
+    },
+    {
+      "epoch": 4.1499457700650755,
+      "grad_norm": 0.16248080134391785,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 3061
+    },
+    {
+      "epoch": 4.151301518438178,
+      "grad_norm": 0.1900682896375656,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 3062
+    },
+    {
+      "epoch": 4.15265726681128,
+      "grad_norm": 0.16430655121803284,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 3063
+    },
+    {
+      "epoch": 4.154013015184382,
+      "grad_norm": 0.17080210149288177,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3064
+    },
+    {
+      "epoch": 4.155368763557484,
+      "grad_norm": 0.16991880536079407,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 3065
+    },
+    {
+      "epoch": 4.156724511930586,
+      "grad_norm": 0.17976613342761993,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3066
+    },
+    {
+      "epoch": 4.158080260303688,
+      "grad_norm": 0.17202334105968475,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 3067
+    },
+    {
+      "epoch": 4.159436008676789,
+      "grad_norm": 0.15483422577381134,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 3068
+    },
+    {
+      "epoch": 4.160791757049892,
+      "grad_norm": 0.1812533289194107,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 3069
+    },
+    {
+      "epoch": 4.162147505422993,
+      "grad_norm": 0.21884404122829437,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 3070
+    },
+    {
+      "epoch": 4.163503253796096,
+      "grad_norm": 0.16797062754631042,
+      "learning_rate": 0.0001,
+      "loss": 0.1372,
+      "step": 3071
+    },
+    {
+      "epoch": 4.164859002169197,
+      "grad_norm": 0.16521568596363068,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 3072
+    },
+    {
+      "epoch": 4.1662147505422995,
+      "grad_norm": 0.15489068627357483,
+      "learning_rate": 0.0001,
+      "loss": 0.1319,
+      "step": 3073
+    },
+    {
+      "epoch": 4.167570498915401,
+      "grad_norm": 0.17856468260288239,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 3074
+    },
+    {
+      "epoch": 4.168926247288503,
+      "grad_norm": 0.1809462159872055,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 3075
+    },
+    {
+      "epoch": 4.170281995661605,
+      "grad_norm": 0.17608648538589478,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 3076
+    },
+    {
+      "epoch": 4.171637744034707,
+      "grad_norm": 0.18130168318748474,
+      "learning_rate": 0.0001,
+      "loss": 0.1568,
+      "step": 3077
+    },
+    {
+      "epoch": 4.172993492407809,
+      "grad_norm": 0.1824263036251068,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 3078
+    },
+    {
+      "epoch": 4.174349240780911,
+      "grad_norm": 0.15585611760616302,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 3079
+    },
+    {
+      "epoch": 4.1757049891540134,
+      "grad_norm": 0.16141767799854279,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 3080
+    },
+    {
+      "epoch": 4.177060737527115,
+      "grad_norm": 0.1554366648197174,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 3081
+    },
+    {
+      "epoch": 4.178416485900217,
+      "grad_norm": 0.18599915504455566,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 3082
+    },
+    {
+      "epoch": 4.179772234273319,
+      "grad_norm": 0.20063552260398865,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 3083
+    },
+    {
+      "epoch": 4.181127982646421,
+      "grad_norm": 0.18506528437137604,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 3084
+    },
+    {
+      "epoch": 4.182483731019523,
+      "grad_norm": 0.17456820607185364,
+      "learning_rate": 0.0001,
+      "loss": 0.1397,
+      "step": 3085
+    },
+    {
+      "epoch": 4.183839479392625,
+      "grad_norm": 0.17862217128276825,
+      "learning_rate": 0.0001,
+      "loss": 0.1433,
+      "step": 3086
+    },
+    {
+      "epoch": 4.1851952277657265,
+      "grad_norm": 0.18172699213027954,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3087
+    },
+    {
+      "epoch": 4.186550976138829,
+      "grad_norm": 0.18565739691257477,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3088
+    },
+    {
+      "epoch": 4.18790672451193,
+      "grad_norm": 0.17048317193984985,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 3089
+    },
+    {
+      "epoch": 4.189262472885033,
+      "grad_norm": 0.19380158185958862,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 3090
+    },
+    {
+      "epoch": 4.190618221258134,
+      "grad_norm": 0.18565168976783752,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3091
+    },
+    {
+      "epoch": 4.191973969631237,
+      "grad_norm": 0.17944537103176117,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 3092
+    },
+    {
+      "epoch": 4.193329718004338,
+      "grad_norm": 0.14856137335300446,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 3093
+    },
+    {
+      "epoch": 4.19468546637744,
+      "grad_norm": 0.2340715378522873,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 3094
+    },
+    {
+      "epoch": 4.196041214750542,
+      "grad_norm": 0.16668421030044556,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 3095
+    },
+    {
+      "epoch": 4.197396963123644,
+      "grad_norm": 0.1781020611524582,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 3096
+    },
+    {
+      "epoch": 4.198752711496746,
+      "grad_norm": 0.13747841119766235,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 3097
+    },
+    {
+      "epoch": 4.200108459869848,
+      "grad_norm": 0.15311746299266815,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 3098
+    },
+    {
+      "epoch": 4.2014642082429505,
+      "grad_norm": 0.16055209934711456,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 3099
+    },
+    {
+      "epoch": 4.202819956616052,
+      "grad_norm": 0.16564621031284332,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 3100
+    },
+    {
+      "epoch": 4.204175704989154,
+      "grad_norm": 0.22573426365852356,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 3101
+    },
+    {
+      "epoch": 4.205531453362256,
+      "grad_norm": 0.19351111352443695,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 3102
+    },
+    {
+      "epoch": 4.206887201735358,
+      "grad_norm": 0.1578178107738495,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3103
+    },
+    {
+      "epoch": 4.20824295010846,
+      "grad_norm": 0.17999866604804993,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3104
+    },
+    {
+      "epoch": 4.209598698481562,
+      "grad_norm": 0.19283939898014069,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 3105
+    },
+    {
+      "epoch": 4.210954446854664,
+      "grad_norm": 0.15185318887233734,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 3106
+    },
+    {
+      "epoch": 4.212310195227766,
+      "grad_norm": 0.17106957733631134,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 3107
+    },
+    {
+      "epoch": 4.213665943600867,
+      "grad_norm": 0.18885555863380432,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3108
+    },
+    {
+      "epoch": 4.21502169197397,
+      "grad_norm": 0.17076760530471802,
+      "learning_rate": 0.0001,
+      "loss": 0.1365,
+      "step": 3109
+    },
+    {
+      "epoch": 4.216377440347071,
+      "grad_norm": 0.19526001811027527,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3110
+    },
+    {
+      "epoch": 4.217733188720174,
+      "grad_norm": 0.16689838469028473,
+      "learning_rate": 0.0001,
+      "loss": 0.1343,
+      "step": 3111
+    },
+    {
+      "epoch": 4.219088937093275,
+      "grad_norm": 0.16316482424736023,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3112
+    },
+    {
+      "epoch": 4.2204446854663775,
+      "grad_norm": 0.19420979917049408,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 3113
+    },
+    {
+      "epoch": 4.221800433839479,
+      "grad_norm": 0.2032974511384964,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3114
+    },
+    {
+      "epoch": 4.223156182212581,
+      "grad_norm": 0.1723061352968216,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 3115
+    },
+    {
+      "epoch": 4.224511930585683,
+      "grad_norm": 0.1779283583164215,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 3116
+    },
+    {
+      "epoch": 4.225867678958785,
+      "grad_norm": 0.2101929634809494,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3117
+    },
+    {
+      "epoch": 4.227223427331888,
+      "grad_norm": 0.1752268671989441,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 3118
+    },
+    {
+      "epoch": 4.228579175704989,
+      "grad_norm": 0.16272756457328796,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 3119
+    },
+    {
+      "epoch": 4.2299349240780915,
+      "grad_norm": 0.16360008716583252,
+      "learning_rate": 0.0001,
+      "loss": 0.1427,
+      "step": 3120
+    },
+    {
+      "epoch": 4.231290672451193,
+      "grad_norm": 0.15849001705646515,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 3121
+    },
+    {
+      "epoch": 4.232646420824295,
+      "grad_norm": 0.18035705387592316,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 3122
+    },
+    {
+      "epoch": 4.234002169197397,
+      "grad_norm": 0.179602712392807,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3123
+    },
+    {
+      "epoch": 4.235357917570499,
+      "grad_norm": 0.2349834144115448,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 3124
+    },
+    {
+      "epoch": 4.236713665943601,
+      "grad_norm": 0.17465510964393616,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3125
+    },
+    {
+      "epoch": 4.238069414316703,
+      "grad_norm": 0.21605047583580017,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3126
+    },
+    {
+      "epoch": 4.2394251626898045,
+      "grad_norm": 0.16568467020988464,
+      "learning_rate": 0.0001,
+      "loss": 0.1326,
+      "step": 3127
+    },
+    {
+      "epoch": 4.240780911062907,
+      "grad_norm": 0.17276495695114136,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 3128
+    },
+    {
+      "epoch": 4.242136659436008,
+      "grad_norm": 0.1747943013906479,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 3129
+    },
+    {
+      "epoch": 4.243492407809111,
+      "grad_norm": 0.17195089161396027,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3130
+    },
+    {
+      "epoch": 4.244848156182212,
+      "grad_norm": 0.17979048192501068,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 3131
+    },
+    {
+      "epoch": 4.246203904555315,
+      "grad_norm": 0.1780099868774414,
+      "learning_rate": 0.0001,
+      "loss": 0.1353,
+      "step": 3132
+    },
+    {
+      "epoch": 4.247559652928416,
+      "grad_norm": 0.20652218163013458,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 3133
+    },
+    {
+      "epoch": 4.2489154013015185,
+      "grad_norm": 0.18512703478336334,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 3134
+    },
+    {
+      "epoch": 4.250271149674621,
+      "grad_norm": 0.16563622653484344,
+      "learning_rate": 0.0001,
+      "loss": 0.1511,
+      "step": 3135
+    },
+    {
+      "epoch": 4.251626898047722,
+      "grad_norm": 0.19591963291168213,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 3136
+    },
+    {
+      "epoch": 4.252982646420825,
+      "grad_norm": 0.15947504341602325,
+      "learning_rate": 0.0001,
+      "loss": 0.1401,
+      "step": 3137
+    },
+    {
+      "epoch": 4.254338394793926,
+      "grad_norm": 0.15142714977264404,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3138
+    },
+    {
+      "epoch": 4.2556941431670285,
+      "grad_norm": 0.1635928601026535,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 3139
+    },
+    {
+      "epoch": 4.25704989154013,
+      "grad_norm": 0.16013139486312866,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 3140
+    },
+    {
+      "epoch": 4.258405639913232,
+      "grad_norm": 0.15780827403068542,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 3141
+    },
+    {
+      "epoch": 4.259761388286334,
+      "grad_norm": 0.15048468112945557,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 3142
+    },
+    {
+      "epoch": 4.261117136659436,
+      "grad_norm": 0.17292137444019318,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 3143
+    },
+    {
+      "epoch": 4.262472885032538,
+      "grad_norm": 0.15934589505195618,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 3144
+    },
+    {
+      "epoch": 4.26382863340564,
+      "grad_norm": 0.21587221324443817,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 3145
+    },
+    {
+      "epoch": 4.265184381778742,
+      "grad_norm": 0.17930743098258972,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 3146
+    },
+    {
+      "epoch": 4.266540130151844,
+      "grad_norm": 0.15549559891223907,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 3147
+    },
+    {
+      "epoch": 4.2678958785249455,
+      "grad_norm": 0.1762574315071106,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 3148
+    },
+    {
+      "epoch": 4.269251626898048,
+      "grad_norm": 0.20732329785823822,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 3149
+    },
+    {
+      "epoch": 4.270607375271149,
+      "grad_norm": 0.1625312864780426,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 3150
+    },
+    {
+      "epoch": 4.271963123644252,
+      "grad_norm": 0.16769258677959442,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 3151
+    },
+    {
+      "epoch": 4.273318872017353,
+      "grad_norm": 0.19604326784610748,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 3152
+    },
+    {
+      "epoch": 4.2746746203904555,
+      "grad_norm": 0.1918765753507614,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 3153
+    },
+    {
+      "epoch": 4.276030368763557,
+      "grad_norm": 0.18067851662635803,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 3154
+    },
+    {
+      "epoch": 4.277386117136659,
+      "grad_norm": 0.17459744215011597,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 3155
+    },
+    {
+      "epoch": 4.278741865509762,
+      "grad_norm": 0.1897064596414566,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 3156
+    },
+    {
+      "epoch": 4.280097613882863,
+      "grad_norm": 0.19426338374614716,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 3157
+    },
+    {
+      "epoch": 4.281453362255966,
+      "grad_norm": 0.16044163703918457,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 3158
+    },
+    {
+      "epoch": 4.282809110629067,
+      "grad_norm": 0.1960867941379547,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 3159
+    },
+    {
+      "epoch": 4.2841648590021695,
+      "grad_norm": 0.19970008730888367,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 3160
+    },
+    {
+      "epoch": 4.285520607375271,
+      "grad_norm": 0.17267131805419922,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3161
+    },
+    {
+      "epoch": 4.286876355748373,
+      "grad_norm": 0.17330409586429596,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 3162
+    },
+    {
+      "epoch": 4.288232104121475,
+      "grad_norm": 0.16853345930576324,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3163
+    },
+    {
+      "epoch": 4.289587852494577,
+      "grad_norm": 0.17296352982521057,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 3164
+    },
+    {
+      "epoch": 4.290943600867679,
+      "grad_norm": 0.19540084898471832,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 3165
+    },
+    {
+      "epoch": 4.292299349240781,
+      "grad_norm": 0.17911715805530548,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 3166
+    },
+    {
+      "epoch": 4.2936550976138825,
+      "grad_norm": 0.19178162515163422,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 3167
+    },
+    {
+      "epoch": 4.295010845986985,
+      "grad_norm": 0.174775168299675,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 3168
+    },
+    {
+      "epoch": 4.296366594360086,
+      "grad_norm": 0.19943968951702118,
+      "learning_rate": 0.0001,
+      "loss": 0.148,
+      "step": 3169
+    },
+    {
+      "epoch": 4.297722342733189,
+      "grad_norm": 0.1918841302394867,
+      "learning_rate": 0.0001,
+      "loss": 0.1654,
+      "step": 3170
+    },
+    {
+      "epoch": 4.29907809110629,
+      "grad_norm": 0.15276604890823364,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3171
+    },
+    {
+      "epoch": 4.300433839479393,
+      "grad_norm": 0.14646317064762115,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 3172
+    },
+    {
+      "epoch": 4.301789587852495,
+      "grad_norm": 0.18695466220378876,
+      "learning_rate": 0.0001,
+      "loss": 0.154,
+      "step": 3173
+    },
+    {
+      "epoch": 4.3031453362255965,
+      "grad_norm": 0.16479003429412842,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 3174
+    },
+    {
+      "epoch": 4.304501084598699,
+      "grad_norm": 0.16377784311771393,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 3175
+    },
+    {
+      "epoch": 4.3058568329718,
+      "grad_norm": 0.17996813356876373,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 3176
+    },
+    {
+      "epoch": 4.307212581344903,
+      "grad_norm": 0.16615524888038635,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 3177
+    },
+    {
+      "epoch": 4.308568329718004,
+      "grad_norm": 0.17665554583072662,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3178
+    },
+    {
+      "epoch": 4.309924078091107,
+      "grad_norm": 0.14471717178821564,
+      "learning_rate": 0.0001,
+      "loss": 0.1392,
+      "step": 3179
+    },
+    {
+      "epoch": 4.311279826464208,
+      "grad_norm": 0.15263982117176056,
+      "learning_rate": 0.0001,
+      "loss": 0.1415,
+      "step": 3180
+    },
+    {
+      "epoch": 4.31263557483731,
+      "grad_norm": 0.16039493680000305,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 3181
+    },
+    {
+      "epoch": 4.313991323210412,
+      "grad_norm": 0.20201976597309113,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 3182
+    },
+    {
+      "epoch": 4.315347071583514,
+      "grad_norm": 0.18662229180335999,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 3183
+    },
+    {
+      "epoch": 4.316702819956616,
+      "grad_norm": 0.15628182888031006,
+      "learning_rate": 0.0001,
+      "loss": 0.1478,
+      "step": 3184
+    },
+    {
+      "epoch": 4.318058568329718,
+      "grad_norm": 0.194144606590271,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 3185
+    },
+    {
+      "epoch": 4.31941431670282,
+      "grad_norm": 0.1699649542570114,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 3186
+    },
+    {
+      "epoch": 4.320770065075922,
+      "grad_norm": 0.17831355333328247,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 3187
+    },
+    {
+      "epoch": 4.3221258134490235,
+      "grad_norm": 0.1611872911453247,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 3188
+    },
+    {
+      "epoch": 4.323481561822126,
+      "grad_norm": 0.18529847264289856,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 3189
+    },
+    {
+      "epoch": 4.324837310195228,
+      "grad_norm": 0.16061431169509888,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 3190
+    },
+    {
+      "epoch": 4.32619305856833,
+      "grad_norm": 0.18216833472251892,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3191
+    },
+    {
+      "epoch": 4.327548806941432,
+      "grad_norm": 0.16705365478992462,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3192
+    },
+    {
+      "epoch": 4.3289045553145336,
+      "grad_norm": 0.18242287635803223,
+      "learning_rate": 0.0001,
+      "loss": 0.1439,
+      "step": 3193
+    },
+    {
+      "epoch": 4.330260303687636,
+      "grad_norm": 0.16203604638576508,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 3194
+    },
+    {
+      "epoch": 4.331616052060737,
+      "grad_norm": 0.18025241792201996,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3195
+    },
+    {
+      "epoch": 4.33297180043384,
+      "grad_norm": 0.20001383125782013,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 3196
+    },
+    {
+      "epoch": 4.334327548806941,
+      "grad_norm": 0.18947085738182068,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3197
+    },
+    {
+      "epoch": 4.335683297180044,
+      "grad_norm": 0.1779194176197052,
+      "learning_rate": 0.0001,
+      "loss": 0.1337,
+      "step": 3198
+    },
+    {
+      "epoch": 4.337039045553145,
+      "grad_norm": 0.16770419478416443,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 3199
+    },
+    {
+      "epoch": 4.3383947939262475,
+      "grad_norm": 0.1906682550907135,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3200
+    },
+    {
+      "epoch": 4.339750542299349,
+      "grad_norm": 0.17314310371875763,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 3201
+    },
+    {
+      "epoch": 4.341106290672451,
+      "grad_norm": 0.1851377934217453,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 3202
+    },
+    {
+      "epoch": 4.342462039045553,
+      "grad_norm": 0.2056530863046646,
+      "learning_rate": 0.0001,
+      "loss": 0.155,
+      "step": 3203
+    },
+    {
+      "epoch": 4.343817787418655,
+      "grad_norm": 0.17806705832481384,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3204
+    },
+    {
+      "epoch": 4.345173535791757,
+      "grad_norm": 0.1901869773864746,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 3205
+    },
+    {
+      "epoch": 4.346529284164859,
+      "grad_norm": 0.17157796025276184,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 3206
+    },
+    {
+      "epoch": 4.3478850325379605,
+      "grad_norm": 0.17560768127441406,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 3207
+    },
+    {
+      "epoch": 4.349240780911063,
+      "grad_norm": 0.16859450936317444,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3208
+    },
+    {
+      "epoch": 4.350596529284164,
+      "grad_norm": 0.1756477952003479,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 3209
+    },
+    {
+      "epoch": 4.351952277657267,
+      "grad_norm": 0.18380112946033478,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3210
+    },
+    {
+      "epoch": 4.353308026030369,
+      "grad_norm": 0.18705224990844727,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 3211
+    },
+    {
+      "epoch": 4.354663774403471,
+      "grad_norm": 0.23231928050518036,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 3212
+    },
+    {
+      "epoch": 4.356019522776573,
+      "grad_norm": 0.18645626306533813,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 3213
+    },
+    {
+      "epoch": 4.3573752711496745,
+      "grad_norm": 0.16493472456932068,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 3214
+    },
+    {
+      "epoch": 4.358731019522777,
+      "grad_norm": 0.18472078442573547,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 3215
+    },
+    {
+      "epoch": 4.360086767895878,
+      "grad_norm": 0.1856292188167572,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 3216
+    },
+    {
+      "epoch": 4.361442516268981,
+      "grad_norm": 0.16342054307460785,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 3217
+    },
+    {
+      "epoch": 4.362798264642082,
+      "grad_norm": 0.1711912751197815,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 3218
+    },
+    {
+      "epoch": 4.364154013015185,
+      "grad_norm": 0.19055688381195068,
+      "learning_rate": 0.0001,
+      "loss": 0.1563,
+      "step": 3219
+    },
+    {
+      "epoch": 4.365509761388286,
+      "grad_norm": 0.2573765218257904,
+      "learning_rate": 0.0001,
+      "loss": 0.1342,
+      "step": 3220
+    },
+    {
+      "epoch": 4.366865509761388,
+      "grad_norm": 0.17054328322410583,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3221
+    },
+    {
+      "epoch": 4.36822125813449,
+      "grad_norm": 0.17018720507621765,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 3222
+    },
+    {
+      "epoch": 4.369577006507592,
+      "grad_norm": 0.1550099402666092,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 3223
+    },
+    {
+      "epoch": 4.370932754880694,
+      "grad_norm": 0.14917075634002686,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 3224
+    },
+    {
+      "epoch": 4.372288503253796,
+      "grad_norm": 0.18531842529773712,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 3225
+    },
+    {
+      "epoch": 4.373644251626898,
+      "grad_norm": 0.18064923584461212,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 3226
+    },
+    {
+      "epoch": 4.375,
+      "grad_norm": 0.17196619510650635,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 3227
+    },
+    {
+      "epoch": 4.376355748373102,
+      "grad_norm": 0.16069740056991577,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 3228
+    },
+    {
+      "epoch": 4.377711496746204,
+      "grad_norm": 0.17057248950004578,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3229
+    },
+    {
+      "epoch": 4.379067245119306,
+      "grad_norm": 0.1679837852716446,
+      "learning_rate": 0.0001,
+      "loss": 0.1439,
+      "step": 3230
+    },
+    {
+      "epoch": 4.380422993492408,
+      "grad_norm": 0.16454172134399414,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 3231
+    },
+    {
+      "epoch": 4.38177874186551,
+      "grad_norm": 0.19864045083522797,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 3232
+    },
+    {
+      "epoch": 4.383134490238612,
+      "grad_norm": 0.1739833950996399,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 3233
+    },
+    {
+      "epoch": 4.384490238611714,
+      "grad_norm": 0.19095352292060852,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 3234
+    },
+    {
+      "epoch": 4.385845986984815,
+      "grad_norm": 0.15727178752422333,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3235
+    },
+    {
+      "epoch": 4.387201735357918,
+      "grad_norm": 0.18433906137943268,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 3236
+    },
+    {
+      "epoch": 4.388557483731019,
+      "grad_norm": 0.1907864212989807,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 3237
+    },
+    {
+      "epoch": 4.389913232104122,
+      "grad_norm": 0.187800794839859,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 3238
+    },
+    {
+      "epoch": 4.391268980477223,
+      "grad_norm": 0.16032393276691437,
+      "learning_rate": 0.0001,
+      "loss": 0.148,
+      "step": 3239
+    },
+    {
+      "epoch": 4.3926247288503255,
+      "grad_norm": 0.1793687641620636,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3240
+    },
+    {
+      "epoch": 4.393980477223427,
+      "grad_norm": 0.208247572183609,
+      "learning_rate": 0.0001,
+      "loss": 0.1489,
+      "step": 3241
+    },
+    {
+      "epoch": 4.395336225596529,
+      "grad_norm": 0.17244546115398407,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 3242
+    },
+    {
+      "epoch": 4.396691973969631,
+      "grad_norm": 0.1959066092967987,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 3243
+    },
+    {
+      "epoch": 4.398047722342733,
+      "grad_norm": 0.16346968710422516,
+      "learning_rate": 0.0001,
+      "loss": 0.1401,
+      "step": 3244
+    },
+    {
+      "epoch": 4.399403470715836,
+      "grad_norm": 0.21487069129943848,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 3245
+    },
+    {
+      "epoch": 4.400759219088937,
+      "grad_norm": 0.21206115186214447,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 3246
+    },
+    {
+      "epoch": 4.4021149674620395,
+      "grad_norm": 0.1916428655385971,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 3247
+    },
+    {
+      "epoch": 4.403470715835141,
+      "grad_norm": 0.1867111325263977,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 3248
+    },
+    {
+      "epoch": 4.404826464208243,
+      "grad_norm": 0.16039283573627472,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 3249
+    },
+    {
+      "epoch": 4.406182212581345,
+      "grad_norm": 0.18900015950202942,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 3250
+    },
+    {
+      "epoch": 4.407537960954447,
+      "grad_norm": 0.16531610488891602,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 3251
+    },
+    {
+      "epoch": 4.408893709327549,
+      "grad_norm": 0.16516783833503723,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 3252
+    },
+    {
+      "epoch": 4.410249457700651,
+      "grad_norm": 0.21621641516685486,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 3253
+    },
+    {
+      "epoch": 4.4116052060737525,
+      "grad_norm": 0.16775530576705933,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 3254
+    },
+    {
+      "epoch": 4.412960954446855,
+      "grad_norm": 0.1821138858795166,
+      "learning_rate": 0.0001,
+      "loss": 0.1724,
+      "step": 3255
+    },
+    {
+      "epoch": 4.414316702819956,
+      "grad_norm": 0.15448616445064545,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 3256
+    },
+    {
+      "epoch": 4.415672451193059,
+      "grad_norm": 0.15979130566120148,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 3257
+    },
+    {
+      "epoch": 4.41702819956616,
+      "grad_norm": 0.16981519758701324,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 3258
+    },
+    {
+      "epoch": 4.418383947939263,
+      "grad_norm": 0.17391841113567352,
+      "learning_rate": 0.0001,
+      "loss": 0.1539,
+      "step": 3259
+    },
+    {
+      "epoch": 4.419739696312364,
+      "grad_norm": 0.19012778997421265,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3260
+    },
+    {
+      "epoch": 4.4210954446854664,
+      "grad_norm": 0.17741554975509644,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 3261
+    },
+    {
+      "epoch": 4.422451193058568,
+      "grad_norm": 0.17173880338668823,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 3262
+    },
+    {
+      "epoch": 4.42380694143167,
+      "grad_norm": 0.15365886688232422,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3263
+    },
+    {
+      "epoch": 4.425162689804772,
+      "grad_norm": 0.16097646951675415,
+      "learning_rate": 0.0001,
+      "loss": 0.1314,
+      "step": 3264
+    },
+    {
+      "epoch": 4.426518438177874,
+      "grad_norm": 0.16785593330860138,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 3265
+    },
+    {
+      "epoch": 4.4278741865509765,
+      "grad_norm": 0.1696942001581192,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3266
+    },
+    {
+      "epoch": 4.429229934924078,
+      "grad_norm": 0.21592114865779877,
+      "learning_rate": 0.0001,
+      "loss": 0.1615,
+      "step": 3267
+    },
+    {
+      "epoch": 4.43058568329718,
+      "grad_norm": 0.17126482725143433,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3268
+    },
+    {
+      "epoch": 4.431941431670282,
+      "grad_norm": 0.16739684343338013,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 3269
+    },
+    {
+      "epoch": 4.433297180043384,
+      "grad_norm": 0.17433249950408936,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 3270
+    },
+    {
+      "epoch": 4.434652928416486,
+      "grad_norm": 0.2067904770374298,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 3271
+    },
+    {
+      "epoch": 4.436008676789588,
+      "grad_norm": 0.17751920223236084,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 3272
+    },
+    {
+      "epoch": 4.43736442516269,
+      "grad_norm": 0.19393254816532135,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 3273
+    },
+    {
+      "epoch": 4.438720173535792,
+      "grad_norm": 0.16970616579055786,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 3274
+    },
+    {
+      "epoch": 4.440075921908893,
+      "grad_norm": 0.16536763310432434,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 3275
+    },
+    {
+      "epoch": 4.441431670281996,
+      "grad_norm": 0.20129263401031494,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3276
+    },
+    {
+      "epoch": 4.442787418655097,
+      "grad_norm": 0.16185995936393738,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 3277
+    },
+    {
+      "epoch": 4.4441431670282,
+      "grad_norm": 0.18997161090373993,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 3278
+    },
+    {
+      "epoch": 4.445498915401301,
+      "grad_norm": 0.19148215651512146,
+      "learning_rate": 0.0001,
+      "loss": 0.1388,
+      "step": 3279
+    },
+    {
+      "epoch": 4.4468546637744035,
+      "grad_norm": 0.19252318143844604,
+      "learning_rate": 0.0001,
+      "loss": 0.1478,
+      "step": 3280
+    },
+    {
+      "epoch": 4.448210412147505,
+      "grad_norm": 0.18372027575969696,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 3281
+    },
+    {
+      "epoch": 4.449566160520607,
+      "grad_norm": 0.18284083902835846,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3282
+    },
+    {
+      "epoch": 4.45092190889371,
+      "grad_norm": 0.22145332396030426,
+      "learning_rate": 0.0001,
+      "loss": 0.153,
+      "step": 3283
+    },
+    {
+      "epoch": 4.452277657266811,
+      "grad_norm": 0.18066543340682983,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 3284
+    },
+    {
+      "epoch": 4.453633405639914,
+      "grad_norm": 0.16980405151844025,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 3285
+    },
+    {
+      "epoch": 4.454989154013015,
+      "grad_norm": 0.15454034507274628,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 3286
+    },
+    {
+      "epoch": 4.4563449023861175,
+      "grad_norm": 0.1893301010131836,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 3287
+    },
+    {
+      "epoch": 4.457700650759219,
+      "grad_norm": 0.20530694723129272,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 3288
+    },
+    {
+      "epoch": 4.459056399132321,
+      "grad_norm": 0.171835795044899,
+      "learning_rate": 0.0001,
+      "loss": 0.1472,
+      "step": 3289
+    },
+    {
+      "epoch": 4.460412147505423,
+      "grad_norm": 0.3054157495498657,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 3290
+    },
+    {
+      "epoch": 4.461767895878525,
+      "grad_norm": 0.1811966747045517,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 3291
+    },
+    {
+      "epoch": 4.463123644251627,
+      "grad_norm": 0.18700146675109863,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 3292
+    },
+    {
+      "epoch": 4.464479392624729,
+      "grad_norm": 0.17815159261226654,
+      "learning_rate": 0.0001,
+      "loss": 0.1453,
+      "step": 3293
+    },
+    {
+      "epoch": 4.4658351409978305,
+      "grad_norm": 0.18773239850997925,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3294
+    },
+    {
+      "epoch": 4.467190889370933,
+      "grad_norm": 0.17627693712711334,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 3295
+    },
+    {
+      "epoch": 4.468546637744034,
+      "grad_norm": 0.16243170201778412,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 3296
+    },
+    {
+      "epoch": 4.469902386117137,
+      "grad_norm": 0.1708793044090271,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 3297
+    },
+    {
+      "epoch": 4.471258134490238,
+      "grad_norm": 0.16761872172355652,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 3298
+    },
+    {
+      "epoch": 4.472613882863341,
+      "grad_norm": 0.19431306421756744,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 3299
+    },
+    {
+      "epoch": 4.473969631236443,
+      "grad_norm": 0.1837160438299179,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3300
+    },
+    {
+      "epoch": 4.4753253796095445,
+      "grad_norm": 0.19466720521450043,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 3301
+    },
+    {
+      "epoch": 4.476681127982647,
+      "grad_norm": 0.18154746294021606,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3302
+    },
+    {
+      "epoch": 4.478036876355748,
+      "grad_norm": 0.1561909019947052,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 3303
+    },
+    {
+      "epoch": 4.479392624728851,
+      "grad_norm": 0.17040710151195526,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 3304
+    },
+    {
+      "epoch": 4.480748373101952,
+      "grad_norm": 0.1682850569486618,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 3305
+    },
+    {
+      "epoch": 4.4821041214750545,
+      "grad_norm": 0.1879897266626358,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3306
+    },
+    {
+      "epoch": 4.483459869848156,
+      "grad_norm": 0.16870401799678802,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 3307
+    },
+    {
+      "epoch": 4.484815618221258,
+      "grad_norm": 0.17076797783374786,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 3308
+    },
+    {
+      "epoch": 4.48617136659436,
+      "grad_norm": 0.18567629158496857,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 3309
+    },
+    {
+      "epoch": 4.487527114967462,
+      "grad_norm": 0.16917207837104797,
+      "learning_rate": 0.0001,
+      "loss": 0.1437,
+      "step": 3310
+    },
+    {
+      "epoch": 4.488882863340564,
+      "grad_norm": 0.16012635827064514,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 3311
+    },
+    {
+      "epoch": 4.490238611713666,
+      "grad_norm": 0.14483149349689484,
+      "learning_rate": 0.0001,
+      "loss": 0.1359,
+      "step": 3312
+    },
+    {
+      "epoch": 4.491594360086768,
+      "grad_norm": 0.15184737741947174,
+      "learning_rate": 0.0001,
+      "loss": 0.1524,
+      "step": 3313
+    },
+    {
+      "epoch": 4.49295010845987,
+      "grad_norm": 0.17364099621772766,
+      "learning_rate": 0.0001,
+      "loss": 0.1637,
+      "step": 3314
+    },
+    {
+      "epoch": 4.4943058568329715,
+      "grad_norm": 0.178785502910614,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3315
+    },
+    {
+      "epoch": 4.495661605206074,
+      "grad_norm": 0.1760556697845459,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3316
+    },
+    {
+      "epoch": 4.497017353579175,
+      "grad_norm": 0.19433532655239105,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 3317
+    },
+    {
+      "epoch": 4.498373101952278,
+      "grad_norm": 0.15920431911945343,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 3318
+    },
+    {
+      "epoch": 4.499728850325379,
+      "grad_norm": 0.17181362211704254,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3319
+    },
+    {
+      "epoch": 4.5010845986984815,
+      "grad_norm": 0.16277746856212616,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 3320
+    },
+    {
+      "epoch": 4.502440347071584,
+      "grad_norm": 0.16528603434562683,
+      "learning_rate": 0.0001,
+      "loss": 0.1271,
+      "step": 3321
+    },
+    {
+      "epoch": 4.503796095444685,
+      "grad_norm": 0.18113455176353455,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 3322
+    },
+    {
+      "epoch": 4.505151843817788,
+      "grad_norm": 0.2011893093585968,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3323
+    },
+    {
+      "epoch": 4.506507592190889,
+      "grad_norm": 0.19137655198574066,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 3324
+    },
+    {
+      "epoch": 4.507863340563992,
+      "grad_norm": 0.1821868121623993,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 3325
+    },
+    {
+      "epoch": 4.509219088937093,
+      "grad_norm": 0.18148669600486755,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3326
+    },
+    {
+      "epoch": 4.5105748373101955,
+      "grad_norm": 0.18293723464012146,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 3327
+    },
+    {
+      "epoch": 4.511930585683297,
+      "grad_norm": 0.17502564191818237,
+      "learning_rate": 0.0001,
+      "loss": 0.147,
+      "step": 3328
+    },
+    {
+      "epoch": 4.513286334056399,
+      "grad_norm": 0.19371357560157776,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 3329
+    },
+    {
+      "epoch": 4.514642082429501,
+      "grad_norm": 0.14756040275096893,
+      "learning_rate": 0.0001,
+      "loss": 0.1268,
+      "step": 3330
+    },
+    {
+      "epoch": 4.515997830802603,
+      "grad_norm": 0.19293458759784698,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 3331
+    },
+    {
+      "epoch": 4.517353579175705,
+      "grad_norm": 0.17000806331634521,
+      "learning_rate": 0.0001,
+      "loss": 0.1338,
+      "step": 3332
+    },
+    {
+      "epoch": 4.518709327548807,
+      "grad_norm": 0.17879806458950043,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 3333
+    },
+    {
+      "epoch": 4.5200650759219085,
+      "grad_norm": 0.16809603571891785,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 3334
+    },
+    {
+      "epoch": 4.521420824295011,
+      "grad_norm": 0.1683455854654312,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 3335
+    },
+    {
+      "epoch": 4.522776572668112,
+      "grad_norm": 0.15783575177192688,
+      "learning_rate": 0.0001,
+      "loss": 0.1399,
+      "step": 3336
+    },
+    {
+      "epoch": 4.524132321041215,
+      "grad_norm": 0.17252196371555328,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 3337
+    },
+    {
+      "epoch": 4.525488069414317,
+      "grad_norm": 0.1648230403661728,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3338
+    },
+    {
+      "epoch": 4.526843817787419,
+      "grad_norm": 0.19594421982765198,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 3339
+    },
+    {
+      "epoch": 4.528199566160521,
+      "grad_norm": 0.1899336874485016,
+      "learning_rate": 0.0001,
+      "loss": 0.1452,
+      "step": 3340
+    },
+    {
+      "epoch": 4.5295553145336225,
+      "grad_norm": 0.1642763912677765,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 3341
+    },
+    {
+      "epoch": 4.530911062906725,
+      "grad_norm": 0.18766261637210846,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 3342
+    },
+    {
+      "epoch": 4.532266811279826,
+      "grad_norm": 0.1994369626045227,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 3343
+    },
+    {
+      "epoch": 4.533622559652929,
+      "grad_norm": 0.17733345925807953,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 3344
+    },
+    {
+      "epoch": 4.53497830802603,
+      "grad_norm": 0.1821119636297226,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 3345
+    },
+    {
+      "epoch": 4.536334056399133,
+      "grad_norm": 0.16333520412445068,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 3346
+    },
+    {
+      "epoch": 4.537689804772234,
+      "grad_norm": 0.15246249735355377,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 3347
+    },
+    {
+      "epoch": 4.539045553145336,
+      "grad_norm": 0.17232660949230194,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 3348
+    },
+    {
+      "epoch": 4.540401301518438,
+      "grad_norm": 0.20209482312202454,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 3349
+    },
+    {
+      "epoch": 4.54175704989154,
+      "grad_norm": 0.16611602902412415,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 3350
+    },
+    {
+      "epoch": 4.543112798264642,
+      "grad_norm": 0.17314590513706207,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 3351
+    },
+    {
+      "epoch": 4.544468546637744,
+      "grad_norm": 0.201426163315773,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3352
+    },
+    {
+      "epoch": 4.545824295010846,
+      "grad_norm": 0.1743067055940628,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 3353
+    },
+    {
+      "epoch": 4.547180043383948,
+      "grad_norm": 0.1711207926273346,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 3354
+    },
+    {
+      "epoch": 4.54853579175705,
+      "grad_norm": 0.18258152902126312,
+      "learning_rate": 0.0001,
+      "loss": 0.1506,
+      "step": 3355
+    },
+    {
+      "epoch": 4.549891540130152,
+      "grad_norm": 0.22108352184295654,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 3356
+    },
+    {
+      "epoch": 4.551247288503253,
+      "grad_norm": 0.20009806752204895,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 3357
+    },
+    {
+      "epoch": 4.552603036876356,
+      "grad_norm": 0.16064049303531647,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 3358
+    },
+    {
+      "epoch": 4.553958785249458,
+      "grad_norm": 0.17387248575687408,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 3359
+    },
+    {
+      "epoch": 4.55531453362256,
+      "grad_norm": 0.2855769395828247,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 3360
+    },
+    {
+      "epoch": 4.556670281995662,
+      "grad_norm": 0.17106030881404877,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 3361
+    },
+    {
+      "epoch": 4.558026030368763,
+      "grad_norm": 0.19091838598251343,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 3362
+    },
+    {
+      "epoch": 4.559381778741866,
+      "grad_norm": 0.18551349639892578,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 3363
+    },
+    {
+      "epoch": 4.560737527114967,
+      "grad_norm": 0.1667039692401886,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3364
+    },
+    {
+      "epoch": 4.56209327548807,
+      "grad_norm": 0.1821744590997696,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 3365
+    },
+    {
+      "epoch": 4.563449023861171,
+      "grad_norm": 0.15762914717197418,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 3366
+    },
+    {
+      "epoch": 4.5648047722342735,
+      "grad_norm": 0.16440901160240173,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3367
+    },
+    {
+      "epoch": 4.566160520607375,
+      "grad_norm": 0.1936037391424179,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 3368
+    },
+    {
+      "epoch": 4.567516268980477,
+      "grad_norm": 0.19556547701358795,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 3369
+    },
+    {
+      "epoch": 4.568872017353579,
+      "grad_norm": 0.1439594328403473,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 3370
+    },
+    {
+      "epoch": 4.570227765726681,
+      "grad_norm": 0.16945002973079681,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3371
+    },
+    {
+      "epoch": 4.571583514099783,
+      "grad_norm": 0.20213951170444489,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 3372
+    },
+    {
+      "epoch": 4.572939262472885,
+      "grad_norm": 0.1705339401960373,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "step": 3373
+    },
+    {
+      "epoch": 4.5742950108459866,
+      "grad_norm": 0.17926301062107086,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 3374
+    },
+    {
+      "epoch": 4.575650759219089,
+      "grad_norm": 0.19325128197669983,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3375
+    },
+    {
+      "epoch": 4.577006507592191,
+      "grad_norm": 0.18984535336494446,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 3376
+    },
+    {
+      "epoch": 4.578362255965293,
+      "grad_norm": 0.17717087268829346,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 3377
+    },
+    {
+      "epoch": 4.579718004338395,
+      "grad_norm": 0.15235331654548645,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 3378
+    },
+    {
+      "epoch": 4.581073752711497,
+      "grad_norm": 0.1869913637638092,
+      "learning_rate": 0.0001,
+      "loss": 0.1526,
+      "step": 3379
+    },
+    {
+      "epoch": 4.582429501084599,
+      "grad_norm": 0.16048920154571533,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 3380
+    },
+    {
+      "epoch": 4.5837852494577005,
+      "grad_norm": 0.16753724217414856,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 3381
+    },
+    {
+      "epoch": 4.585140997830803,
+      "grad_norm": 0.19183646142482758,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3382
+    },
+    {
+      "epoch": 4.586496746203904,
+      "grad_norm": 0.20102903246879578,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 3383
+    },
+    {
+      "epoch": 4.587852494577007,
+      "grad_norm": 0.18148520588874817,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 3384
+    },
+    {
+      "epoch": 4.589208242950108,
+      "grad_norm": 0.1939910352230072,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 3385
+    },
+    {
+      "epoch": 4.590563991323211,
+      "grad_norm": 0.16409459710121155,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 3386
+    },
+    {
+      "epoch": 4.591919739696312,
+      "grad_norm": 0.17499399185180664,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 3387
+    },
+    {
+      "epoch": 4.593275488069414,
+      "grad_norm": 0.16636013984680176,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 3388
+    },
+    {
+      "epoch": 4.594631236442516,
+      "grad_norm": 0.19517302513122559,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 3389
+    },
+    {
+      "epoch": 4.595986984815618,
+      "grad_norm": 0.1995604783296585,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 3390
+    },
+    {
+      "epoch": 4.59734273318872,
+      "grad_norm": 0.17369505763053894,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 3391
+    },
+    {
+      "epoch": 4.598698481561822,
+      "grad_norm": 0.15999364852905273,
+      "learning_rate": 0.0001,
+      "loss": 0.1363,
+      "step": 3392
+    },
+    {
+      "epoch": 4.6000542299349245,
+      "grad_norm": 0.22685717046260834,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 3393
+    },
+    {
+      "epoch": 4.601409978308026,
+      "grad_norm": 0.17896604537963867,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3394
+    },
+    {
+      "epoch": 4.6027657266811275,
+      "grad_norm": 0.16349247097969055,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 3395
+    },
+    {
+      "epoch": 4.60412147505423,
+      "grad_norm": 0.17090563476085663,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 3396
+    },
+    {
+      "epoch": 4.605477223427332,
+      "grad_norm": 0.17118322849273682,
+      "learning_rate": 0.0001,
+      "loss": 0.1415,
+      "step": 3397
+    },
+    {
+      "epoch": 4.606832971800434,
+      "grad_norm": 0.22109238803386688,
+      "learning_rate": 0.0001,
+      "loss": 0.1642,
+      "step": 3398
+    },
+    {
+      "epoch": 4.608188720173536,
+      "grad_norm": 0.17618823051452637,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 3399
+    },
+    {
+      "epoch": 4.609544468546638,
+      "grad_norm": 0.16229498386383057,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 3400
+    },
+    {
+      "epoch": 4.61090021691974,
+      "grad_norm": 0.196502223610878,
+      "learning_rate": 0.0001,
+      "loss": 0.1451,
+      "step": 3401
+    },
+    {
+      "epoch": 4.612255965292841,
+      "grad_norm": 0.18354085087776184,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 3402
+    },
+    {
+      "epoch": 4.613611713665944,
+      "grad_norm": 0.19125568866729736,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 3403
+    },
+    {
+      "epoch": 4.614967462039045,
+      "grad_norm": 0.16297075152397156,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 3404
+    },
+    {
+      "epoch": 4.616323210412148,
+      "grad_norm": 0.16922065615653992,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 3405
+    },
+    {
+      "epoch": 4.617678958785249,
+      "grad_norm": 0.17112493515014648,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3406
+    },
+    {
+      "epoch": 4.6190347071583515,
+      "grad_norm": 0.19858039915561676,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 3407
+    },
+    {
+      "epoch": 4.620390455531453,
+      "grad_norm": 0.1747017204761505,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 3408
+    },
+    {
+      "epoch": 4.621746203904555,
+      "grad_norm": 0.17435871064662933,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 3409
+    },
+    {
+      "epoch": 4.623101952277658,
+      "grad_norm": 0.15747769176959991,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3410
+    },
+    {
+      "epoch": 4.624457700650759,
+      "grad_norm": 0.20545510947704315,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 3411
+    },
+    {
+      "epoch": 4.625813449023861,
+      "grad_norm": 0.18162085115909576,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 3412
+    },
+    {
+      "epoch": 4.627169197396963,
+      "grad_norm": 0.17437314987182617,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3413
+    },
+    {
+      "epoch": 4.6285249457700655,
+      "grad_norm": 0.1813751608133316,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 3414
+    },
+    {
+      "epoch": 4.629880694143167,
+      "grad_norm": 0.19293993711471558,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 3415
+    },
+    {
+      "epoch": 4.631236442516269,
+      "grad_norm": 0.1433936357498169,
+      "learning_rate": 0.0001,
+      "loss": 0.1255,
+      "step": 3416
+    },
+    {
+      "epoch": 4.632592190889371,
+      "grad_norm": 0.17155782878398895,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 3417
+    },
+    {
+      "epoch": 4.633947939262473,
+      "grad_norm": 0.18624886870384216,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 3418
+    },
+    {
+      "epoch": 4.635303687635575,
+      "grad_norm": 0.16862282156944275,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3419
+    },
+    {
+      "epoch": 4.636659436008677,
+      "grad_norm": 0.20334602892398834,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 3420
+    },
+    {
+      "epoch": 4.6380151843817785,
+      "grad_norm": 0.19085368514060974,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 3421
+    },
+    {
+      "epoch": 4.639370932754881,
+      "grad_norm": 0.16873827576637268,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 3422
+    },
+    {
+      "epoch": 4.640726681127982,
+      "grad_norm": 0.17224439978599548,
+      "learning_rate": 0.0001,
+      "loss": 0.1451,
+      "step": 3423
+    },
+    {
+      "epoch": 4.642082429501085,
+      "grad_norm": 0.19711093604564667,
+      "learning_rate": 0.0001,
+      "loss": 0.1565,
+      "step": 3424
+    },
+    {
+      "epoch": 4.643438177874186,
+      "grad_norm": 0.17229218780994415,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3425
+    },
+    {
+      "epoch": 4.644793926247289,
+      "grad_norm": 0.22238600254058838,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 3426
+    },
+    {
+      "epoch": 4.64614967462039,
+      "grad_norm": 0.1649092733860016,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3427
+    },
+    {
+      "epoch": 4.6475054229934925,
+      "grad_norm": 0.23536570370197296,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 3428
+    },
+    {
+      "epoch": 4.648861171366594,
+      "grad_norm": 0.21269582211971283,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 3429
+    },
+    {
+      "epoch": 4.650216919739696,
+      "grad_norm": 0.17638219892978668,
+      "learning_rate": 0.0001,
+      "loss": 0.1399,
+      "step": 3430
+    },
+    {
+      "epoch": 4.651572668112799,
+      "grad_norm": 0.15687817335128784,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "step": 3431
+    },
+    {
+      "epoch": 4.6529284164859,
+      "grad_norm": 0.17040227353572845,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3432
+    },
+    {
+      "epoch": 4.6542841648590025,
+      "grad_norm": 0.1859876811504364,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 3433
+    },
+    {
+      "epoch": 4.655639913232104,
+      "grad_norm": 0.16652995347976685,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 3434
+    },
+    {
+      "epoch": 4.656995661605206,
+      "grad_norm": 0.1726629137992859,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 3435
+    },
+    {
+      "epoch": 4.658351409978308,
+      "grad_norm": 0.17365217208862305,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 3436
+    },
+    {
+      "epoch": 4.65970715835141,
+      "grad_norm": 0.1618271917104721,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 3437
+    },
+    {
+      "epoch": 4.661062906724512,
+      "grad_norm": 0.18575508892536163,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 3438
+    },
+    {
+      "epoch": 4.662418655097614,
+      "grad_norm": 0.15571406483650208,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 3439
+    },
+    {
+      "epoch": 4.663774403470716,
+      "grad_norm": 0.18498443067073822,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 3440
+    },
+    {
+      "epoch": 4.665130151843818,
+      "grad_norm": 0.16301600635051727,
+      "learning_rate": 0.0001,
+      "loss": 0.146,
+      "step": 3441
+    },
+    {
+      "epoch": 4.6664859002169194,
+      "grad_norm": 0.1723099797964096,
+      "learning_rate": 0.0001,
+      "loss": 0.1443,
+      "step": 3442
+    },
+    {
+      "epoch": 4.667841648590022,
+      "grad_norm": 0.16776974499225616,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 3443
+    },
+    {
+      "epoch": 4.669197396963123,
+      "grad_norm": 0.16515573859214783,
+      "learning_rate": 0.0001,
+      "loss": 0.1682,
+      "step": 3444
+    },
+    {
+      "epoch": 4.670553145336226,
+      "grad_norm": 0.17750214040279388,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 3445
+    },
+    {
+      "epoch": 4.671908893709327,
+      "grad_norm": 0.1787509024143219,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 3446
+    },
+    {
+      "epoch": 4.6732646420824295,
+      "grad_norm": 0.1701103001832962,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 3447
+    },
+    {
+      "epoch": 4.674620390455532,
+      "grad_norm": 0.17335344851016998,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 3448
+    },
+    {
+      "epoch": 4.675976138828633,
+      "grad_norm": 0.15562671422958374,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 3449
+    },
+    {
+      "epoch": 4.677331887201735,
+      "grad_norm": 0.19219082593917847,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 3450
+    },
+    {
+      "epoch": 4.678687635574837,
+      "grad_norm": 0.1920401155948639,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 3451
+    },
+    {
+      "epoch": 4.68004338394794,
+      "grad_norm": 0.16627298295497894,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3452
+    },
+    {
+      "epoch": 4.681399132321041,
+      "grad_norm": 0.21542319655418396,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 3453
+    },
+    {
+      "epoch": 4.6827548806941435,
+      "grad_norm": 0.18504860997200012,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 3454
+    },
+    {
+      "epoch": 4.684110629067245,
+      "grad_norm": 0.19127188622951508,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 3455
+    },
+    {
+      "epoch": 4.685466377440347,
+      "grad_norm": 0.17136254906654358,
+      "learning_rate": 0.0001,
+      "loss": 0.1511,
+      "step": 3456
+    },
+    {
+      "epoch": 4.686822125813449,
+      "grad_norm": 0.18820467591285706,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 3457
+    },
+    {
+      "epoch": 4.688177874186551,
+      "grad_norm": 0.182331845164299,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 3458
+    },
+    {
+      "epoch": 4.689533622559653,
+      "grad_norm": 0.1661399006843567,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3459
+    },
+    {
+      "epoch": 4.690889370932755,
+      "grad_norm": 0.2126038670539856,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 3460
+    },
+    {
+      "epoch": 4.6922451193058565,
+      "grad_norm": 0.1822144091129303,
+      "learning_rate": 0.0001,
+      "loss": 0.1399,
+      "step": 3461
+    },
+    {
+      "epoch": 4.693600867678959,
+      "grad_norm": 0.20370501279830933,
+      "learning_rate": 0.0001,
+      "loss": 0.1553,
+      "step": 3462
+    },
+    {
+      "epoch": 4.69495661605206,
+      "grad_norm": 0.19539442658424377,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 3463
+    },
+    {
+      "epoch": 4.696312364425163,
+      "grad_norm": 0.19825349748134613,
+      "learning_rate": 0.0001,
+      "loss": 0.158,
+      "step": 3464
+    },
+    {
+      "epoch": 4.697668112798265,
+      "grad_norm": 0.17113910615444183,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 3465
+    },
+    {
+      "epoch": 4.699023861171367,
+      "grad_norm": 0.16679087281227112,
+      "learning_rate": 0.0001,
+      "loss": 0.1493,
+      "step": 3466
+    },
+    {
+      "epoch": 4.700379609544468,
+      "grad_norm": 0.20282718539237976,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 3467
+    },
+    {
+      "epoch": 4.7017353579175705,
+      "grad_norm": 0.17271924018859863,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3468
+    },
+    {
+      "epoch": 4.703091106290673,
+      "grad_norm": 0.21581482887268066,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 3469
+    },
+    {
+      "epoch": 4.704446854663774,
+      "grad_norm": 0.15045958757400513,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 3470
+    },
+    {
+      "epoch": 4.705802603036877,
+      "grad_norm": 0.19110536575317383,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 3471
+    },
+    {
+      "epoch": 4.707158351409978,
+      "grad_norm": 0.19086085259914398,
+      "learning_rate": 0.0001,
+      "loss": 0.1451,
+      "step": 3472
+    },
+    {
+      "epoch": 4.7085140997830806,
+      "grad_norm": 0.17367899417877197,
+      "learning_rate": 0.0001,
+      "loss": 0.1622,
+      "step": 3473
+    },
+    {
+      "epoch": 4.709869848156182,
+      "grad_norm": 0.1808701455593109,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 3474
+    },
+    {
+      "epoch": 4.711225596529284,
+      "grad_norm": 0.18992313742637634,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3475
+    },
+    {
+      "epoch": 4.712581344902386,
+      "grad_norm": 0.17572295665740967,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 3476
+    },
+    {
+      "epoch": 4.713937093275488,
+      "grad_norm": 0.15331797301769257,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 3477
+    },
+    {
+      "epoch": 4.71529284164859,
+      "grad_norm": 0.17957516014575958,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 3478
+    },
+    {
+      "epoch": 4.716648590021692,
+      "grad_norm": 0.16567547619342804,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 3479
+    },
+    {
+      "epoch": 4.718004338394794,
+      "grad_norm": 0.19090387225151062,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 3480
+    },
+    {
+      "epoch": 4.719360086767896,
+      "grad_norm": 0.1786288470029831,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3481
+    },
+    {
+      "epoch": 4.7207158351409975,
+      "grad_norm": 0.18172308802604675,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3482
+    },
+    {
+      "epoch": 4.7220715835141,
+      "grad_norm": 0.1887499839067459,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3483
+    },
+    {
+      "epoch": 4.723427331887201,
+      "grad_norm": 0.20102041959762573,
+      "learning_rate": 0.0001,
+      "loss": 0.1545,
+      "step": 3484
+    },
+    {
+      "epoch": 4.724783080260304,
+      "grad_norm": 0.20029442012310028,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 3485
+    },
+    {
+      "epoch": 4.726138828633406,
+      "grad_norm": 0.17953082919120789,
+      "learning_rate": 0.0001,
+      "loss": 0.1559,
+      "step": 3486
+    },
+    {
+      "epoch": 4.7274945770065075,
+      "grad_norm": 0.16086991131305695,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 3487
+    },
+    {
+      "epoch": 4.72885032537961,
+      "grad_norm": 0.15167208015918732,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "step": 3488
+    },
+    {
+      "epoch": 4.730206073752711,
+      "grad_norm": 0.14360961318016052,
+      "learning_rate": 0.0001,
+      "loss": 0.1238,
+      "step": 3489
+    },
+    {
+      "epoch": 4.731561822125814,
+      "grad_norm": 0.16596636176109314,
+      "learning_rate": 0.0001,
+      "loss": 0.1629,
+      "step": 3490
+    },
+    {
+      "epoch": 4.732917570498915,
+      "grad_norm": 0.14974261820316315,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 3491
+    },
+    {
+      "epoch": 4.734273318872018,
+      "grad_norm": 0.15755216777324677,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 3492
+    },
+    {
+      "epoch": 4.735629067245119,
+      "grad_norm": 0.21578851342201233,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 3493
+    },
+    {
+      "epoch": 4.7369848156182215,
+      "grad_norm": 0.1468820571899414,
+      "learning_rate": 0.0001,
+      "loss": 0.1322,
+      "step": 3494
+    },
+    {
+      "epoch": 4.738340563991323,
+      "grad_norm": 0.1658543199300766,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 3495
+    },
+    {
+      "epoch": 4.739696312364425,
+      "grad_norm": 0.16640399396419525,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 3496
+    },
+    {
+      "epoch": 4.741052060737527,
+      "grad_norm": 0.2138807624578476,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 3497
+    },
+    {
+      "epoch": 4.742407809110629,
+      "grad_norm": 0.2052323967218399,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3498
+    },
+    {
+      "epoch": 4.743763557483731,
+      "grad_norm": 0.1978110671043396,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3499
+    },
+    {
+      "epoch": 4.745119305856833,
+      "grad_norm": 0.16758909821510315,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3500
+    },
+    {
+      "epoch": 4.745119305856833,
+      "eval_loss": 0.18996892869472504,
+      "eval_runtime": 594.8254,
+      "eval_samples_per_second": 4.65,
+      "eval_steps_per_second": 1.163,
+      "step": 3500
+    },
+    {
+      "epoch": 4.7464750542299345,
+      "grad_norm": 0.1803526133298874,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 3501
+    },
+    {
+      "epoch": 4.747830802603037,
+      "grad_norm": 0.20030874013900757,
+      "learning_rate": 0.0001,
+      "loss": 0.1535,
+      "step": 3502
+    },
+    {
+      "epoch": 4.749186550976139,
+      "grad_norm": 0.1619904786348343,
+      "learning_rate": 0.0001,
+      "loss": 0.1338,
+      "step": 3503
+    },
+    {
+      "epoch": 4.750542299349241,
+      "grad_norm": 0.18657103180885315,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 3504
+    },
+    {
+      "epoch": 4.751898047722342,
+      "grad_norm": 0.1900104433298111,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3505
+    },
+    {
+      "epoch": 4.753253796095445,
+      "grad_norm": 0.17833976447582245,
+      "learning_rate": 0.0001,
+      "loss": 0.1645,
+      "step": 3506
+    },
+    {
+      "epoch": 4.754609544468547,
+      "grad_norm": 0.1625334769487381,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 3507
+    },
+    {
+      "epoch": 4.7559652928416485,
+      "grad_norm": 0.1539725512266159,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 3508
+    },
+    {
+      "epoch": 4.757321041214751,
+      "grad_norm": 0.1665370762348175,
+      "learning_rate": 0.0001,
+      "loss": 0.1445,
+      "step": 3509
+    },
+    {
+      "epoch": 4.758676789587852,
+      "grad_norm": 0.19304019212722778,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 3510
+    },
+    {
+      "epoch": 4.760032537960955,
+      "grad_norm": 0.17237302660942078,
+      "learning_rate": 0.0001,
+      "loss": 0.154,
+      "step": 3511
+    },
+    {
+      "epoch": 4.761388286334056,
+      "grad_norm": 0.1516687124967575,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 3512
+    },
+    {
+      "epoch": 4.762744034707159,
+      "grad_norm": 0.16323266923427582,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 3513
+    },
+    {
+      "epoch": 4.76409978308026,
+      "grad_norm": 0.1482943892478943,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 3514
+    },
+    {
+      "epoch": 4.765455531453362,
+      "grad_norm": 0.1698738932609558,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 3515
+    },
+    {
+      "epoch": 4.766811279826464,
+      "grad_norm": 0.17037464678287506,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 3516
+    },
+    {
+      "epoch": 4.768167028199566,
+      "grad_norm": 0.1612706035375595,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 3517
+    },
+    {
+      "epoch": 4.769522776572668,
+      "grad_norm": 0.17152543365955353,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 3518
+    },
+    {
+      "epoch": 4.77087852494577,
+      "grad_norm": 0.26533105969429016,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 3519
+    },
+    {
+      "epoch": 4.7722342733188725,
+      "grad_norm": 0.3484690189361572,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 3520
+    },
+    {
+      "epoch": 4.773590021691974,
+      "grad_norm": 0.20017659664154053,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 3521
+    },
+    {
+      "epoch": 4.7749457700650755,
+      "grad_norm": 0.19287154078483582,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 3522
+    },
+    {
+      "epoch": 4.776301518438178,
+      "grad_norm": 0.1786482334136963,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 3523
+    },
+    {
+      "epoch": 4.77765726681128,
+      "grad_norm": 0.17277711629867554,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 3524
+    },
+    {
+      "epoch": 4.779013015184382,
+      "grad_norm": 0.19284272193908691,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 3525
+    },
+    {
+      "epoch": 4.780368763557484,
+      "grad_norm": 0.1662152111530304,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3526
+    },
+    {
+      "epoch": 4.781724511930586,
+      "grad_norm": 0.21459072828292847,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 3527
+    },
+    {
+      "epoch": 4.783080260303688,
+      "grad_norm": 0.18564818799495697,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 3528
+    },
+    {
+      "epoch": 4.784436008676789,
+      "grad_norm": 0.19011226296424866,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 3529
+    },
+    {
+      "epoch": 4.785791757049892,
+      "grad_norm": 0.15489433705806732,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 3530
+    },
+    {
+      "epoch": 4.787147505422993,
+      "grad_norm": 0.16612376272678375,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 3531
+    },
+    {
+      "epoch": 4.788503253796096,
+      "grad_norm": 0.19624672830104828,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 3532
+    },
+    {
+      "epoch": 4.789859002169197,
+      "grad_norm": 0.2198767215013504,
+      "learning_rate": 0.0001,
+      "loss": 0.1624,
+      "step": 3533
+    },
+    {
+      "epoch": 4.7912147505422995,
+      "grad_norm": 0.2075909823179245,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 3534
+    },
+    {
+      "epoch": 4.792570498915401,
+      "grad_norm": 0.181952103972435,
+      "learning_rate": 0.0001,
+      "loss": 0.1412,
+      "step": 3535
+    },
+    {
+      "epoch": 4.793926247288503,
+      "grad_norm": 0.15723367035388947,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3536
+    },
+    {
+      "epoch": 4.795281995661605,
+      "grad_norm": 0.19536957144737244,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 3537
+    },
+    {
+      "epoch": 4.796637744034707,
+      "grad_norm": 0.16806697845458984,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3538
+    },
+    {
+      "epoch": 4.797993492407809,
+      "grad_norm": 0.1607080101966858,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3539
+    },
+    {
+      "epoch": 4.799349240780911,
+      "grad_norm": 0.16113168001174927,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 3540
+    },
+    {
+      "epoch": 4.8007049891540134,
+      "grad_norm": 0.20462468266487122,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 3541
+    },
+    {
+      "epoch": 4.802060737527115,
+      "grad_norm": 0.20951411128044128,
+      "learning_rate": 0.0001,
+      "loss": 0.152,
+      "step": 3542
+    },
+    {
+      "epoch": 4.803416485900217,
+      "grad_norm": 0.22279970347881317,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 3543
+    },
+    {
+      "epoch": 4.804772234273319,
+      "grad_norm": 0.180355504155159,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 3544
+    },
+    {
+      "epoch": 4.806127982646421,
+      "grad_norm": 0.18398134410381317,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 3545
+    },
+    {
+      "epoch": 4.807483731019523,
+      "grad_norm": 0.15936703979969025,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 3546
+    },
+    {
+      "epoch": 4.808839479392625,
+      "grad_norm": 0.1565820872783661,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 3547
+    },
+    {
+      "epoch": 4.8101952277657265,
+      "grad_norm": 0.18602265417575836,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3548
+    },
+    {
+      "epoch": 4.811550976138829,
+      "grad_norm": 0.1894860416650772,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 3549
+    },
+    {
+      "epoch": 4.81290672451193,
+      "grad_norm": 0.20733734965324402,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 3550
+    },
+    {
+      "epoch": 4.814262472885033,
+      "grad_norm": 0.16897177696228027,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 3551
+    },
+    {
+      "epoch": 4.815618221258134,
+      "grad_norm": 0.1810941845178604,
+      "learning_rate": 0.0001,
+      "loss": 0.1715,
+      "step": 3552
+    },
+    {
+      "epoch": 4.816973969631237,
+      "grad_norm": 0.17076313495635986,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 3553
+    },
+    {
+      "epoch": 4.818329718004338,
+      "grad_norm": 0.19699819386005402,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 3554
+    },
+    {
+      "epoch": 4.81968546637744,
+      "grad_norm": 0.1822241097688675,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 3555
+    },
+    {
+      "epoch": 4.821041214750542,
+      "grad_norm": 0.1974359005689621,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 3556
+    },
+    {
+      "epoch": 4.822396963123644,
+      "grad_norm": 0.17581897974014282,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 3557
+    },
+    {
+      "epoch": 4.823752711496747,
+      "grad_norm": 0.17086133360862732,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 3558
+    },
+    {
+      "epoch": 4.825108459869848,
+      "grad_norm": 0.19532613456249237,
+      "learning_rate": 0.0001,
+      "loss": 0.1661,
+      "step": 3559
+    },
+    {
+      "epoch": 4.82646420824295,
+      "grad_norm": 0.17742235958576202,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3560
+    },
+    {
+      "epoch": 4.827819956616052,
+      "grad_norm": 0.16471728682518005,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3561
+    },
+    {
+      "epoch": 4.829175704989154,
+      "grad_norm": 0.1573915183544159,
+      "learning_rate": 0.0001,
+      "loss": 0.1511,
+      "step": 3562
+    },
+    {
+      "epoch": 4.830531453362256,
+      "grad_norm": 0.15450486540794373,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 3563
+    },
+    {
+      "epoch": 4.831887201735358,
+      "grad_norm": 0.1593848168849945,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3564
+    },
+    {
+      "epoch": 4.83324295010846,
+      "grad_norm": 0.16488870978355408,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 3565
+    },
+    {
+      "epoch": 4.834598698481562,
+      "grad_norm": 0.18728254735469818,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 3566
+    },
+    {
+      "epoch": 4.835954446854664,
+      "grad_norm": 0.3224685490131378,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3567
+    },
+    {
+      "epoch": 4.837310195227766,
+      "grad_norm": 0.18077979981899261,
+      "learning_rate": 0.0001,
+      "loss": 0.1571,
+      "step": 3568
+    },
+    {
+      "epoch": 4.838665943600867,
+      "grad_norm": 0.1770423948764801,
+      "learning_rate": 0.0001,
+      "loss": 0.1552,
+      "step": 3569
+    },
+    {
+      "epoch": 4.84002169197397,
+      "grad_norm": 0.17360299825668335,
+      "learning_rate": 0.0001,
+      "loss": 0.1578,
+      "step": 3570
+    },
+    {
+      "epoch": 4.841377440347071,
+      "grad_norm": 0.17958727478981018,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 3571
+    },
+    {
+      "epoch": 4.842733188720174,
+      "grad_norm": 0.16022983193397522,
+      "learning_rate": 0.0001,
+      "loss": 0.1343,
+      "step": 3572
+    },
+    {
+      "epoch": 4.844088937093275,
+      "grad_norm": 0.1763676255941391,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 3573
+    },
+    {
+      "epoch": 4.8454446854663775,
+      "grad_norm": 0.21843454241752625,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 3574
+    },
+    {
+      "epoch": 4.846800433839479,
+      "grad_norm": 0.1680547446012497,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 3575
+    },
+    {
+      "epoch": 4.848156182212581,
+      "grad_norm": 0.16468442976474762,
+      "learning_rate": 0.0001,
+      "loss": 0.1498,
+      "step": 3576
+    },
+    {
+      "epoch": 4.849511930585683,
+      "grad_norm": 0.18719074130058289,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3577
+    },
+    {
+      "epoch": 4.850867678958785,
+      "grad_norm": 0.19062285125255585,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 3578
+    },
+    {
+      "epoch": 4.852223427331888,
+      "grad_norm": 0.16870160400867462,
+      "learning_rate": 0.0001,
+      "loss": 0.1366,
+      "step": 3579
+    },
+    {
+      "epoch": 4.853579175704989,
+      "grad_norm": 0.19802866876125336,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 3580
+    },
+    {
+      "epoch": 4.8549349240780915,
+      "grad_norm": 0.16935677826404572,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 3581
+    },
+    {
+      "epoch": 4.856290672451193,
+      "grad_norm": 0.1808127611875534,
+      "learning_rate": 0.0001,
+      "loss": 0.1564,
+      "step": 3582
+    },
+    {
+      "epoch": 4.857646420824295,
+      "grad_norm": 0.17421944439411163,
+      "learning_rate": 0.0001,
+      "loss": 0.1388,
+      "step": 3583
+    },
+    {
+      "epoch": 4.859002169197397,
+      "grad_norm": 0.1732134222984314,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 3584
+    },
+    {
+      "epoch": 4.860357917570499,
+      "grad_norm": 0.1917235553264618,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 3585
+    },
+    {
+      "epoch": 4.861713665943601,
+      "grad_norm": 0.1784203201532364,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 3586
+    },
+    {
+      "epoch": 4.863069414316703,
+      "grad_norm": 0.1725122332572937,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3587
+    },
+    {
+      "epoch": 4.8644251626898045,
+      "grad_norm": 0.17237123847007751,
+      "learning_rate": 0.0001,
+      "loss": 0.1614,
+      "step": 3588
+    },
+    {
+      "epoch": 4.865780911062907,
+      "grad_norm": 0.14557112753391266,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 3589
+    },
+    {
+      "epoch": 4.867136659436008,
+      "grad_norm": 0.17193172872066498,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 3590
+    },
+    {
+      "epoch": 4.868492407809111,
+      "grad_norm": 0.15810562670230865,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3591
+    },
+    {
+      "epoch": 4.869848156182212,
+      "grad_norm": 0.17525851726531982,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 3592
+    },
+    {
+      "epoch": 4.871203904555315,
+      "grad_norm": 0.15603366494178772,
+      "learning_rate": 0.0001,
+      "loss": 0.1447,
+      "step": 3593
+    },
+    {
+      "epoch": 4.872559652928416,
+      "grad_norm": 0.17024736106395721,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 3594
+    },
+    {
+      "epoch": 4.8739154013015185,
+      "grad_norm": 0.17609041929244995,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 3595
+    },
+    {
+      "epoch": 4.875271149674621,
+      "grad_norm": 0.20263245701789856,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 3596
+    },
+    {
+      "epoch": 4.876626898047722,
+      "grad_norm": 0.16922537982463837,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 3597
+    },
+    {
+      "epoch": 4.877982646420824,
+      "grad_norm": 0.20815037190914154,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 3598
+    },
+    {
+      "epoch": 4.879338394793926,
+      "grad_norm": 0.18487748503684998,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 3599
+    },
+    {
+      "epoch": 4.8806941431670285,
+      "grad_norm": 0.17414698004722595,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 3600
+    },
+    {
+      "epoch": 4.88204989154013,
+      "grad_norm": 0.1767132729291916,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 3601
+    },
+    {
+      "epoch": 4.883405639913232,
+      "grad_norm": 0.1815786510705948,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3602
+    },
+    {
+      "epoch": 4.884761388286334,
+      "grad_norm": 0.16805146634578705,
+      "learning_rate": 0.0001,
+      "loss": 0.151,
+      "step": 3603
+    },
+    {
+      "epoch": 4.886117136659436,
+      "grad_norm": 0.17632882297039032,
+      "learning_rate": 0.0001,
+      "loss": 0.1502,
+      "step": 3604
+    },
+    {
+      "epoch": 4.887472885032538,
+      "grad_norm": 0.17220023274421692,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 3605
+    },
+    {
+      "epoch": 4.88882863340564,
+      "grad_norm": 0.17020520567893982,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 3606
+    },
+    {
+      "epoch": 4.890184381778742,
+      "grad_norm": 0.17332690954208374,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3607
+    },
+    {
+      "epoch": 4.891540130151844,
+      "grad_norm": 0.23671475052833557,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 3608
+    },
+    {
+      "epoch": 4.8928958785249455,
+      "grad_norm": 0.1602783352136612,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 3609
+    },
+    {
+      "epoch": 4.894251626898048,
+      "grad_norm": 0.1773342341184616,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3610
+    },
+    {
+      "epoch": 4.895607375271149,
+      "grad_norm": 0.16765156388282776,
+      "learning_rate": 0.0001,
+      "loss": 0.1501,
+      "step": 3611
+    },
+    {
+      "epoch": 4.896963123644252,
+      "grad_norm": 0.15519852936267853,
+      "learning_rate": 0.0001,
+      "loss": 0.1437,
+      "step": 3612
+    },
+    {
+      "epoch": 4.898318872017354,
+      "grad_norm": 0.1597619503736496,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 3613
+    },
+    {
+      "epoch": 4.8996746203904555,
+      "grad_norm": 0.16426722705364227,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 3614
+    },
+    {
+      "epoch": 4.901030368763557,
+      "grad_norm": 0.19765938818454742,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 3615
+    },
+    {
+      "epoch": 4.902386117136659,
+      "grad_norm": 0.17634209990501404,
+      "learning_rate": 0.0001,
+      "loss": 0.1567,
+      "step": 3616
+    },
+    {
+      "epoch": 4.903741865509762,
+      "grad_norm": 0.16385629773139954,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3617
+    },
+    {
+      "epoch": 4.905097613882863,
+      "grad_norm": 0.19646531343460083,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 3618
+    },
+    {
+      "epoch": 4.906453362255966,
+      "grad_norm": 0.1666743904352188,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 3619
+    },
+    {
+      "epoch": 4.907809110629067,
+      "grad_norm": 0.18106748163700104,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3620
+    },
+    {
+      "epoch": 4.9091648590021695,
+      "grad_norm": 0.21364550292491913,
+      "learning_rate": 0.0001,
+      "loss": 0.1573,
+      "step": 3621
+    },
+    {
+      "epoch": 4.910520607375271,
+      "grad_norm": 0.24629782140254974,
+      "learning_rate": 0.0001,
+      "loss": 0.1521,
+      "step": 3622
+    },
+    {
+      "epoch": 4.911876355748373,
+      "grad_norm": 0.19388675689697266,
+      "learning_rate": 0.0001,
+      "loss": 0.157,
+      "step": 3623
+    },
+    {
+      "epoch": 4.913232104121475,
+      "grad_norm": 0.2235429435968399,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 3624
+    },
+    {
+      "epoch": 4.914587852494577,
+      "grad_norm": 0.21154628694057465,
+      "learning_rate": 0.0001,
+      "loss": 0.1518,
+      "step": 3625
+    },
+    {
+      "epoch": 4.915943600867679,
+      "grad_norm": 0.15410004556179047,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3626
+    },
+    {
+      "epoch": 4.917299349240781,
+      "grad_norm": 0.1838187426328659,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 3627
+    },
+    {
+      "epoch": 4.9186550976138825,
+      "grad_norm": 0.16143520176410675,
+      "learning_rate": 0.0001,
+      "loss": 0.1427,
+      "step": 3628
+    },
+    {
+      "epoch": 4.920010845986985,
+      "grad_norm": 0.16722097992897034,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 3629
+    },
+    {
+      "epoch": 4.921366594360086,
+      "grad_norm": 0.17140308022499084,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3630
+    },
+    {
+      "epoch": 4.922722342733189,
+      "grad_norm": 0.13132096827030182,
+      "learning_rate": 0.0001,
+      "loss": 0.1221,
+      "step": 3631
+    },
+    {
+      "epoch": 4.92407809110629,
+      "grad_norm": 0.19125492870807648,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 3632
+    },
+    {
+      "epoch": 4.925433839479393,
+      "grad_norm": 0.167622908949852,
+      "learning_rate": 0.0001,
+      "loss": 0.1479,
+      "step": 3633
+    },
+    {
+      "epoch": 4.926789587852495,
+      "grad_norm": 0.16791361570358276,
+      "learning_rate": 0.0001,
+      "loss": 0.1326,
+      "step": 3634
+    },
+    {
+      "epoch": 4.9281453362255965,
+      "grad_norm": 0.17410235106945038,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3635
+    },
+    {
+      "epoch": 4.929501084598699,
+      "grad_norm": 0.19615580141544342,
+      "learning_rate": 0.0001,
+      "loss": 0.1592,
+      "step": 3636
+    },
+    {
+      "epoch": 4.9308568329718,
+      "grad_norm": 0.1862279176712036,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 3637
+    },
+    {
+      "epoch": 4.932212581344903,
+      "grad_norm": 0.1809968650341034,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 3638
+    },
+    {
+      "epoch": 4.933568329718004,
+      "grad_norm": 0.23079454898834229,
+      "learning_rate": 0.0001,
+      "loss": 0.1513,
+      "step": 3639
+    },
+    {
+      "epoch": 4.934924078091107,
+      "grad_norm": 0.15342378616333008,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 3640
+    },
+    {
+      "epoch": 4.936279826464208,
+      "grad_norm": 0.17765669524669647,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 3641
+    },
+    {
+      "epoch": 4.93763557483731,
+      "grad_norm": 0.1723620444536209,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 3642
+    },
+    {
+      "epoch": 4.938991323210412,
+      "grad_norm": 0.18482692539691925,
+      "learning_rate": 0.0001,
+      "loss": 0.1315,
+      "step": 3643
+    },
+    {
+      "epoch": 4.940347071583514,
+      "grad_norm": 0.15473949909210205,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3644
+    },
+    {
+      "epoch": 4.941702819956616,
+      "grad_norm": 0.171922966837883,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3645
+    },
+    {
+      "epoch": 4.943058568329718,
+      "grad_norm": 0.1608002781867981,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3646
+    },
+    {
+      "epoch": 4.94441431670282,
+      "grad_norm": 0.19476306438446045,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "step": 3647
+    },
+    {
+      "epoch": 4.945770065075922,
+      "grad_norm": 0.16973794996738434,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 3648
+    },
+    {
+      "epoch": 4.9471258134490235,
+      "grad_norm": 0.16994601488113403,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3649
+    },
+    {
+      "epoch": 4.948481561822126,
+      "grad_norm": 0.19343267381191254,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 3650
+    },
+    {
+      "epoch": 4.949837310195228,
+      "grad_norm": 0.21064072847366333,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3651
+    },
+    {
+      "epoch": 4.95119305856833,
+      "grad_norm": 0.23989996314048767,
+      "learning_rate": 0.0001,
+      "loss": 0.1633,
+      "step": 3652
+    },
+    {
+      "epoch": 4.952548806941431,
+      "grad_norm": 0.1799103319644928,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 3653
+    },
+    {
+      "epoch": 4.9539045553145336,
+      "grad_norm": 0.3055696189403534,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 3654
+    },
+    {
+      "epoch": 4.955260303687636,
+      "grad_norm": 0.21054604649543762,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 3655
+    },
+    {
+      "epoch": 4.956616052060737,
+      "grad_norm": 0.1809500902891159,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3656
+    },
+    {
+      "epoch": 4.95797180043384,
+      "grad_norm": 0.17708100378513336,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 3657
+    },
+    {
+      "epoch": 4.959327548806941,
+      "grad_norm": 0.1890614926815033,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3658
+    },
+    {
+      "epoch": 4.960683297180044,
+      "grad_norm": 0.1907016634941101,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 3659
+    },
+    {
+      "epoch": 4.962039045553145,
+      "grad_norm": 0.17200808227062225,
+      "learning_rate": 0.0001,
+      "loss": 0.1488,
+      "step": 3660
+    },
+    {
+      "epoch": 4.9633947939262475,
+      "grad_norm": 0.19260121881961823,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 3661
+    },
+    {
+      "epoch": 4.964750542299349,
+      "grad_norm": 0.1791544258594513,
+      "learning_rate": 0.0001,
+      "loss": 0.1605,
+      "step": 3662
+    },
+    {
+      "epoch": 4.966106290672451,
+      "grad_norm": 0.2451677769422531,
+      "learning_rate": 0.0001,
+      "loss": 0.1522,
+      "step": 3663
+    },
+    {
+      "epoch": 4.967462039045553,
+      "grad_norm": 0.14601077139377594,
+      "learning_rate": 0.0001,
+      "loss": 0.129,
+      "step": 3664
+    },
+    {
+      "epoch": 4.968817787418655,
+      "grad_norm": 0.21724118292331696,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 3665
+    },
+    {
+      "epoch": 4.970173535791757,
+      "grad_norm": 0.183166965842247,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3666
+    },
+    {
+      "epoch": 4.971529284164859,
+      "grad_norm": 0.17056751251220703,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3667
+    },
+    {
+      "epoch": 4.972885032537961,
+      "grad_norm": 0.19645896553993225,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 3668
+    },
+    {
+      "epoch": 4.974240780911063,
+      "grad_norm": 0.19081225991249084,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 3669
+    },
+    {
+      "epoch": 4.975596529284164,
+      "grad_norm": 0.16865240037441254,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 3670
+    },
+    {
+      "epoch": 4.976952277657267,
+      "grad_norm": 0.18388719856739044,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 3671
+    },
+    {
+      "epoch": 4.978308026030369,
+      "grad_norm": 0.17200466990470886,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3672
+    },
+    {
+      "epoch": 4.979663774403471,
+      "grad_norm": 0.15525725483894348,
+      "learning_rate": 0.0001,
+      "loss": 0.1561,
+      "step": 3673
+    },
+    {
+      "epoch": 4.981019522776573,
+      "grad_norm": 0.18589025735855103,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 3674
+    },
+    {
+      "epoch": 4.9823752711496745,
+      "grad_norm": 0.17635135352611542,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 3675
+    },
+    {
+      "epoch": 4.983731019522777,
+      "grad_norm": 0.1862964779138565,
+      "learning_rate": 0.0001,
+      "loss": 0.1543,
+      "step": 3676
+    },
+    {
+      "epoch": 4.985086767895878,
+      "grad_norm": 0.1759718358516693,
+      "learning_rate": 0.0001,
+      "loss": 0.1481,
+      "step": 3677
+    },
+    {
+      "epoch": 4.986442516268981,
+      "grad_norm": 0.19465459883213043,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 3678
+    },
+    {
+      "epoch": 4.987798264642082,
+      "grad_norm": 0.20093528926372528,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 3679
+    },
+    {
+      "epoch": 4.989154013015185,
+      "grad_norm": 0.2585608959197998,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 3680
+    },
+    {
+      "epoch": 4.990509761388286,
+      "grad_norm": 0.19867657124996185,
+      "learning_rate": 0.0001,
+      "loss": 0.1612,
+      "step": 3681
+    },
+    {
+      "epoch": 4.991865509761388,
+      "grad_norm": 0.20651739835739136,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 3682
+    },
+    {
+      "epoch": 4.99322125813449,
+      "grad_norm": 0.2533813416957855,
+      "learning_rate": 0.0001,
+      "loss": 0.1608,
+      "step": 3683
+    },
+    {
+      "epoch": 4.994577006507592,
+      "grad_norm": 0.21776250004768372,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3684
+    },
+    {
+      "epoch": 4.995932754880694,
+      "grad_norm": 0.2116597443819046,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3685
+    },
+    {
+      "epoch": 4.997288503253796,
+      "grad_norm": 0.1929144561290741,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 3686
+    },
+    {
+      "epoch": 4.998644251626898,
+      "grad_norm": 0.17420583963394165,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 3687
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 0.17896969616413116,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 3688
+    },
+    {
+      "epoch": 5.001355748373102,
+      "grad_norm": 0.1459037959575653,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 3689
+    },
+    {
+      "epoch": 5.002711496746204,
+      "grad_norm": 0.16055841743946075,
+      "learning_rate": 0.0001,
+      "loss": 0.1312,
+      "step": 3690
+    },
+    {
+      "epoch": 5.004067245119306,
+      "grad_norm": 0.1742427945137024,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 3691
+    },
+    {
+      "epoch": 5.005422993492408,
+      "grad_norm": 0.1574975550174713,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 3692
+    },
+    {
+      "epoch": 5.00677874186551,
+      "grad_norm": 0.17318610846996307,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3693
+    },
+    {
+      "epoch": 5.008134490238612,
+      "grad_norm": 0.17581161856651306,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3694
+    },
+    {
+      "epoch": 5.009490238611714,
+      "grad_norm": 0.17843496799468994,
+      "learning_rate": 0.0001,
+      "loss": 0.127,
+      "step": 3695
+    },
+    {
+      "epoch": 5.010845986984815,
+      "grad_norm": 0.18170936405658722,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 3696
+    },
+    {
+      "epoch": 5.012201735357918,
+      "grad_norm": 0.17674800753593445,
+      "learning_rate": 0.0001,
+      "loss": 0.1212,
+      "step": 3697
+    },
+    {
+      "epoch": 5.013557483731019,
+      "grad_norm": 0.1857871562242508,
+      "learning_rate": 0.0001,
+      "loss": 0.133,
+      "step": 3698
+    },
+    {
+      "epoch": 5.014913232104122,
+      "grad_norm": 0.21306639909744263,
+      "learning_rate": 0.0001,
+      "loss": 0.1365,
+      "step": 3699
+    },
+    {
+      "epoch": 5.016268980477223,
+      "grad_norm": 0.2321872115135193,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 3700
+    },
+    {
+      "epoch": 5.0176247288503255,
+      "grad_norm": 0.20657402276992798,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 3701
+    },
+    {
+      "epoch": 5.018980477223427,
+      "grad_norm": 0.20073236525058746,
+      "learning_rate": 0.0001,
+      "loss": 0.1294,
+      "step": 3702
+    },
+    {
+      "epoch": 5.020336225596529,
+      "grad_norm": 0.22149407863616943,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3703
+    },
+    {
+      "epoch": 5.021691973969631,
+      "grad_norm": 0.19679497182369232,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 3704
+    },
+    {
+      "epoch": 5.023047722342733,
+      "grad_norm": 0.20500943064689636,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 3705
+    },
+    {
+      "epoch": 5.024403470715835,
+      "grad_norm": 0.1662627011537552,
+      "learning_rate": 0.0001,
+      "loss": 0.1169,
+      "step": 3706
+    },
+    {
+      "epoch": 5.025759219088937,
+      "grad_norm": 0.17743834853172302,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 3707
+    },
+    {
+      "epoch": 5.0271149674620395,
+      "grad_norm": 0.19327175617218018,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3708
+    },
+    {
+      "epoch": 5.028470715835141,
+      "grad_norm": 0.18815995752811432,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3709
+    },
+    {
+      "epoch": 5.029826464208243,
+      "grad_norm": 0.16772347688674927,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 3710
+    },
+    {
+      "epoch": 5.031182212581345,
+      "grad_norm": 0.18108178675174713,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 3711
+    },
+    {
+      "epoch": 5.032537960954447,
+      "grad_norm": 0.17932546138763428,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 3712
+    },
+    {
+      "epoch": 5.033893709327549,
+      "grad_norm": 0.18960824608802795,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 3713
+    },
+    {
+      "epoch": 5.035249457700651,
+      "grad_norm": 0.18402527272701263,
+      "learning_rate": 0.0001,
+      "loss": 0.1427,
+      "step": 3714
+    },
+    {
+      "epoch": 5.0366052060737525,
+      "grad_norm": 0.1648482233285904,
+      "learning_rate": 0.0001,
+      "loss": 0.1297,
+      "step": 3715
+    },
+    {
+      "epoch": 5.037960954446855,
+      "grad_norm": 0.19778962433338165,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 3716
+    },
+    {
+      "epoch": 5.039316702819956,
+      "grad_norm": 0.19581440091133118,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "step": 3717
+    },
+    {
+      "epoch": 5.040672451193059,
+      "grad_norm": 0.19724906980991364,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 3718
+    },
+    {
+      "epoch": 5.04202819956616,
+      "grad_norm": 0.20006047189235687,
+      "learning_rate": 0.0001,
+      "loss": 0.1321,
+      "step": 3719
+    },
+    {
+      "epoch": 5.043383947939263,
+      "grad_norm": 0.20289820432662964,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 3720
+    },
+    {
+      "epoch": 5.044739696312364,
+      "grad_norm": 0.18855296075344086,
+      "learning_rate": 0.0001,
+      "loss": 0.1304,
+      "step": 3721
+    },
+    {
+      "epoch": 5.0460954446854664,
+      "grad_norm": 0.21181556582450867,
+      "learning_rate": 0.0001,
+      "loss": 0.132,
+      "step": 3722
+    },
+    {
+      "epoch": 5.047451193058568,
+      "grad_norm": 0.22952499985694885,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 3723
+    },
+    {
+      "epoch": 5.04880694143167,
+      "grad_norm": 0.2055777907371521,
+      "learning_rate": 0.0001,
+      "loss": 0.1448,
+      "step": 3724
+    },
+    {
+      "epoch": 5.050162689804772,
+      "grad_norm": 0.19826354086399078,
+      "learning_rate": 0.0001,
+      "loss": 0.1352,
+      "step": 3725
+    },
+    {
+      "epoch": 5.051518438177874,
+      "grad_norm": 0.18285784125328064,
+      "learning_rate": 0.0001,
+      "loss": 0.1255,
+      "step": 3726
+    },
+    {
+      "epoch": 5.0528741865509765,
+      "grad_norm": 0.19297826290130615,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 3727
+    },
+    {
+      "epoch": 5.054229934924078,
+      "grad_norm": 0.20574496686458588,
+      "learning_rate": 0.0001,
+      "loss": 0.1397,
+      "step": 3728
+    },
+    {
+      "epoch": 5.05558568329718,
+      "grad_norm": 0.20263321697711945,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "step": 3729
+    },
+    {
+      "epoch": 5.056941431670282,
+      "grad_norm": 0.2362242043018341,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 3730
+    },
+    {
+      "epoch": 5.058297180043384,
+      "grad_norm": 0.1986127495765686,
+      "learning_rate": 0.0001,
+      "loss": 0.1312,
+      "step": 3731
+    },
+    {
+      "epoch": 5.059652928416486,
+      "grad_norm": 0.16935548186302185,
+      "learning_rate": 0.0001,
+      "loss": 0.1294,
+      "step": 3732
+    },
+    {
+      "epoch": 5.061008676789588,
+      "grad_norm": 0.18661154806613922,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 3733
+    },
+    {
+      "epoch": 5.06236442516269,
+      "grad_norm": 0.1805429756641388,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3734
+    },
+    {
+      "epoch": 5.063720173535792,
+      "grad_norm": 0.1880904734134674,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "step": 3735
+    },
+    {
+      "epoch": 5.065075921908893,
+      "grad_norm": 0.18620361387729645,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "step": 3736
+    },
+    {
+      "epoch": 5.066431670281996,
+      "grad_norm": 0.1917521059513092,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "step": 3737
+    },
+    {
+      "epoch": 5.067787418655097,
+      "grad_norm": 0.18394559621810913,
+      "learning_rate": 0.0001,
+      "loss": 0.1261,
+      "step": 3738
+    },
+    {
+      "epoch": 5.0691431670282,
+      "grad_norm": 0.24504998326301575,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 3739
+    },
+    {
+      "epoch": 5.070498915401301,
+      "grad_norm": 0.20707863569259644,
+      "learning_rate": 0.0001,
+      "loss": 0.1293,
+      "step": 3740
+    },
+    {
+      "epoch": 5.0718546637744035,
+      "grad_norm": 0.220600888133049,
+      "learning_rate": 0.0001,
+      "loss": 0.1264,
+      "step": 3741
+    },
+    {
+      "epoch": 5.073210412147505,
+      "grad_norm": 0.19943201541900635,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3742
+    },
+    {
+      "epoch": 5.074566160520607,
+      "grad_norm": 0.22103723883628845,
+      "learning_rate": 0.0001,
+      "loss": 0.1246,
+      "step": 3743
+    },
+    {
+      "epoch": 5.07592190889371,
+      "grad_norm": 0.22077922523021698,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 3744
+    },
+    {
+      "epoch": 5.077277657266811,
+      "grad_norm": 0.20367443561553955,
+      "learning_rate": 0.0001,
+      "loss": 0.1238,
+      "step": 3745
+    },
+    {
+      "epoch": 5.078633405639914,
+      "grad_norm": 0.20239482820034027,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 3746
+    },
+    {
+      "epoch": 5.079989154013015,
+      "grad_norm": 0.2258267104625702,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 3747
+    },
+    {
+      "epoch": 5.0813449023861175,
+      "grad_norm": 0.19551678001880646,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 3748
+    },
+    {
+      "epoch": 5.082700650759219,
+      "grad_norm": 0.1813683807849884,
+      "learning_rate": 0.0001,
+      "loss": 0.1332,
+      "step": 3749
+    },
+    {
+      "epoch": 5.084056399132321,
+      "grad_norm": 0.21899621188640594,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 3750
+    },
+    {
+      "epoch": 5.085412147505423,
+      "grad_norm": 0.1948169320821762,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3751
+    },
+    {
+      "epoch": 5.086767895878525,
+      "grad_norm": 0.2121468037366867,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3752
+    },
+    {
+      "epoch": 5.088123644251627,
+      "grad_norm": 0.17412881553173065,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3753
+    },
+    {
+      "epoch": 5.089479392624729,
+      "grad_norm": 0.19968664646148682,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 3754
+    },
+    {
+      "epoch": 5.0908351409978305,
+      "grad_norm": 0.1773080676794052,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 3755
+    },
+    {
+      "epoch": 5.092190889370933,
+      "grad_norm": 0.18925388157367706,
+      "learning_rate": 0.0001,
+      "loss": 0.1282,
+      "step": 3756
+    },
+    {
+      "epoch": 5.093546637744034,
+      "grad_norm": 0.18845148384571075,
+      "learning_rate": 0.0001,
+      "loss": 0.1267,
+      "step": 3757
+    },
+    {
+      "epoch": 5.094902386117137,
+      "grad_norm": 0.1691565364599228,
+      "learning_rate": 0.0001,
+      "loss": 0.1224,
+      "step": 3758
+    },
+    {
+      "epoch": 5.096258134490238,
+      "grad_norm": 0.17583447694778442,
+      "learning_rate": 0.0001,
+      "loss": 0.1219,
+      "step": 3759
+    },
+    {
+      "epoch": 5.097613882863341,
+      "grad_norm": 0.20752811431884766,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 3760
+    },
+    {
+      "epoch": 5.098969631236442,
+      "grad_norm": 0.1866748183965683,
+      "learning_rate": 0.0001,
+      "loss": 0.1307,
+      "step": 3761
+    },
+    {
+      "epoch": 5.1003253796095445,
+      "grad_norm": 0.2122664898633957,
+      "learning_rate": 0.0001,
+      "loss": 0.1355,
+      "step": 3762
+    },
+    {
+      "epoch": 5.101681127982647,
+      "grad_norm": 0.18646736443042755,
+      "learning_rate": 0.0001,
+      "loss": 0.123,
+      "step": 3763
+    },
+    {
+      "epoch": 5.103036876355748,
+      "grad_norm": 0.2090400606393814,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 3764
+    },
+    {
+      "epoch": 5.104392624728851,
+      "grad_norm": 0.1950397789478302,
+      "learning_rate": 0.0001,
+      "loss": 0.1279,
+      "step": 3765
+    },
+    {
+      "epoch": 5.105748373101952,
+      "grad_norm": 0.2392561286687851,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3766
+    },
+    {
+      "epoch": 5.1071041214750545,
+      "grad_norm": 0.20499782264232635,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 3767
+    },
+    {
+      "epoch": 5.108459869848156,
+      "grad_norm": 0.23551443219184875,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3768
+    },
+    {
+      "epoch": 5.109815618221258,
+      "grad_norm": 0.22145359218120575,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 3769
+    },
+    {
+      "epoch": 5.11117136659436,
+      "grad_norm": 0.2017410695552826,
+      "learning_rate": 0.0001,
+      "loss": 0.1347,
+      "step": 3770
+    },
+    {
+      "epoch": 5.112527114967462,
+      "grad_norm": 0.20491015911102295,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 3771
+    },
+    {
+      "epoch": 5.113882863340564,
+      "grad_norm": 0.19087745249271393,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3772
+    },
+    {
+      "epoch": 5.115238611713666,
+      "grad_norm": 0.17719033360481262,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 3773
+    },
+    {
+      "epoch": 5.116594360086768,
+      "grad_norm": 0.222449392080307,
+      "learning_rate": 0.0001,
+      "loss": 0.1429,
+      "step": 3774
+    },
+    {
+      "epoch": 5.11795010845987,
+      "grad_norm": 0.18914712965488434,
+      "learning_rate": 0.0001,
+      "loss": 0.1192,
+      "step": 3775
+    },
+    {
+      "epoch": 5.1193058568329715,
+      "grad_norm": 0.22027070820331573,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 3776
+    },
+    {
+      "epoch": 5.120661605206074,
+      "grad_norm": 0.2910611927509308,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 3777
+    },
+    {
+      "epoch": 5.122017353579175,
+      "grad_norm": 0.2078401893377304,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 3778
+    },
+    {
+      "epoch": 5.123373101952278,
+      "grad_norm": 0.16397380828857422,
+      "learning_rate": 0.0001,
+      "loss": 0.1222,
+      "step": 3779
+    },
+    {
+      "epoch": 5.124728850325379,
+      "grad_norm": 0.21659260988235474,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3780
+    },
+    {
+      "epoch": 5.1260845986984815,
+      "grad_norm": 0.19768762588500977,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 3781
+    },
+    {
+      "epoch": 5.127440347071584,
+      "grad_norm": 0.20528782904148102,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 3782
+    },
+    {
+      "epoch": 5.128796095444685,
+      "grad_norm": 0.20931823551654816,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 3783
+    },
+    {
+      "epoch": 5.130151843817788,
+      "grad_norm": 0.1983269900083542,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3784
+    },
+    {
+      "epoch": 5.131507592190889,
+      "grad_norm": 0.2187405377626419,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 3785
+    },
+    {
+      "epoch": 5.132863340563992,
+      "grad_norm": 0.2203279435634613,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 3786
+    },
+    {
+      "epoch": 5.134219088937093,
+      "grad_norm": 0.19385920464992523,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3787
+    },
+    {
+      "epoch": 5.1355748373101955,
+      "grad_norm": 0.181990846991539,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 3788
+    },
+    {
+      "epoch": 5.136930585683297,
+      "grad_norm": 0.20600350201129913,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 3789
+    },
+    {
+      "epoch": 5.138286334056399,
+      "grad_norm": 0.18734464049339294,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 3790
+    },
+    {
+      "epoch": 5.139642082429501,
+      "grad_norm": 0.21551395952701569,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 3791
+    },
+    {
+      "epoch": 5.140997830802603,
+      "grad_norm": 0.18591450154781342,
+      "learning_rate": 0.0001,
+      "loss": 0.1337,
+      "step": 3792
+    },
+    {
+      "epoch": 5.142353579175705,
+      "grad_norm": 0.20778079330921173,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 3793
+    },
+    {
+      "epoch": 5.143709327548807,
+      "grad_norm": 0.20774921774864197,
+      "learning_rate": 0.0001,
+      "loss": 0.125,
+      "step": 3794
+    },
+    {
+      "epoch": 5.1450650759219085,
+      "grad_norm": 0.17566008865833282,
+      "learning_rate": 0.0001,
+      "loss": 0.1339,
+      "step": 3795
+    },
+    {
+      "epoch": 5.146420824295011,
+      "grad_norm": 0.1849910020828247,
+      "learning_rate": 0.0001,
+      "loss": 0.1306,
+      "step": 3796
+    },
+    {
+      "epoch": 5.147776572668112,
+      "grad_norm": 0.18262450397014618,
+      "learning_rate": 0.0001,
+      "loss": 0.1304,
+      "step": 3797
+    },
+    {
+      "epoch": 5.149132321041215,
+      "grad_norm": 0.1993110626935959,
+      "learning_rate": 0.0001,
+      "loss": 0.1503,
+      "step": 3798
+    },
+    {
+      "epoch": 5.150488069414317,
+      "grad_norm": 0.21482586860656738,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 3799
+    },
+    {
+      "epoch": 5.151843817787419,
+      "grad_norm": 0.20116373896598816,
+      "learning_rate": 0.0001,
+      "loss": 0.1497,
+      "step": 3800
+    },
+    {
+      "epoch": 5.153199566160521,
+      "grad_norm": 0.20637138187885284,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 3801
+    },
+    {
+      "epoch": 5.1545553145336225,
+      "grad_norm": 0.1744101643562317,
+      "learning_rate": 0.0001,
+      "loss": 0.1274,
+      "step": 3802
+    },
+    {
+      "epoch": 5.155911062906725,
+      "grad_norm": 0.20255570113658905,
+      "learning_rate": 0.0001,
+      "loss": 0.1215,
+      "step": 3803
+    },
+    {
+      "epoch": 5.157266811279826,
+      "grad_norm": 0.1982630044221878,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 3804
+    },
+    {
+      "epoch": 5.158622559652929,
+      "grad_norm": 0.18022654950618744,
+      "learning_rate": 0.0001,
+      "loss": 0.1269,
+      "step": 3805
+    },
+    {
+      "epoch": 5.15997830802603,
+      "grad_norm": 0.21304933726787567,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 3806
+    },
+    {
+      "epoch": 5.161334056399133,
+      "grad_norm": 0.20757918059825897,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 3807
+    },
+    {
+      "epoch": 5.162689804772234,
+      "grad_norm": 0.2010778784751892,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 3808
+    },
+    {
+      "epoch": 5.164045553145336,
+      "grad_norm": 0.17674343287944794,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3809
+    },
+    {
+      "epoch": 5.165401301518438,
+      "grad_norm": 0.21640251576900482,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3810
+    },
+    {
+      "epoch": 5.16675704989154,
+      "grad_norm": 0.17253616452217102,
+      "learning_rate": 0.0001,
+      "loss": 0.1236,
+      "step": 3811
+    },
+    {
+      "epoch": 5.168112798264642,
+      "grad_norm": 0.2024357169866562,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 3812
+    },
+    {
+      "epoch": 5.169468546637744,
+      "grad_norm": 0.18557903170585632,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3813
+    },
+    {
+      "epoch": 5.170824295010846,
+      "grad_norm": 0.19443611800670624,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 3814
+    },
+    {
+      "epoch": 5.172180043383948,
+      "grad_norm": 0.18382392823696136,
+      "learning_rate": 0.0001,
+      "loss": 0.1361,
+      "step": 3815
+    },
+    {
+      "epoch": 5.1735357917570495,
+      "grad_norm": 0.19892576336860657,
+      "learning_rate": 0.0001,
+      "loss": 0.1295,
+      "step": 3816
+    },
+    {
+      "epoch": 5.174891540130152,
+      "grad_norm": 0.1782553791999817,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 3817
+    },
+    {
+      "epoch": 5.176247288503254,
+      "grad_norm": 0.18879933655261993,
+      "learning_rate": 0.0001,
+      "loss": 0.1207,
+      "step": 3818
+    },
+    {
+      "epoch": 5.177603036876356,
+      "grad_norm": 0.17705178260803223,
+      "learning_rate": 0.0001,
+      "loss": 0.1257,
+      "step": 3819
+    },
+    {
+      "epoch": 5.178958785249458,
+      "grad_norm": 0.22203262150287628,
+      "learning_rate": 0.0001,
+      "loss": 0.1337,
+      "step": 3820
+    },
+    {
+      "epoch": 5.18031453362256,
+      "grad_norm": 0.2567465901374817,
+      "learning_rate": 0.0001,
+      "loss": 0.1283,
+      "step": 3821
+    },
+    {
+      "epoch": 5.181670281995662,
+      "grad_norm": 0.22039654850959778,
+      "learning_rate": 0.0001,
+      "loss": 0.1358,
+      "step": 3822
+    },
+    {
+      "epoch": 5.183026030368763,
+      "grad_norm": 0.21611815690994263,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 3823
+    },
+    {
+      "epoch": 5.184381778741866,
+      "grad_norm": 0.21296539902687073,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3824
+    },
+    {
+      "epoch": 5.185737527114967,
+      "grad_norm": 0.2032531201839447,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 3825
+    },
+    {
+      "epoch": 5.18709327548807,
+      "grad_norm": 0.19675280153751373,
+      "learning_rate": 0.0001,
+      "loss": 0.1388,
+      "step": 3826
+    },
+    {
+      "epoch": 5.188449023861171,
+      "grad_norm": 0.2226826697587967,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 3827
+    },
+    {
+      "epoch": 5.1898047722342735,
+      "grad_norm": 0.17776957154273987,
+      "learning_rate": 0.0001,
+      "loss": 0.129,
+      "step": 3828
+    },
+    {
+      "epoch": 5.191160520607375,
+      "grad_norm": 0.19979839026927948,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3829
+    },
+    {
+      "epoch": 5.192516268980477,
+      "grad_norm": 0.19759666919708252,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 3830
+    },
+    {
+      "epoch": 5.193872017353579,
+      "grad_norm": 0.20706439018249512,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 3831
+    },
+    {
+      "epoch": 5.195227765726681,
+      "grad_norm": 0.18861521780490875,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 3832
+    },
+    {
+      "epoch": 5.196583514099783,
+      "grad_norm": 0.17032714188098907,
+      "learning_rate": 0.0001,
+      "loss": 0.1352,
+      "step": 3833
+    },
+    {
+      "epoch": 5.197939262472885,
+      "grad_norm": 0.19058646261692047,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 3834
+    },
+    {
+      "epoch": 5.1992950108459866,
+      "grad_norm": 0.21811529994010925,
+      "learning_rate": 0.0001,
+      "loss": 0.1445,
+      "step": 3835
+    },
+    {
+      "epoch": 5.200650759219089,
+      "grad_norm": 0.17997701466083527,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 3836
+    },
+    {
+      "epoch": 5.202006507592191,
+      "grad_norm": 0.21861736476421356,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 3837
+    },
+    {
+      "epoch": 5.203362255965293,
+      "grad_norm": 0.1728283017873764,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 3838
+    },
+    {
+      "epoch": 5.204718004338395,
+      "grad_norm": 0.21308064460754395,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3839
+    },
+    {
+      "epoch": 5.206073752711497,
+      "grad_norm": 0.17866861820220947,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 3840
+    },
+    {
+      "epoch": 5.207429501084599,
+      "grad_norm": 0.19965428113937378,
+      "learning_rate": 0.0001,
+      "loss": 0.1243,
+      "step": 3841
+    },
+    {
+      "epoch": 5.2087852494577005,
+      "grad_norm": 0.17260149121284485,
+      "learning_rate": 0.0001,
+      "loss": 0.1318,
+      "step": 3842
+    },
+    {
+      "epoch": 5.210140997830803,
+      "grad_norm": 0.2013937383890152,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 3843
+    },
+    {
+      "epoch": 5.211496746203904,
+      "grad_norm": 0.22428515553474426,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3844
+    },
+    {
+      "epoch": 5.212852494577007,
+      "grad_norm": 0.2237040102481842,
+      "learning_rate": 0.0001,
+      "loss": 0.1463,
+      "step": 3845
+    },
+    {
+      "epoch": 5.214208242950108,
+      "grad_norm": 0.1743355244398117,
+      "learning_rate": 0.0001,
+      "loss": 0.1154,
+      "step": 3846
+    },
+    {
+      "epoch": 5.215563991323211,
+      "grad_norm": 0.20919054746627808,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 3847
+    },
+    {
+      "epoch": 5.216919739696312,
+      "grad_norm": 0.20588727295398712,
+      "learning_rate": 0.0001,
+      "loss": 0.1281,
+      "step": 3848
+    },
+    {
+      "epoch": 5.218275488069414,
+      "grad_norm": 0.19741295278072357,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 3849
+    },
+    {
+      "epoch": 5.219631236442516,
+      "grad_norm": 0.180435910820961,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3850
+    },
+    {
+      "epoch": 5.220986984815618,
+      "grad_norm": 0.19891154766082764,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 3851
+    },
+    {
+      "epoch": 5.22234273318872,
+      "grad_norm": 0.17790701985359192,
+      "learning_rate": 0.0001,
+      "loss": 0.1321,
+      "step": 3852
+    },
+    {
+      "epoch": 5.223698481561822,
+      "grad_norm": 0.18890883028507233,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 3853
+    },
+    {
+      "epoch": 5.2250542299349245,
+      "grad_norm": 0.2474263310432434,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3854
+    },
+    {
+      "epoch": 5.226409978308026,
+      "grad_norm": 0.1706131100654602,
+      "learning_rate": 0.0001,
+      "loss": 0.117,
+      "step": 3855
+    },
+    {
+      "epoch": 5.227765726681128,
+      "grad_norm": 0.2217845916748047,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 3856
+    },
+    {
+      "epoch": 5.22912147505423,
+      "grad_norm": 0.21447904407978058,
+      "learning_rate": 0.0001,
+      "loss": 0.129,
+      "step": 3857
+    },
+    {
+      "epoch": 5.230477223427332,
+      "grad_norm": 0.1871328204870224,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 3858
+    },
+    {
+      "epoch": 5.231832971800434,
+      "grad_norm": 0.18111349642276764,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 3859
+    },
+    {
+      "epoch": 5.233188720173536,
+      "grad_norm": 0.17041108012199402,
+      "learning_rate": 0.0001,
+      "loss": 0.1284,
+      "step": 3860
+    },
+    {
+      "epoch": 5.234544468546638,
+      "grad_norm": 0.17362888157367706,
+      "learning_rate": 0.0001,
+      "loss": 0.1275,
+      "step": 3861
+    },
+    {
+      "epoch": 5.23590021691974,
+      "grad_norm": 0.2017630636692047,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 3862
+    },
+    {
+      "epoch": 5.237255965292841,
+      "grad_norm": 0.1782131940126419,
+      "learning_rate": 0.0001,
+      "loss": 0.1332,
+      "step": 3863
+    },
+    {
+      "epoch": 5.238611713665944,
+      "grad_norm": 0.21806348860263824,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 3864
+    },
+    {
+      "epoch": 5.239967462039045,
+      "grad_norm": 0.2075176239013672,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 3865
+    },
+    {
+      "epoch": 5.241323210412148,
+      "grad_norm": 0.25359222292900085,
+      "learning_rate": 0.0001,
+      "loss": 0.1525,
+      "step": 3866
+    },
+    {
+      "epoch": 5.242678958785249,
+      "grad_norm": 0.2238093912601471,
+      "learning_rate": 0.0001,
+      "loss": 0.1335,
+      "step": 3867
+    },
+    {
+      "epoch": 5.2440347071583515,
+      "grad_norm": 0.18724693357944489,
+      "learning_rate": 0.0001,
+      "loss": 0.1437,
+      "step": 3868
+    },
+    {
+      "epoch": 5.245390455531453,
+      "grad_norm": 0.21190552413463593,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 3869
+    },
+    {
+      "epoch": 5.246746203904555,
+      "grad_norm": 0.18589691817760468,
+      "learning_rate": 0.0001,
+      "loss": 0.1162,
+      "step": 3870
+    },
+    {
+      "epoch": 5.248101952277657,
+      "grad_norm": 0.21094201505184174,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3871
+    },
+    {
+      "epoch": 5.249457700650759,
+      "grad_norm": 0.2128743678331375,
+      "learning_rate": 0.0001,
+      "loss": 0.1439,
+      "step": 3872
+    },
+    {
+      "epoch": 5.250813449023861,
+      "grad_norm": 0.1944763958454132,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 3873
+    },
+    {
+      "epoch": 5.252169197396963,
+      "grad_norm": 0.19475007057189941,
+      "learning_rate": 0.0001,
+      "loss": 0.129,
+      "step": 3874
+    },
+    {
+      "epoch": 5.2535249457700655,
+      "grad_norm": 0.1987103372812271,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 3875
+    },
+    {
+      "epoch": 5.254880694143167,
+      "grad_norm": 0.2075999677181244,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 3876
+    },
+    {
+      "epoch": 5.256236442516269,
+      "grad_norm": 0.21752449870109558,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3877
+    },
+    {
+      "epoch": 5.257592190889371,
+      "grad_norm": 0.20933745801448822,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 3878
+    },
+    {
+      "epoch": 5.258947939262473,
+      "grad_norm": 0.18841655552387238,
+      "learning_rate": 0.0001,
+      "loss": 0.1321,
+      "step": 3879
+    },
+    {
+      "epoch": 5.260303687635575,
+      "grad_norm": 0.18272264301776886,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 3880
+    },
+    {
+      "epoch": 5.261659436008677,
+      "grad_norm": 0.18497809767723083,
+      "learning_rate": 0.0001,
+      "loss": 0.1298,
+      "step": 3881
+    },
+    {
+      "epoch": 5.2630151843817785,
+      "grad_norm": 0.21539273858070374,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 3882
+    },
+    {
+      "epoch": 5.264370932754881,
+      "grad_norm": 0.2073226273059845,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 3883
+    },
+    {
+      "epoch": 5.265726681127982,
+      "grad_norm": 0.17169463634490967,
+      "learning_rate": 0.0001,
+      "loss": 0.122,
+      "step": 3884
+    },
+    {
+      "epoch": 5.267082429501085,
+      "grad_norm": 0.25557956099510193,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 3885
+    },
+    {
+      "epoch": 5.268438177874186,
+      "grad_norm": 0.20945961773395538,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3886
+    },
+    {
+      "epoch": 5.269793926247289,
+      "grad_norm": 0.18738842010498047,
+      "learning_rate": 0.0001,
+      "loss": 0.1233,
+      "step": 3887
+    },
+    {
+      "epoch": 5.27114967462039,
+      "grad_norm": 0.20277082920074463,
+      "learning_rate": 0.0001,
+      "loss": 0.1352,
+      "step": 3888
+    },
+    {
+      "epoch": 5.2725054229934925,
+      "grad_norm": 0.21748986840248108,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3889
+    },
+    {
+      "epoch": 5.273861171366594,
+      "grad_norm": 0.2121599018573761,
+      "learning_rate": 0.0001,
+      "loss": 0.1345,
+      "step": 3890
+    },
+    {
+      "epoch": 5.275216919739696,
+      "grad_norm": 0.20960605144500732,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3891
+    },
+    {
+      "epoch": 5.276572668112799,
+      "grad_norm": 0.1859261393547058,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 3892
+    },
+    {
+      "epoch": 5.2779284164859,
+      "grad_norm": 0.21012559533119202,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 3893
+    },
+    {
+      "epoch": 5.2792841648590025,
+      "grad_norm": 0.18233902752399445,
+      "learning_rate": 0.0001,
+      "loss": 0.1249,
+      "step": 3894
+    },
+    {
+      "epoch": 5.280639913232104,
+      "grad_norm": 0.2126597911119461,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 3895
+    },
+    {
+      "epoch": 5.281995661605206,
+      "grad_norm": 0.2382776439189911,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 3896
+    },
+    {
+      "epoch": 5.283351409978308,
+      "grad_norm": 0.2580108642578125,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 3897
+    },
+    {
+      "epoch": 5.28470715835141,
+      "grad_norm": 0.1761651337146759,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 3898
+    },
+    {
+      "epoch": 5.286062906724512,
+      "grad_norm": 0.170660600066185,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 3899
+    },
+    {
+      "epoch": 5.287418655097614,
+      "grad_norm": 0.19068709015846252,
+      "learning_rate": 0.0001,
+      "loss": 0.1288,
+      "step": 3900
+    },
+    {
+      "epoch": 5.288774403470716,
+      "grad_norm": 0.20653203129768372,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3901
+    },
+    {
+      "epoch": 5.290130151843818,
+      "grad_norm": 0.1782461553812027,
+      "learning_rate": 0.0001,
+      "loss": 0.1304,
+      "step": 3902
+    },
+    {
+      "epoch": 5.2914859002169194,
+      "grad_norm": 0.16326598823070526,
+      "learning_rate": 0.0001,
+      "loss": 0.118,
+      "step": 3903
+    },
+    {
+      "epoch": 5.292841648590022,
+      "grad_norm": 0.24215014278888702,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 3904
+    },
+    {
+      "epoch": 5.294197396963123,
+      "grad_norm": 0.17270983755588531,
+      "learning_rate": 0.0001,
+      "loss": 0.1222,
+      "step": 3905
+    },
+    {
+      "epoch": 5.295553145336226,
+      "grad_norm": 0.18183240294456482,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 3906
+    },
+    {
+      "epoch": 5.296908893709327,
+      "grad_norm": 0.1884816735982895,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 3907
+    },
+    {
+      "epoch": 5.2982646420824295,
+      "grad_norm": 0.20102311670780182,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 3908
+    },
+    {
+      "epoch": 5.299620390455532,
+      "grad_norm": 0.19160257279872894,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 3909
+    },
+    {
+      "epoch": 5.300976138828633,
+      "grad_norm": 0.14438055455684662,
+      "learning_rate": 0.0001,
+      "loss": 0.1174,
+      "step": 3910
+    },
+    {
+      "epoch": 5.302331887201736,
+      "grad_norm": 0.17439775168895721,
+      "learning_rate": 0.0001,
+      "loss": 0.1317,
+      "step": 3911
+    },
+    {
+      "epoch": 5.303687635574837,
+      "grad_norm": 0.2277466356754303,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3912
+    },
+    {
+      "epoch": 5.30504338394794,
+      "grad_norm": 0.1815563440322876,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 3913
+    },
+    {
+      "epoch": 5.306399132321041,
+      "grad_norm": 0.18681283295154572,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 3914
+    },
+    {
+      "epoch": 5.3077548806941435,
+      "grad_norm": 0.18376539647579193,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 3915
+    },
+    {
+      "epoch": 5.309110629067245,
+      "grad_norm": 0.2074599266052246,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 3916
+    },
+    {
+      "epoch": 5.310466377440347,
+      "grad_norm": 0.19143739342689514,
+      "learning_rate": 0.0001,
+      "loss": 0.1275,
+      "step": 3917
+    },
+    {
+      "epoch": 5.311822125813449,
+      "grad_norm": 0.2400282919406891,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 3918
+    },
+    {
+      "epoch": 5.313177874186551,
+      "grad_norm": 0.24576151371002197,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 3919
+    },
+    {
+      "epoch": 5.314533622559653,
+      "grad_norm": 0.18895545601844788,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 3920
+    },
+    {
+      "epoch": 5.315889370932755,
+      "grad_norm": 0.18899565935134888,
+      "learning_rate": 0.0001,
+      "loss": 0.1361,
+      "step": 3921
+    },
+    {
+      "epoch": 5.3172451193058565,
+      "grad_norm": 0.21902668476104736,
+      "learning_rate": 0.0001,
+      "loss": 0.1285,
+      "step": 3922
+    },
+    {
+      "epoch": 5.318600867678959,
+      "grad_norm": 0.1917247623205185,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 3923
+    },
+    {
+      "epoch": 5.31995661605206,
+      "grad_norm": 0.2379962056875229,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 3924
+    },
+    {
+      "epoch": 5.321312364425163,
+      "grad_norm": 0.1867963671684265,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 3925
+    },
+    {
+      "epoch": 5.322668112798264,
+      "grad_norm": 0.20400063693523407,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 3926
+    },
+    {
+      "epoch": 5.324023861171367,
+      "grad_norm": 0.17210537195205688,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 3927
+    },
+    {
+      "epoch": 5.325379609544468,
+      "grad_norm": 0.20037761330604553,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 3928
+    },
+    {
+      "epoch": 5.3267353579175705,
+      "grad_norm": 0.20432761311531067,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 3929
+    },
+    {
+      "epoch": 5.328091106290673,
+      "grad_norm": 0.18991869688034058,
+      "learning_rate": 0.0001,
+      "loss": 0.1265,
+      "step": 3930
+    },
+    {
+      "epoch": 5.329446854663774,
+      "grad_norm": 0.20604266226291656,
+      "learning_rate": 0.0001,
+      "loss": 0.1366,
+      "step": 3931
+    },
+    {
+      "epoch": 5.330802603036877,
+      "grad_norm": 0.17990173399448395,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "step": 3932
+    },
+    {
+      "epoch": 5.332158351409978,
+      "grad_norm": 0.17874684929847717,
+      "learning_rate": 0.0001,
+      "loss": 0.1359,
+      "step": 3933
+    },
+    {
+      "epoch": 5.3335140997830806,
+      "grad_norm": 0.18908998370170593,
+      "learning_rate": 0.0001,
+      "loss": 0.1401,
+      "step": 3934
+    },
+    {
+      "epoch": 5.334869848156182,
+      "grad_norm": 0.19610393047332764,
+      "learning_rate": 0.0001,
+      "loss": 0.142,
+      "step": 3935
+    },
+    {
+      "epoch": 5.336225596529284,
+      "grad_norm": 0.18661242723464966,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 3936
+    },
+    {
+      "epoch": 5.337581344902386,
+      "grad_norm": 0.1833367943763733,
+      "learning_rate": 0.0001,
+      "loss": 0.1278,
+      "step": 3937
+    },
+    {
+      "epoch": 5.338937093275488,
+      "grad_norm": 0.20360694825649261,
+      "learning_rate": 0.0001,
+      "loss": 0.1542,
+      "step": 3938
+    },
+    {
+      "epoch": 5.34029284164859,
+      "grad_norm": 0.21344871819019318,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3939
+    },
+    {
+      "epoch": 5.341648590021692,
+      "grad_norm": 0.18398606777191162,
+      "learning_rate": 0.0001,
+      "loss": 0.1332,
+      "step": 3940
+    },
+    {
+      "epoch": 5.343004338394794,
+      "grad_norm": 0.18641281127929688,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 3941
+    },
+    {
+      "epoch": 5.344360086767896,
+      "grad_norm": 0.17405757308006287,
+      "learning_rate": 0.0001,
+      "loss": 0.1309,
+      "step": 3942
+    },
+    {
+      "epoch": 5.3457158351409975,
+      "grad_norm": 0.17762739956378937,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 3943
+    },
+    {
+      "epoch": 5.3470715835141,
+      "grad_norm": 0.1969188004732132,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 3944
+    },
+    {
+      "epoch": 5.348427331887201,
+      "grad_norm": 0.1781027466058731,
+      "learning_rate": 0.0001,
+      "loss": 0.1365,
+      "step": 3945
+    },
+    {
+      "epoch": 5.349783080260304,
+      "grad_norm": 0.25040480494499207,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 3946
+    },
+    {
+      "epoch": 5.351138828633406,
+      "grad_norm": 0.21542155742645264,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 3947
+    },
+    {
+      "epoch": 5.3524945770065075,
+      "grad_norm": 0.20837867259979248,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 3948
+    },
+    {
+      "epoch": 5.35385032537961,
+      "grad_norm": 0.19424563646316528,
+      "learning_rate": 0.0001,
+      "loss": 0.1182,
+      "step": 3949
+    },
+    {
+      "epoch": 5.355206073752711,
+      "grad_norm": 0.20837070047855377,
+      "learning_rate": 0.0001,
+      "loss": 0.1338,
+      "step": 3950
+    },
+    {
+      "epoch": 5.356561822125814,
+      "grad_norm": 0.2176591455936432,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 3951
+    },
+    {
+      "epoch": 5.357917570498915,
+      "grad_norm": 0.1889396607875824,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 3952
+    },
+    {
+      "epoch": 5.359273318872018,
+      "grad_norm": 0.186717689037323,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "step": 3953
+    },
+    {
+      "epoch": 5.360629067245119,
+      "grad_norm": 0.20618684589862823,
+      "learning_rate": 0.0001,
+      "loss": 0.1401,
+      "step": 3954
+    },
+    {
+      "epoch": 5.3619848156182215,
+      "grad_norm": 0.18949072062969208,
+      "learning_rate": 0.0001,
+      "loss": 0.1318,
+      "step": 3955
+    },
+    {
+      "epoch": 5.363340563991323,
+      "grad_norm": 0.17766791582107544,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 3956
+    },
+    {
+      "epoch": 5.364696312364425,
+      "grad_norm": 0.18118247389793396,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 3957
+    },
+    {
+      "epoch": 5.366052060737527,
+      "grad_norm": 0.21373578906059265,
+      "learning_rate": 0.0001,
+      "loss": 0.1443,
+      "step": 3958
+    },
+    {
+      "epoch": 5.367407809110629,
+      "grad_norm": 0.25244903564453125,
+      "learning_rate": 0.0001,
+      "loss": 0.1331,
+      "step": 3959
+    },
+    {
+      "epoch": 5.368763557483731,
+      "grad_norm": 0.20814213156700134,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 3960
+    },
+    {
+      "epoch": 5.370119305856833,
+      "grad_norm": 0.1908901035785675,
+      "learning_rate": 0.0001,
+      "loss": 0.1308,
+      "step": 3961
+    },
+    {
+      "epoch": 5.3714750542299345,
+      "grad_norm": 0.18240942060947418,
+      "learning_rate": 0.0001,
+      "loss": 0.1235,
+      "step": 3962
+    },
+    {
+      "epoch": 5.372830802603037,
+      "grad_norm": 0.23230840265750885,
+      "learning_rate": 0.0001,
+      "loss": 0.1426,
+      "step": 3963
+    },
+    {
+      "epoch": 5.374186550976139,
+      "grad_norm": 0.19476883113384247,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 3964
+    },
+    {
+      "epoch": 5.375542299349241,
+      "grad_norm": 0.20221437513828278,
+      "learning_rate": 0.0001,
+      "loss": 0.1329,
+      "step": 3965
+    },
+    {
+      "epoch": 5.376898047722343,
+      "grad_norm": 0.20423078536987305,
+      "learning_rate": 0.0001,
+      "loss": 0.132,
+      "step": 3966
+    },
+    {
+      "epoch": 5.378253796095445,
+      "grad_norm": 0.19254781305789948,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3967
+    },
+    {
+      "epoch": 5.379609544468547,
+      "grad_norm": 0.19743528962135315,
+      "learning_rate": 0.0001,
+      "loss": 0.1462,
+      "step": 3968
+    },
+    {
+      "epoch": 5.3809652928416485,
+      "grad_norm": 0.17651034891605377,
+      "learning_rate": 0.0001,
+      "loss": 0.121,
+      "step": 3969
+    },
+    {
+      "epoch": 5.382321041214751,
+      "grad_norm": 0.2094426304101944,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 3970
+    },
+    {
+      "epoch": 5.383676789587852,
+      "grad_norm": 0.18866796791553497,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 3971
+    },
+    {
+      "epoch": 5.385032537960955,
+      "grad_norm": 0.251818984746933,
+      "learning_rate": 0.0001,
+      "loss": 0.1473,
+      "step": 3972
+    },
+    {
+      "epoch": 5.386388286334056,
+      "grad_norm": 0.18851402401924133,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 3973
+    },
+    {
+      "epoch": 5.387744034707159,
+      "grad_norm": 0.20244912803173065,
+      "learning_rate": 0.0001,
+      "loss": 0.1537,
+      "step": 3974
+    },
+    {
+      "epoch": 5.38909978308026,
+      "grad_norm": 0.21718864142894745,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 3975
+    },
+    {
+      "epoch": 5.390455531453362,
+      "grad_norm": 0.17539137601852417,
+      "learning_rate": 0.0001,
+      "loss": 0.1355,
+      "step": 3976
+    },
+    {
+      "epoch": 5.391811279826464,
+      "grad_norm": 0.19120380282402039,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "step": 3977
+    },
+    {
+      "epoch": 5.393167028199566,
+      "grad_norm": 0.15712253749370575,
+      "learning_rate": 0.0001,
+      "loss": 0.1211,
+      "step": 3978
+    },
+    {
+      "epoch": 5.394522776572668,
+      "grad_norm": 0.18818052113056183,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3979
+    },
+    {
+      "epoch": 5.39587852494577,
+      "grad_norm": 0.18515652418136597,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 3980
+    },
+    {
+      "epoch": 5.397234273318872,
+      "grad_norm": 0.19622701406478882,
+      "learning_rate": 0.0001,
+      "loss": 0.1317,
+      "step": 3981
+    },
+    {
+      "epoch": 5.398590021691974,
+      "grad_norm": 0.1767326295375824,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 3982
+    },
+    {
+      "epoch": 5.3999457700650755,
+      "grad_norm": 0.19720593094825745,
+      "learning_rate": 0.0001,
+      "loss": 0.1325,
+      "step": 3983
+    },
+    {
+      "epoch": 5.401301518438178,
+      "grad_norm": 0.17478252947330475,
+      "learning_rate": 0.0001,
+      "loss": 0.1257,
+      "step": 3984
+    },
+    {
+      "epoch": 5.40265726681128,
+      "grad_norm": 0.20106370747089386,
+      "learning_rate": 0.0001,
+      "loss": 0.1353,
+      "step": 3985
+    },
+    {
+      "epoch": 5.404013015184382,
+      "grad_norm": 0.20802229642868042,
+      "learning_rate": 0.0001,
+      "loss": 0.135,
+      "step": 3986
+    },
+    {
+      "epoch": 5.405368763557484,
+      "grad_norm": 0.22888657450675964,
+      "learning_rate": 0.0001,
+      "loss": 0.1468,
+      "step": 3987
+    },
+    {
+      "epoch": 5.406724511930586,
+      "grad_norm": 0.20527762174606323,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 3988
+    },
+    {
+      "epoch": 5.408080260303688,
+      "grad_norm": 0.1980275958776474,
+      "learning_rate": 0.0001,
+      "loss": 0.1285,
+      "step": 3989
+    },
+    {
+      "epoch": 5.409436008676789,
+      "grad_norm": 0.18148015439510345,
+      "learning_rate": 0.0001,
+      "loss": 0.1279,
+      "step": 3990
+    },
+    {
+      "epoch": 5.410791757049892,
+      "grad_norm": 0.18459658324718475,
+      "learning_rate": 0.0001,
+      "loss": 0.1372,
+      "step": 3991
+    },
+    {
+      "epoch": 5.412147505422993,
+      "grad_norm": 0.24354663491249084,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 3992
+    },
+    {
+      "epoch": 5.413503253796096,
+      "grad_norm": 0.18468743562698364,
+      "learning_rate": 0.0001,
+      "loss": 0.1375,
+      "step": 3993
+    },
+    {
+      "epoch": 5.414859002169197,
+      "grad_norm": 0.2016223967075348,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 3994
+    },
+    {
+      "epoch": 5.4162147505422995,
+      "grad_norm": 0.19039924442768097,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 3995
+    },
+    {
+      "epoch": 5.417570498915401,
+      "grad_norm": 0.1783359944820404,
+      "learning_rate": 0.0001,
+      "loss": 0.1288,
+      "step": 3996
+    },
+    {
+      "epoch": 5.418926247288503,
+      "grad_norm": 0.23778297007083893,
+      "learning_rate": 0.0001,
+      "loss": 0.1428,
+      "step": 3997
+    },
+    {
+      "epoch": 5.420281995661605,
+      "grad_norm": 0.20209278166294098,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 3998
+    },
+    {
+      "epoch": 5.421637744034707,
+      "grad_norm": 0.17456962168216705,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 3999
+    },
+    {
+      "epoch": 5.422993492407809,
+      "grad_norm": 0.16228465735912323,
+      "learning_rate": 0.0001,
+      "loss": 0.1287,
+      "step": 4000
+    },
+    {
+      "epoch": 5.422993492407809,
+      "eval_loss": 0.19098033010959625,
+      "eval_runtime": 589.8477,
+      "eval_samples_per_second": 4.689,
+      "eval_steps_per_second": 1.173,
+      "step": 4000
+    },
+    {
+      "epoch": 5.424349240780911,
+      "grad_norm": 0.21068115532398224,
+      "learning_rate": 0.0001,
+      "loss": 0.1436,
+      "step": 4001
+    },
+    {
+      "epoch": 5.4257049891540134,
+      "grad_norm": 0.19872987270355225,
+      "learning_rate": 0.0001,
+      "loss": 0.1412,
+      "step": 4002
+    },
+    {
+      "epoch": 5.427060737527115,
+      "grad_norm": 0.1796189397573471,
+      "learning_rate": 0.0001,
+      "loss": 0.1427,
+      "step": 4003
+    },
+    {
+      "epoch": 5.428416485900217,
+      "grad_norm": 0.20879217982292175,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 4004
+    },
+    {
+      "epoch": 5.429772234273319,
+      "grad_norm": 0.16824156045913696,
+      "learning_rate": 0.0001,
+      "loss": 0.1311,
+      "step": 4005
+    },
+    {
+      "epoch": 5.431127982646421,
+      "grad_norm": 0.19574154913425446,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 4006
+    },
+    {
+      "epoch": 5.432483731019523,
+      "grad_norm": 0.18009383976459503,
+      "learning_rate": 0.0001,
+      "loss": 0.1223,
+      "step": 4007
+    },
+    {
+      "epoch": 5.433839479392625,
+      "grad_norm": 0.1906512975692749,
+      "learning_rate": 0.0001,
+      "loss": 0.1308,
+      "step": 4008
+    },
+    {
+      "epoch": 5.4351952277657265,
+      "grad_norm": 0.2073734998703003,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4009
+    },
+    {
+      "epoch": 5.436550976138829,
+      "grad_norm": 0.19430607557296753,
+      "learning_rate": 0.0001,
+      "loss": 0.1324,
+      "step": 4010
+    },
+    {
+      "epoch": 5.43790672451193,
+      "grad_norm": 0.19753706455230713,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 4011
+    },
+    {
+      "epoch": 5.439262472885033,
+      "grad_norm": 0.21959741413593292,
+      "learning_rate": 0.0001,
+      "loss": 0.1547,
+      "step": 4012
+    },
+    {
+      "epoch": 5.440618221258134,
+      "grad_norm": 0.2150670886039734,
+      "learning_rate": 0.0001,
+      "loss": 0.1375,
+      "step": 4013
+    },
+    {
+      "epoch": 5.441973969631237,
+      "grad_norm": 0.21116280555725098,
+      "learning_rate": 0.0001,
+      "loss": 0.1356,
+      "step": 4014
+    },
+    {
+      "epoch": 5.443329718004338,
+      "grad_norm": 0.21926277875900269,
+      "learning_rate": 0.0001,
+      "loss": 0.138,
+      "step": 4015
+    },
+    {
+      "epoch": 5.44468546637744,
+      "grad_norm": 0.20461031794548035,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 4016
+    },
+    {
+      "epoch": 5.446041214750542,
+      "grad_norm": 0.19065773487091064,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 4017
+    },
+    {
+      "epoch": 5.447396963123644,
+      "grad_norm": 0.1985783874988556,
+      "learning_rate": 0.0001,
+      "loss": 0.1546,
+      "step": 4018
+    },
+    {
+      "epoch": 5.448752711496747,
+      "grad_norm": 0.19591084122657776,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 4019
+    },
+    {
+      "epoch": 5.450108459869848,
+      "grad_norm": 0.203099325299263,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 4020
+    },
+    {
+      "epoch": 5.45146420824295,
+      "grad_norm": 0.19864605367183685,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 4021
+    },
+    {
+      "epoch": 5.452819956616052,
+      "grad_norm": 0.19800104200839996,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 4022
+    },
+    {
+      "epoch": 5.454175704989154,
+      "grad_norm": 0.19500888884067535,
+      "learning_rate": 0.0001,
+      "loss": 0.1223,
+      "step": 4023
+    },
+    {
+      "epoch": 5.455531453362256,
+      "grad_norm": 0.19037114083766937,
+      "learning_rate": 0.0001,
+      "loss": 0.127,
+      "step": 4024
+    },
+    {
+      "epoch": 5.456887201735358,
+      "grad_norm": 0.1756601631641388,
+      "learning_rate": 0.0001,
+      "loss": 0.1174,
+      "step": 4025
+    },
+    {
+      "epoch": 5.45824295010846,
+      "grad_norm": 0.21155746281147003,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 4026
+    },
+    {
+      "epoch": 5.459598698481562,
+      "grad_norm": 0.19837328791618347,
+      "learning_rate": 0.0001,
+      "loss": 0.1312,
+      "step": 4027
+    },
+    {
+      "epoch": 5.460954446854664,
+      "grad_norm": 0.20850351452827454,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 4028
+    },
+    {
+      "epoch": 5.462310195227766,
+      "grad_norm": 0.20173344016075134,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 4029
+    },
+    {
+      "epoch": 5.463665943600867,
+      "grad_norm": 0.18221519887447357,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "step": 4030
+    },
+    {
+      "epoch": 5.46502169197397,
+      "grad_norm": 0.18215306103229523,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 4031
+    },
+    {
+      "epoch": 5.466377440347071,
+      "grad_norm": 0.22437463700771332,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 4032
+    },
+    {
+      "epoch": 5.467733188720174,
+      "grad_norm": 0.19852352142333984,
+      "learning_rate": 0.0001,
+      "loss": 0.1483,
+      "step": 4033
+    },
+    {
+      "epoch": 5.469088937093275,
+      "grad_norm": 0.18465155363082886,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 4034
+    },
+    {
+      "epoch": 5.4704446854663775,
+      "grad_norm": 0.21004405617713928,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 4035
+    },
+    {
+      "epoch": 5.471800433839479,
+      "grad_norm": 0.1905849426984787,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 4036
+    },
+    {
+      "epoch": 5.473156182212581,
+      "grad_norm": 0.19453226029872894,
+      "learning_rate": 0.0001,
+      "loss": 0.1491,
+      "step": 4037
+    },
+    {
+      "epoch": 5.474511930585683,
+      "grad_norm": 0.233594611287117,
+      "learning_rate": 0.0001,
+      "loss": 0.1402,
+      "step": 4038
+    },
+    {
+      "epoch": 5.475867678958785,
+      "grad_norm": 0.18251436948776245,
+      "learning_rate": 0.0001,
+      "loss": 0.1308,
+      "step": 4039
+    },
+    {
+      "epoch": 5.477223427331888,
+      "grad_norm": 0.1808748096227646,
+      "learning_rate": 0.0001,
+      "loss": 0.1305,
+      "step": 4040
+    },
+    {
+      "epoch": 5.478579175704989,
+      "grad_norm": 0.2097751647233963,
+      "learning_rate": 0.0001,
+      "loss": 0.1359,
+      "step": 4041
+    },
+    {
+      "epoch": 5.4799349240780915,
+      "grad_norm": 0.2144276648759842,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 4042
+    },
+    {
+      "epoch": 5.481290672451193,
+      "grad_norm": 0.18750561773777008,
+      "learning_rate": 0.0001,
+      "loss": 0.1373,
+      "step": 4043
+    },
+    {
+      "epoch": 5.482646420824295,
+      "grad_norm": 0.188527911901474,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 4044
+    },
+    {
+      "epoch": 5.484002169197397,
+      "grad_norm": 0.19539350271224976,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 4045
+    },
+    {
+      "epoch": 5.485357917570499,
+      "grad_norm": 0.1595081388950348,
+      "learning_rate": 0.0001,
+      "loss": 0.1321,
+      "step": 4046
+    },
+    {
+      "epoch": 5.486713665943601,
+      "grad_norm": 0.2283356785774231,
+      "learning_rate": 0.0001,
+      "loss": 0.1449,
+      "step": 4047
+    },
+    {
+      "epoch": 5.488069414316703,
+      "grad_norm": 0.19843368232250214,
+      "learning_rate": 0.0001,
+      "loss": 0.1431,
+      "step": 4048
+    },
+    {
+      "epoch": 5.4894251626898045,
+      "grad_norm": 0.206586092710495,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 4049
+    },
+    {
+      "epoch": 5.490780911062907,
+      "grad_norm": 0.1979714184999466,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 4050
+    },
+    {
+      "epoch": 5.492136659436008,
+      "grad_norm": 0.18403495848178864,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 4051
+    },
+    {
+      "epoch": 5.493492407809111,
+      "grad_norm": 0.19826795160770416,
+      "learning_rate": 0.0001,
+      "loss": 0.1446,
+      "step": 4052
+    },
+    {
+      "epoch": 5.494848156182212,
+      "grad_norm": 0.1908699870109558,
+      "learning_rate": 0.0001,
+      "loss": 0.1459,
+      "step": 4053
+    },
+    {
+      "epoch": 5.496203904555315,
+      "grad_norm": 0.2052282840013504,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 4054
+    },
+    {
+      "epoch": 5.497559652928416,
+      "grad_norm": 0.1749231368303299,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 4055
+    },
+    {
+      "epoch": 5.4989154013015185,
+      "grad_norm": 0.19660651683807373,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 4056
+    },
+    {
+      "epoch": 5.500271149674621,
+      "grad_norm": 0.20411533117294312,
+      "learning_rate": 0.0001,
+      "loss": 0.1415,
+      "step": 4057
+    },
+    {
+      "epoch": 5.501626898047722,
+      "grad_norm": 0.17525528371334076,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 4058
+    },
+    {
+      "epoch": 5.502982646420824,
+      "grad_norm": 0.22735746204853058,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 4059
+    },
+    {
+      "epoch": 5.504338394793926,
+      "grad_norm": 0.1785789132118225,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "step": 4060
+    },
+    {
+      "epoch": 5.5056941431670285,
+      "grad_norm": 0.18217726051807404,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 4061
+    },
+    {
+      "epoch": 5.50704989154013,
+      "grad_norm": 0.17354296147823334,
+      "learning_rate": 0.0001,
+      "loss": 0.1297,
+      "step": 4062
+    },
+    {
+      "epoch": 5.508405639913232,
+      "grad_norm": 0.6741007566452026,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 4063
+    },
+    {
+      "epoch": 5.509761388286334,
+      "grad_norm": 0.19516399502754211,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 4064
+    },
+    {
+      "epoch": 5.511117136659436,
+      "grad_norm": 0.22792811691761017,
+      "learning_rate": 0.0001,
+      "loss": 0.1529,
+      "step": 4065
+    },
+    {
+      "epoch": 5.512472885032538,
+      "grad_norm": 0.19647610187530518,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "step": 4066
+    },
+    {
+      "epoch": 5.51382863340564,
+      "grad_norm": 0.20056557655334473,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 4067
+    },
+    {
+      "epoch": 5.515184381778742,
+      "grad_norm": 0.19963711500167847,
+      "learning_rate": 0.0001,
+      "loss": 0.1442,
+      "step": 4068
+    },
+    {
+      "epoch": 5.516540130151844,
+      "grad_norm": 0.19459176063537598,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 4069
+    },
+    {
+      "epoch": 5.5178958785249455,
+      "grad_norm": 0.19609899818897247,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 4070
+    },
+    {
+      "epoch": 5.519251626898048,
+      "grad_norm": 0.17973268032073975,
+      "learning_rate": 0.0001,
+      "loss": 0.1407,
+      "step": 4071
+    },
+    {
+      "epoch": 5.520607375271149,
+      "grad_norm": 0.22362099587917328,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 4072
+    },
+    {
+      "epoch": 5.521963123644252,
+      "grad_norm": 0.1736082285642624,
+      "learning_rate": 0.0001,
+      "loss": 0.1377,
+      "step": 4073
+    },
+    {
+      "epoch": 5.523318872017354,
+      "grad_norm": 0.22075651586055756,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 4074
+    },
+    {
+      "epoch": 5.5246746203904555,
+      "grad_norm": 0.19705471396446228,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 4075
+    },
+    {
+      "epoch": 5.526030368763557,
+      "grad_norm": 0.17228716611862183,
+      "learning_rate": 0.0001,
+      "loss": 0.1274,
+      "step": 4076
+    },
+    {
+      "epoch": 5.527386117136659,
+      "grad_norm": 0.2133665531873703,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 4077
+    },
+    {
+      "epoch": 5.528741865509762,
+      "grad_norm": 0.200433149933815,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 4078
+    },
+    {
+      "epoch": 5.530097613882863,
+      "grad_norm": 0.21091310679912567,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 4079
+    },
+    {
+      "epoch": 5.531453362255966,
+      "grad_norm": 0.21932649612426758,
+      "learning_rate": 0.0001,
+      "loss": 0.1235,
+      "step": 4080
+    },
+    {
+      "epoch": 5.532809110629067,
+      "grad_norm": 0.21369759738445282,
+      "learning_rate": 0.0001,
+      "loss": 0.1422,
+      "step": 4081
+    },
+    {
+      "epoch": 5.5341648590021695,
+      "grad_norm": 0.21681520342826843,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 4082
+    },
+    {
+      "epoch": 5.535520607375271,
+      "grad_norm": 0.19890695810317993,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 4083
+    },
+    {
+      "epoch": 5.536876355748373,
+      "grad_norm": 0.18213006854057312,
+      "learning_rate": 0.0001,
+      "loss": 0.1306,
+      "step": 4084
+    },
+    {
+      "epoch": 5.538232104121475,
+      "grad_norm": 0.19088760018348694,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4085
+    },
+    {
+      "epoch": 5.539587852494577,
+      "grad_norm": 0.22561179101467133,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 4086
+    },
+    {
+      "epoch": 5.540943600867679,
+      "grad_norm": 0.18997599184513092,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 4087
+    },
+    {
+      "epoch": 5.542299349240781,
+      "grad_norm": 0.1746479570865631,
+      "learning_rate": 0.0001,
+      "loss": 0.132,
+      "step": 4088
+    },
+    {
+      "epoch": 5.5436550976138825,
+      "grad_norm": 0.18703065812587738,
+      "learning_rate": 0.0001,
+      "loss": 0.1311,
+      "step": 4089
+    },
+    {
+      "epoch": 5.545010845986985,
+      "grad_norm": 0.19578547775745392,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 4090
+    },
+    {
+      "epoch": 5.546366594360086,
+      "grad_norm": 0.18849121034145355,
+      "learning_rate": 0.0001,
+      "loss": 0.1357,
+      "step": 4091
+    },
+    {
+      "epoch": 5.547722342733189,
+      "grad_norm": 0.17433133721351624,
+      "learning_rate": 0.0001,
+      "loss": 0.1351,
+      "step": 4092
+    },
+    {
+      "epoch": 5.54907809110629,
+      "grad_norm": 0.1956493854522705,
+      "learning_rate": 0.0001,
+      "loss": 0.1253,
+      "step": 4093
+    },
+    {
+      "epoch": 5.550433839479393,
+      "grad_norm": 0.17207588255405426,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 4094
+    },
+    {
+      "epoch": 5.551789587852495,
+      "grad_norm": 0.19887284934520721,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 4095
+    },
+    {
+      "epoch": 5.5531453362255965,
+      "grad_norm": 0.17223981022834778,
+      "learning_rate": 0.0001,
+      "loss": 0.1307,
+      "step": 4096
+    },
+    {
+      "epoch": 5.554501084598699,
+      "grad_norm": 0.188663050532341,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 4097
+    },
+    {
+      "epoch": 5.5558568329718,
+      "grad_norm": 0.18916387856006622,
+      "learning_rate": 0.0001,
+      "loss": 0.1325,
+      "step": 4098
+    },
+    {
+      "epoch": 5.557212581344903,
+      "grad_norm": 0.21452836692333221,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 4099
+    },
+    {
+      "epoch": 5.558568329718004,
+      "grad_norm": 0.2171151340007782,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 4100
+    },
+    {
+      "epoch": 5.559924078091107,
+      "grad_norm": 0.23422391712665558,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 4101
+    },
+    {
+      "epoch": 5.561279826464208,
+      "grad_norm": 0.25326645374298096,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4102
+    },
+    {
+      "epoch": 5.56263557483731,
+      "grad_norm": 0.21761400997638702,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 4103
+    },
+    {
+      "epoch": 5.563991323210412,
+      "grad_norm": 0.21369175612926483,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 4104
+    },
+    {
+      "epoch": 5.565347071583514,
+      "grad_norm": 0.2586635947227478,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 4105
+    },
+    {
+      "epoch": 5.566702819956616,
+      "grad_norm": 0.1835278570652008,
+      "learning_rate": 0.0001,
+      "loss": 0.1366,
+      "step": 4106
+    },
+    {
+      "epoch": 5.568058568329718,
+      "grad_norm": 0.18810692429542542,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4107
+    },
+    {
+      "epoch": 5.56941431670282,
+      "grad_norm": 0.18305975198745728,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 4108
+    },
+    {
+      "epoch": 5.570770065075922,
+      "grad_norm": 0.19861789047718048,
+      "learning_rate": 0.0001,
+      "loss": 0.131,
+      "step": 4109
+    },
+    {
+      "epoch": 5.5721258134490235,
+      "grad_norm": 0.1906837522983551,
+      "learning_rate": 0.0001,
+      "loss": 0.1331,
+      "step": 4110
+    },
+    {
+      "epoch": 5.573481561822126,
+      "grad_norm": 0.18296459317207336,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 4111
+    },
+    {
+      "epoch": 5.574837310195228,
+      "grad_norm": 0.22513502836227417,
+      "learning_rate": 0.0001,
+      "loss": 0.1445,
+      "step": 4112
+    },
+    {
+      "epoch": 5.57619305856833,
+      "grad_norm": 0.17728474736213684,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 4113
+    },
+    {
+      "epoch": 5.577548806941431,
+      "grad_norm": 0.16828052699565887,
+      "learning_rate": 0.0001,
+      "loss": 0.1276,
+      "step": 4114
+    },
+    {
+      "epoch": 5.5789045553145336,
+      "grad_norm": 0.19392414391040802,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 4115
+    },
+    {
+      "epoch": 5.580260303687636,
+      "grad_norm": 0.17794354259967804,
+      "learning_rate": 0.0001,
+      "loss": 0.1434,
+      "step": 4116
+    },
+    {
+      "epoch": 5.581616052060737,
+      "grad_norm": 0.16830387711524963,
+      "learning_rate": 0.0001,
+      "loss": 0.1383,
+      "step": 4117
+    },
+    {
+      "epoch": 5.58297180043384,
+      "grad_norm": 0.1720474511384964,
+      "learning_rate": 0.0001,
+      "loss": 0.1238,
+      "step": 4118
+    },
+    {
+      "epoch": 5.584327548806941,
+      "grad_norm": 0.16751176118850708,
+      "learning_rate": 0.0001,
+      "loss": 0.1237,
+      "step": 4119
+    },
+    {
+      "epoch": 5.585683297180044,
+      "grad_norm": 0.19838947057724,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 4120
+    },
+    {
+      "epoch": 5.587039045553145,
+      "grad_norm": 0.23791953921318054,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 4121
+    },
+    {
+      "epoch": 5.5883947939262475,
+      "grad_norm": 0.2089068740606308,
+      "learning_rate": 0.0001,
+      "loss": 0.1439,
+      "step": 4122
+    },
+    {
+      "epoch": 5.589750542299349,
+      "grad_norm": 0.18199589848518372,
+      "learning_rate": 0.0001,
+      "loss": 0.1461,
+      "step": 4123
+    },
+    {
+      "epoch": 5.591106290672451,
+      "grad_norm": 0.20078489184379578,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4124
+    },
+    {
+      "epoch": 5.592462039045553,
+      "grad_norm": 0.20157070457935333,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 4125
+    },
+    {
+      "epoch": 5.593817787418655,
+      "grad_norm": 0.19163921475410461,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 4126
+    },
+    {
+      "epoch": 5.595173535791757,
+      "grad_norm": 0.18445919454097748,
+      "learning_rate": 0.0001,
+      "loss": 0.1324,
+      "step": 4127
+    },
+    {
+      "epoch": 5.596529284164859,
+      "grad_norm": 0.20193932950496674,
+      "learning_rate": 0.0001,
+      "loss": 0.159,
+      "step": 4128
+    },
+    {
+      "epoch": 5.597885032537961,
+      "grad_norm": 0.19945156574249268,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 4129
+    },
+    {
+      "epoch": 5.599240780911063,
+      "grad_norm": 0.18226633965969086,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 4130
+    },
+    {
+      "epoch": 5.600596529284164,
+      "grad_norm": 0.20945732295513153,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4131
+    },
+    {
+      "epoch": 5.601952277657267,
+      "grad_norm": 0.20842674374580383,
+      "learning_rate": 0.0001,
+      "loss": 0.1536,
+      "step": 4132
+    },
+    {
+      "epoch": 5.603308026030369,
+      "grad_norm": 0.20453740656375885,
+      "learning_rate": 0.0001,
+      "loss": 0.1515,
+      "step": 4133
+    },
+    {
+      "epoch": 5.604663774403471,
+      "grad_norm": 0.19748592376708984,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 4134
+    },
+    {
+      "epoch": 5.606019522776573,
+      "grad_norm": 0.17779751121997833,
+      "learning_rate": 0.0001,
+      "loss": 0.1378,
+      "step": 4135
+    },
+    {
+      "epoch": 5.6073752711496745,
+      "grad_norm": 0.21358351409435272,
+      "learning_rate": 0.0001,
+      "loss": 0.1471,
+      "step": 4136
+    },
+    {
+      "epoch": 5.608731019522777,
+      "grad_norm": 0.20476947724819183,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 4137
+    },
+    {
+      "epoch": 5.610086767895878,
+      "grad_norm": 0.20931439101696014,
+      "learning_rate": 0.0001,
+      "loss": 0.1394,
+      "step": 4138
+    },
+    {
+      "epoch": 5.611442516268981,
+      "grad_norm": 0.20256537199020386,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 4139
+    },
+    {
+      "epoch": 5.612798264642082,
+      "grad_norm": 0.1749372035264969,
+      "learning_rate": 0.0001,
+      "loss": 0.1263,
+      "step": 4140
+    },
+    {
+      "epoch": 5.614154013015185,
+      "grad_norm": 0.16989728808403015,
+      "learning_rate": 0.0001,
+      "loss": 0.1147,
+      "step": 4141
+    },
+    {
+      "epoch": 5.615509761388286,
+      "grad_norm": 0.21411065757274628,
+      "learning_rate": 0.0001,
+      "loss": 0.1465,
+      "step": 4142
+    },
+    {
+      "epoch": 5.616865509761388,
+      "grad_norm": 0.2342263013124466,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 4143
+    },
+    {
+      "epoch": 5.61822125813449,
+      "grad_norm": 0.20519760251045227,
+      "learning_rate": 0.0001,
+      "loss": 0.1494,
+      "step": 4144
+    },
+    {
+      "epoch": 5.619577006507592,
+      "grad_norm": 0.20096264779567719,
+      "learning_rate": 0.0001,
+      "loss": 0.1384,
+      "step": 4145
+    },
+    {
+      "epoch": 5.620932754880694,
+      "grad_norm": 0.17452499270439148,
+      "learning_rate": 0.0001,
+      "loss": 0.1299,
+      "step": 4146
+    },
+    {
+      "epoch": 5.622288503253796,
+      "grad_norm": 0.22272078692913055,
+      "learning_rate": 0.0001,
+      "loss": 0.1396,
+      "step": 4147
+    },
+    {
+      "epoch": 5.623644251626898,
+      "grad_norm": 0.22798803448677063,
+      "learning_rate": 0.0001,
+      "loss": 0.1605,
+      "step": 4148
+    },
+    {
+      "epoch": 5.625,
+      "grad_norm": 0.20973201096057892,
+      "learning_rate": 0.0001,
+      "loss": 0.154,
+      "step": 4149
+    },
+    {
+      "epoch": 5.626355748373102,
+      "grad_norm": 0.19370020925998688,
+      "learning_rate": 0.0001,
+      "loss": 0.1416,
+      "step": 4150
+    },
+    {
+      "epoch": 5.627711496746204,
+      "grad_norm": 0.20400120317935944,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 4151
+    },
+    {
+      "epoch": 5.629067245119306,
+      "grad_norm": 0.1900358945131302,
+      "learning_rate": 0.0001,
+      "loss": 0.1372,
+      "step": 4152
+    },
+    {
+      "epoch": 5.630422993492408,
+      "grad_norm": 0.20619700849056244,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 4153
+    },
+    {
+      "epoch": 5.63177874186551,
+      "grad_norm": 0.20146150887012482,
+      "learning_rate": 0.0001,
+      "loss": 0.1338,
+      "step": 4154
+    },
+    {
+      "epoch": 5.633134490238612,
+      "grad_norm": 0.20281502604484558,
+      "learning_rate": 0.0001,
+      "loss": 0.1549,
+      "step": 4155
+    },
+    {
+      "epoch": 5.634490238611714,
+      "grad_norm": 0.18210162222385406,
+      "learning_rate": 0.0001,
+      "loss": 0.1363,
+      "step": 4156
+    },
+    {
+      "epoch": 5.635845986984815,
+      "grad_norm": 0.19072775542736053,
+      "learning_rate": 0.0001,
+      "loss": 0.1314,
+      "step": 4157
+    },
+    {
+      "epoch": 5.637201735357918,
+      "grad_norm": 0.22012245655059814,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 4158
+    },
+    {
+      "epoch": 5.638557483731019,
+      "grad_norm": 0.19993805885314941,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 4159
+    },
+    {
+      "epoch": 5.639913232104122,
+      "grad_norm": 0.20129556953907013,
+      "learning_rate": 0.0001,
+      "loss": 0.1389,
+      "step": 4160
+    },
+    {
+      "epoch": 5.641268980477223,
+      "grad_norm": 0.18830668926239014,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4161
+    },
+    {
+      "epoch": 5.6426247288503255,
+      "grad_norm": 0.21969063580036163,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 4162
+    },
+    {
+      "epoch": 5.643980477223427,
+      "grad_norm": 0.21191321313381195,
+      "learning_rate": 0.0001,
+      "loss": 0.1319,
+      "step": 4163
+    },
+    {
+      "epoch": 5.645336225596529,
+      "grad_norm": 0.17650389671325684,
+      "learning_rate": 0.0001,
+      "loss": 0.1271,
+      "step": 4164
+    },
+    {
+      "epoch": 5.646691973969631,
+      "grad_norm": 0.19436956942081451,
+      "learning_rate": 0.0001,
+      "loss": 0.1265,
+      "step": 4165
+    },
+    {
+      "epoch": 5.648047722342733,
+      "grad_norm": 0.23487330973148346,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 4166
+    },
+    {
+      "epoch": 5.649403470715836,
+      "grad_norm": 0.23343868553638458,
+      "learning_rate": 0.0001,
+      "loss": 0.137,
+      "step": 4167
+    },
+    {
+      "epoch": 5.650759219088937,
+      "grad_norm": 0.17832249402999878,
+      "learning_rate": 0.0001,
+      "loss": 0.1314,
+      "step": 4168
+    },
+    {
+      "epoch": 5.652114967462039,
+      "grad_norm": 0.16251884400844574,
+      "learning_rate": 0.0001,
+      "loss": 0.1229,
+      "step": 4169
+    },
+    {
+      "epoch": 5.653470715835141,
+      "grad_norm": 0.18108123540878296,
+      "learning_rate": 0.0001,
+      "loss": 0.1344,
+      "step": 4170
+    },
+    {
+      "epoch": 5.654826464208243,
+      "grad_norm": 0.19649073481559753,
+      "learning_rate": 0.0001,
+      "loss": 0.1313,
+      "step": 4171
+    },
+    {
+      "epoch": 5.656182212581345,
+      "grad_norm": 0.20201162993907928,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 4172
+    },
+    {
+      "epoch": 5.657537960954447,
+      "grad_norm": 0.19187133014202118,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 4173
+    },
+    {
+      "epoch": 5.658893709327549,
+      "grad_norm": 0.16954094171524048,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 4174
+    },
+    {
+      "epoch": 5.660249457700651,
+      "grad_norm": 0.20714637637138367,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 4175
+    },
+    {
+      "epoch": 5.6616052060737525,
+      "grad_norm": 0.17160820960998535,
+      "learning_rate": 0.0001,
+      "loss": 0.1266,
+      "step": 4176
+    },
+    {
+      "epoch": 5.662960954446855,
+      "grad_norm": 0.20615343749523163,
+      "learning_rate": 0.0001,
+      "loss": 0.1381,
+      "step": 4177
+    },
+    {
+      "epoch": 5.664316702819956,
+      "grad_norm": 0.17290319502353668,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 4178
+    },
+    {
+      "epoch": 5.665672451193059,
+      "grad_norm": 0.18619176745414734,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 4179
+    },
+    {
+      "epoch": 5.66702819956616,
+      "grad_norm": 0.1546507626771927,
+      "learning_rate": 0.0001,
+      "loss": 0.1275,
+      "step": 4180
+    },
+    {
+      "epoch": 5.668383947939263,
+      "grad_norm": 0.19016698002815247,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4181
+    },
+    {
+      "epoch": 5.669739696312364,
+      "grad_norm": 0.15665952861309052,
+      "learning_rate": 0.0001,
+      "loss": 0.1318,
+      "step": 4182
+    },
+    {
+      "epoch": 5.6710954446854664,
+      "grad_norm": 0.1911107897758484,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 4183
+    },
+    {
+      "epoch": 5.672451193058569,
+      "grad_norm": 0.19488364458084106,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 4184
+    },
+    {
+      "epoch": 5.67380694143167,
+      "grad_norm": 0.30131080746650696,
+      "learning_rate": 0.0001,
+      "loss": 0.1371,
+      "step": 4185
+    },
+    {
+      "epoch": 5.675162689804772,
+      "grad_norm": 0.17258575558662415,
+      "learning_rate": 0.0001,
+      "loss": 0.1315,
+      "step": 4186
+    },
+    {
+      "epoch": 5.676518438177874,
+      "grad_norm": 0.19458982348442078,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 4187
+    },
+    {
+      "epoch": 5.6778741865509765,
+      "grad_norm": 0.19604863226413727,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 4188
+    },
+    {
+      "epoch": 5.679229934924078,
+      "grad_norm": 0.1625247448682785,
+      "learning_rate": 0.0001,
+      "loss": 0.1307,
+      "step": 4189
+    },
+    {
+      "epoch": 5.68058568329718,
+      "grad_norm": 0.18186503648757935,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 4190
+    },
+    {
+      "epoch": 5.681941431670282,
+      "grad_norm": 0.1940801590681076,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 4191
+    },
+    {
+      "epoch": 5.683297180043384,
+      "grad_norm": 0.20764373242855072,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 4192
+    },
+    {
+      "epoch": 5.684652928416486,
+      "grad_norm": 0.21391674876213074,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 4193
+    },
+    {
+      "epoch": 5.686008676789588,
+      "grad_norm": 0.21641704440116882,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 4194
+    },
+    {
+      "epoch": 5.68736442516269,
+      "grad_norm": 0.19196681678295135,
+      "learning_rate": 0.0001,
+      "loss": 0.1464,
+      "step": 4195
+    },
+    {
+      "epoch": 5.688720173535792,
+      "grad_norm": 0.2166317105293274,
+      "learning_rate": 0.0001,
+      "loss": 0.1435,
+      "step": 4196
+    },
+    {
+      "epoch": 5.690075921908893,
+      "grad_norm": 0.19317077100276947,
+      "learning_rate": 0.0001,
+      "loss": 0.1487,
+      "step": 4197
+    },
+    {
+      "epoch": 5.691431670281996,
+      "grad_norm": 0.1861579716205597,
+      "learning_rate": 0.0001,
+      "loss": 0.1292,
+      "step": 4198
+    },
+    {
+      "epoch": 5.692787418655097,
+      "grad_norm": 0.2043808549642563,
+      "learning_rate": 0.0001,
+      "loss": 0.133,
+      "step": 4199
+    },
+    {
+      "epoch": 5.6941431670282,
+      "grad_norm": 0.17921476066112518,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 4200
+    },
+    {
+      "epoch": 5.695498915401301,
+      "grad_norm": 0.20104925334453583,
+      "learning_rate": 0.0001,
+      "loss": 0.1476,
+      "step": 4201
+    },
+    {
+      "epoch": 5.6968546637744035,
+      "grad_norm": 0.18092648684978485,
+      "learning_rate": 0.0001,
+      "loss": 0.1333,
+      "step": 4202
+    },
+    {
+      "epoch": 5.698210412147505,
+      "grad_norm": 0.17052188515663147,
+      "learning_rate": 0.0001,
+      "loss": 0.1164,
+      "step": 4203
+    },
+    {
+      "epoch": 5.699566160520607,
+      "grad_norm": 0.19642221927642822,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 4204
+    },
+    {
+      "epoch": 5.70092190889371,
+      "grad_norm": 0.21498674154281616,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 4205
+    },
+    {
+      "epoch": 5.702277657266811,
+      "grad_norm": 0.20739395916461945,
+      "learning_rate": 0.0001,
+      "loss": 0.15,
+      "step": 4206
+    },
+    {
+      "epoch": 5.703633405639914,
+      "grad_norm": 0.19489920139312744,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 4207
+    },
+    {
+      "epoch": 5.704989154013015,
+      "grad_norm": 0.24173659086227417,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 4208
+    },
+    {
+      "epoch": 5.7063449023861175,
+      "grad_norm": 0.18896205723285675,
+      "learning_rate": 0.0001,
+      "loss": 0.139,
+      "step": 4209
+    },
+    {
+      "epoch": 5.707700650759219,
+      "grad_norm": 0.20157521963119507,
+      "learning_rate": 0.0001,
+      "loss": 0.1405,
+      "step": 4210
+    },
+    {
+      "epoch": 5.709056399132321,
+      "grad_norm": 0.21547766029834747,
+      "learning_rate": 0.0001,
+      "loss": 0.1409,
+      "step": 4211
+    },
+    {
+      "epoch": 5.710412147505423,
+      "grad_norm": 0.17879261076450348,
+      "learning_rate": 0.0001,
+      "loss": 0.1299,
+      "step": 4212
+    },
+    {
+      "epoch": 5.711767895878525,
+      "grad_norm": 0.17945726215839386,
+      "learning_rate": 0.0001,
+      "loss": 0.1367,
+      "step": 4213
+    },
+    {
+      "epoch": 5.713123644251627,
+      "grad_norm": 0.2006184309720993,
+      "learning_rate": 0.0001,
+      "loss": 0.1504,
+      "step": 4214
+    },
+    {
+      "epoch": 5.714479392624729,
+      "grad_norm": 0.19950394332408905,
+      "learning_rate": 0.0001,
+      "loss": 0.1505,
+      "step": 4215
+    },
+    {
+      "epoch": 5.7158351409978305,
+      "grad_norm": 0.18734024465084076,
+      "learning_rate": 0.0001,
+      "loss": 0.1399,
+      "step": 4216
+    },
+    {
+      "epoch": 5.717190889370933,
+      "grad_norm": 0.186662495136261,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 4217
+    },
+    {
+      "epoch": 5.718546637744034,
+      "grad_norm": 0.21753200888633728,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 4218
+    },
+    {
+      "epoch": 5.719902386117137,
+      "grad_norm": 0.1976969987154007,
+      "learning_rate": 0.0001,
+      "loss": 0.1324,
+      "step": 4219
+    },
+    {
+      "epoch": 5.721258134490238,
+      "grad_norm": 0.18096855282783508,
+      "learning_rate": 0.0001,
+      "loss": 0.1402,
+      "step": 4220
+    },
+    {
+      "epoch": 5.722613882863341,
+      "grad_norm": 0.19840629398822784,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 4221
+    },
+    {
+      "epoch": 5.723969631236443,
+      "grad_norm": 0.18321330845355988,
+      "learning_rate": 0.0001,
+      "loss": 0.1301,
+      "step": 4222
+    },
+    {
+      "epoch": 5.7253253796095445,
+      "grad_norm": 0.23323962092399597,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 4223
+    },
+    {
+      "epoch": 5.726681127982646,
+      "grad_norm": 0.2355189323425293,
+      "learning_rate": 0.0001,
+      "loss": 0.1626,
+      "step": 4224
+    },
+    {
+      "epoch": 5.728036876355748,
+      "grad_norm": 0.18643563985824585,
+      "learning_rate": 0.0001,
+      "loss": 0.1432,
+      "step": 4225
+    },
+    {
+      "epoch": 5.729392624728851,
+      "grad_norm": 0.19526715576648712,
+      "learning_rate": 0.0001,
+      "loss": 0.1277,
+      "step": 4226
+    },
+    {
+      "epoch": 5.730748373101952,
+      "grad_norm": 0.22516371309757233,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 4227
+    },
+    {
+      "epoch": 5.7321041214750545,
+      "grad_norm": 0.19453135132789612,
+      "learning_rate": 0.0001,
+      "loss": 0.149,
+      "step": 4228
+    },
+    {
+      "epoch": 5.733459869848156,
+      "grad_norm": 0.2295314371585846,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 4229
+    },
+    {
+      "epoch": 5.734815618221258,
+      "grad_norm": 0.2256745547056198,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 4230
+    },
+    {
+      "epoch": 5.73617136659436,
+      "grad_norm": 0.18518872559070587,
+      "learning_rate": 0.0001,
+      "loss": 0.1361,
+      "step": 4231
+    },
+    {
+      "epoch": 5.737527114967462,
+      "grad_norm": 0.18905901908874512,
+      "learning_rate": 0.0001,
+      "loss": 0.1379,
+      "step": 4232
+    },
+    {
+      "epoch": 5.738882863340564,
+      "grad_norm": 0.19623228907585144,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 4233
+    },
+    {
+      "epoch": 5.740238611713666,
+      "grad_norm": 0.2039419561624527,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 4234
+    },
+    {
+      "epoch": 5.741594360086768,
+      "grad_norm": 0.1722797304391861,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 4235
+    },
+    {
+      "epoch": 5.74295010845987,
+      "grad_norm": 0.17968976497650146,
+      "learning_rate": 0.0001,
+      "loss": 0.13,
+      "step": 4236
+    },
+    {
+      "epoch": 5.7443058568329715,
+      "grad_norm": 0.2015763521194458,
+      "learning_rate": 0.0001,
+      "loss": 0.1467,
+      "step": 4237
+    },
+    {
+      "epoch": 5.745661605206074,
+      "grad_norm": 0.16787894070148468,
+      "learning_rate": 0.0001,
+      "loss": 0.1218,
+      "step": 4238
+    },
+    {
+      "epoch": 5.747017353579176,
+      "grad_norm": 0.16567623615264893,
+      "learning_rate": 0.0001,
+      "loss": 0.1335,
+      "step": 4239
+    },
+    {
+      "epoch": 5.748373101952278,
+      "grad_norm": 0.19809237122535706,
+      "learning_rate": 0.0001,
+      "loss": 0.1532,
+      "step": 4240
+    },
+    {
+      "epoch": 5.749728850325379,
+      "grad_norm": 0.19603615999221802,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 4241
+    },
+    {
+      "epoch": 5.7510845986984815,
+      "grad_norm": 0.18549062311649323,
+      "learning_rate": 0.0001,
+      "loss": 0.1264,
+      "step": 4242
+    },
+    {
+      "epoch": 5.752440347071584,
+      "grad_norm": 0.1797325313091278,
+      "learning_rate": 0.0001,
+      "loss": 0.1322,
+      "step": 4243
+    },
+    {
+      "epoch": 5.753796095444685,
+      "grad_norm": 0.2234465628862381,
+      "learning_rate": 0.0001,
+      "loss": 0.1527,
+      "step": 4244
+    },
+    {
+      "epoch": 5.755151843817788,
+      "grad_norm": 0.20542746782302856,
+      "learning_rate": 0.0001,
+      "loss": 0.1349,
+      "step": 4245
+    },
+    {
+      "epoch": 5.756507592190889,
+      "grad_norm": 0.21355338394641876,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4246
+    },
+    {
+      "epoch": 5.757863340563992,
+      "grad_norm": 0.2113415002822876,
+      "learning_rate": 0.0001,
+      "loss": 0.1348,
+      "step": 4247
+    },
+    {
+      "epoch": 5.759219088937093,
+      "grad_norm": 0.19244085252285004,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4248
+    },
+    {
+      "epoch": 5.7605748373101955,
+      "grad_norm": 0.20356008410453796,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 4249
+    },
+    {
+      "epoch": 5.761930585683297,
+      "grad_norm": 0.1918802559375763,
+      "learning_rate": 0.0001,
+      "loss": 0.1287,
+      "step": 4250
+    },
+    {
+      "epoch": 5.763286334056399,
+      "grad_norm": 0.1797376275062561,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 4251
+    },
+    {
+      "epoch": 5.764642082429501,
+      "grad_norm": 0.2170053869485855,
+      "learning_rate": 0.0001,
+      "loss": 0.1528,
+      "step": 4252
+    },
+    {
+      "epoch": 5.765997830802603,
+      "grad_norm": 0.191667839884758,
+      "learning_rate": 0.0001,
+      "loss": 0.1466,
+      "step": 4253
+    },
+    {
+      "epoch": 5.767353579175705,
+      "grad_norm": 0.16871635615825653,
+      "learning_rate": 0.0001,
+      "loss": 0.1316,
+      "step": 4254
+    },
+    {
+      "epoch": 5.768709327548807,
+      "grad_norm": 0.17284661531448364,
+      "learning_rate": 0.0001,
+      "loss": 0.134,
+      "step": 4255
+    },
+    {
+      "epoch": 5.7700650759219085,
+      "grad_norm": 0.1905890703201294,
+      "learning_rate": 0.0001,
+      "loss": 0.1402,
+      "step": 4256
+    },
+    {
+      "epoch": 5.771420824295011,
+      "grad_norm": 0.2251669466495514,
+      "learning_rate": 0.0001,
+      "loss": 0.1424,
+      "step": 4257
+    },
+    {
+      "epoch": 5.772776572668112,
+      "grad_norm": 0.2452983409166336,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 4258
+    },
+    {
+      "epoch": 5.774132321041215,
+      "grad_norm": 0.16889168322086334,
+      "learning_rate": 0.0001,
+      "loss": 0.132,
+      "step": 4259
+    },
+    {
+      "epoch": 5.775488069414317,
+      "grad_norm": 0.19196026027202606,
+      "learning_rate": 0.0001,
+      "loss": 0.1399,
+      "step": 4260
+    },
+    {
+      "epoch": 5.776843817787419,
+      "grad_norm": 0.1821107119321823,
+      "learning_rate": 0.0001,
+      "loss": 0.1453,
+      "step": 4261
+    },
+    {
+      "epoch": 5.778199566160521,
+      "grad_norm": 0.2409312129020691,
+      "learning_rate": 0.0001,
+      "loss": 0.1495,
+      "step": 4262
+    },
+    {
+      "epoch": 5.7795553145336225,
+      "grad_norm": 0.1966077983379364,
+      "learning_rate": 0.0001,
+      "loss": 0.1458,
+      "step": 4263
+    },
+    {
+      "epoch": 5.780911062906725,
+      "grad_norm": 0.1915093958377838,
+      "learning_rate": 0.0001,
+      "loss": 0.1403,
+      "step": 4264
+    },
+    {
+      "epoch": 5.782266811279826,
+      "grad_norm": 0.198311448097229,
+      "learning_rate": 0.0001,
+      "loss": 0.1396,
+      "step": 4265
+    },
+    {
+      "epoch": 5.783622559652929,
+      "grad_norm": 0.18046166002750397,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 4266
+    },
+    {
+      "epoch": 5.78497830802603,
+      "grad_norm": 0.253396600484848,
+      "learning_rate": 0.0001,
+      "loss": 0.1564,
+      "step": 4267
+    },
+    {
+      "epoch": 5.786334056399133,
+      "grad_norm": 0.19874736666679382,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 4268
+    },
+    {
+      "epoch": 5.787689804772234,
+      "grad_norm": 0.1812356859445572,
+      "learning_rate": 0.0001,
+      "loss": 0.1341,
+      "step": 4269
+    },
+    {
+      "epoch": 5.789045553145336,
+      "grad_norm": 0.2221725732088089,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 4270
+    },
+    {
+      "epoch": 5.790401301518438,
+      "grad_norm": 0.1933864951133728,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 4271
+    },
+    {
+      "epoch": 5.79175704989154,
+      "grad_norm": 0.19647273421287537,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 4272
+    },
+    {
+      "epoch": 5.793112798264642,
+      "grad_norm": 0.20964471995830536,
+      "learning_rate": 0.0001,
+      "loss": 0.1585,
+      "step": 4273
+    },
+    {
+      "epoch": 5.794468546637744,
+      "grad_norm": 0.2278592437505722,
+      "learning_rate": 0.0001,
+      "loss": 0.1482,
+      "step": 4274
+    },
+    {
+      "epoch": 5.795824295010846,
+      "grad_norm": 0.17661483585834503,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 4275
+    },
+    {
+      "epoch": 5.797180043383948,
+      "grad_norm": 0.19317185878753662,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4276
+    },
+    {
+      "epoch": 5.79853579175705,
+      "grad_norm": 0.20447707176208496,
+      "learning_rate": 0.0001,
+      "loss": 0.1541,
+      "step": 4277
+    },
+    {
+      "epoch": 5.799891540130152,
+      "grad_norm": 0.19147798418998718,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 4278
+    },
+    {
+      "epoch": 5.801247288503253,
+      "grad_norm": 0.1940334290266037,
+      "learning_rate": 0.0001,
+      "loss": 0.14,
+      "step": 4279
+    },
+    {
+      "epoch": 5.802603036876356,
+      "grad_norm": 0.16969414055347443,
+      "learning_rate": 0.0001,
+      "loss": 0.1303,
+      "step": 4280
+    },
+    {
+      "epoch": 5.803958785249458,
+      "grad_norm": 0.1775941550731659,
+      "learning_rate": 0.0001,
+      "loss": 0.1401,
+      "step": 4281
+    },
+    {
+      "epoch": 5.80531453362256,
+      "grad_norm": 0.17752115428447723,
+      "learning_rate": 0.0001,
+      "loss": 0.1422,
+      "step": 4282
+    },
+    {
+      "epoch": 5.806670281995662,
+      "grad_norm": 0.20454080402851105,
+      "learning_rate": 0.0001,
+      "loss": 0.1509,
+      "step": 4283
+    },
+    {
+      "epoch": 5.808026030368763,
+      "grad_norm": 0.19171243906021118,
+      "learning_rate": 0.0001,
+      "loss": 0.128,
+      "step": 4284
+    },
+    {
+      "epoch": 5.809381778741866,
+      "grad_norm": 0.2033306211233139,
+      "learning_rate": 0.0001,
+      "loss": 0.1255,
+      "step": 4285
+    },
+    {
+      "epoch": 5.810737527114967,
+      "grad_norm": 0.20183807611465454,
+      "learning_rate": 0.0001,
+      "loss": 0.127,
+      "step": 4286
+    },
+    {
+      "epoch": 5.81209327548807,
+      "grad_norm": 0.23862308263778687,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 4287
+    },
+    {
+      "epoch": 5.813449023861171,
+      "grad_norm": 0.19212111830711365,
+      "learning_rate": 0.0001,
+      "loss": 0.1382,
+      "step": 4288
+    },
+    {
+      "epoch": 5.8148047722342735,
+      "grad_norm": 0.21150749921798706,
+      "learning_rate": 0.0001,
+      "loss": 0.145,
+      "step": 4289
+    },
+    {
+      "epoch": 5.816160520607375,
+      "grad_norm": 0.18475106358528137,
+      "learning_rate": 0.0001,
+      "loss": 0.1457,
+      "step": 4290
+    },
+    {
+      "epoch": 5.817516268980477,
+      "grad_norm": 0.18173694610595703,
+      "learning_rate": 0.0001,
+      "loss": 0.1346,
+      "step": 4291
+    },
+    {
+      "epoch": 5.818872017353579,
+      "grad_norm": 0.1955554485321045,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 4292
+    },
+    {
+      "epoch": 5.820227765726681,
+      "grad_norm": 0.18990109860897064,
+      "learning_rate": 0.0001,
+      "loss": 0.1302,
+      "step": 4293
+    },
+    {
+      "epoch": 5.821583514099783,
+      "grad_norm": 0.2546767592430115,
+      "learning_rate": 0.0001,
+      "loss": 0.1516,
+      "step": 4294
+    },
+    {
+      "epoch": 5.822939262472885,
+      "grad_norm": 0.1767260730266571,
+      "learning_rate": 0.0001,
+      "loss": 0.1347,
+      "step": 4295
+    },
+    {
+      "epoch": 5.8242950108459866,
+      "grad_norm": 0.1851368099451065,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 4296
+    },
+    {
+      "epoch": 5.825650759219089,
+      "grad_norm": 0.2529405951499939,
+      "learning_rate": 0.0001,
+      "loss": 0.1418,
+      "step": 4297
+    },
+    {
+      "epoch": 5.827006507592191,
+      "grad_norm": 0.17316314578056335,
+      "learning_rate": 0.0001,
+      "loss": 0.1328,
+      "step": 4298
+    },
+    {
+      "epoch": 5.828362255965293,
+      "grad_norm": 0.22963330149650574,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 4299
+    },
+    {
+      "epoch": 5.829718004338395,
+      "grad_norm": 0.21335674822330475,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 4300
+    },
+    {
+      "epoch": 5.831073752711497,
+      "grad_norm": 0.18164201080799103,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4301
+    },
+    {
+      "epoch": 5.832429501084599,
+      "grad_norm": 0.18023543059825897,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4302
+    },
+    {
+      "epoch": 5.8337852494577005,
+      "grad_norm": 0.18862104415893555,
+      "learning_rate": 0.0001,
+      "loss": 0.1444,
+      "step": 4303
+    },
+    {
+      "epoch": 5.835140997830803,
+      "grad_norm": 0.1533985137939453,
+      "learning_rate": 0.0001,
+      "loss": 0.1262,
+      "step": 4304
+    },
+    {
+      "epoch": 5.836496746203904,
+      "grad_norm": 0.19740132987499237,
+      "learning_rate": 0.0001,
+      "loss": 0.1475,
+      "step": 4305
+    },
+    {
+      "epoch": 5.837852494577007,
+      "grad_norm": 0.17243118584156036,
+      "learning_rate": 0.0001,
+      "loss": 0.1182,
+      "step": 4306
+    },
+    {
+      "epoch": 5.839208242950108,
+      "grad_norm": 0.19089338183403015,
+      "learning_rate": 0.0001,
+      "loss": 0.1523,
+      "step": 4307
+    },
+    {
+      "epoch": 5.840563991323211,
+      "grad_norm": 0.1931903064250946,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 4308
+    },
+    {
+      "epoch": 5.841919739696312,
+      "grad_norm": 0.17253312468528748,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 4309
+    },
+    {
+      "epoch": 5.843275488069414,
+      "grad_norm": 0.17681695520877838,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 4310
+    },
+    {
+      "epoch": 5.844631236442516,
+      "grad_norm": 0.1737832874059677,
+      "learning_rate": 0.0001,
+      "loss": 0.1249,
+      "step": 4311
+    },
+    {
+      "epoch": 5.845986984815618,
+      "grad_norm": 0.19029942154884338,
+      "learning_rate": 0.0001,
+      "loss": 0.1368,
+      "step": 4312
+    },
+    {
+      "epoch": 5.84734273318872,
+      "grad_norm": 0.22111180424690247,
+      "learning_rate": 0.0001,
+      "loss": 0.1384,
+      "step": 4313
+    },
+    {
+      "epoch": 5.848698481561822,
+      "grad_norm": 0.21650265157222748,
+      "learning_rate": 0.0001,
+      "loss": 0.1469,
+      "step": 4314
+    },
+    {
+      "epoch": 5.8500542299349245,
+      "grad_norm": 0.2931518852710724,
+      "learning_rate": 0.0001,
+      "loss": 0.1581,
+      "step": 4315
+    },
+    {
+      "epoch": 5.851409978308026,
+      "grad_norm": 0.2092655599117279,
+      "learning_rate": 0.0001,
+      "loss": 0.1363,
+      "step": 4316
+    },
+    {
+      "epoch": 5.8527657266811275,
+      "grad_norm": 0.20854857563972473,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 4317
+    },
+    {
+      "epoch": 5.85412147505423,
+      "grad_norm": 0.21228128671646118,
+      "learning_rate": 0.0001,
+      "loss": 0.1558,
+      "step": 4318
+    },
+    {
+      "epoch": 5.855477223427332,
+      "grad_norm": 0.18482787907123566,
+      "learning_rate": 0.0001,
+      "loss": 0.1317,
+      "step": 4319
+    },
+    {
+      "epoch": 5.856832971800434,
+      "grad_norm": 0.2087964415550232,
+      "learning_rate": 0.0001,
+      "loss": 0.1333,
+      "step": 4320
+    },
+    {
+      "epoch": 5.858188720173536,
+      "grad_norm": 0.18515270948410034,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 4321
+    },
+    {
+      "epoch": 5.859544468546638,
+      "grad_norm": 0.1893758475780487,
+      "learning_rate": 0.0001,
+      "loss": 0.1233,
+      "step": 4322
+    },
+    {
+      "epoch": 5.86090021691974,
+      "grad_norm": 0.18985515832901,
+      "learning_rate": 0.0001,
+      "loss": 0.1404,
+      "step": 4323
+    },
+    {
+      "epoch": 5.862255965292841,
+      "grad_norm": 0.1986810266971588,
+      "learning_rate": 0.0001,
+      "loss": 0.138,
+      "step": 4324
+    },
+    {
+      "epoch": 5.863611713665944,
+      "grad_norm": 0.18283404409885406,
+      "learning_rate": 0.0001,
+      "loss": 0.1303,
+      "step": 4325
+    },
+    {
+      "epoch": 5.864967462039045,
+      "grad_norm": 0.18655434250831604,
+      "learning_rate": 0.0001,
+      "loss": 0.1361,
+      "step": 4326
+    },
+    {
+      "epoch": 5.866323210412148,
+      "grad_norm": 0.18271006643772125,
+      "learning_rate": 0.0001,
+      "loss": 0.1364,
+      "step": 4327
+    },
+    {
+      "epoch": 5.867678958785249,
+      "grad_norm": 0.2001565545797348,
+      "learning_rate": 0.0001,
+      "loss": 0.1557,
+      "step": 4328
+    },
+    {
+      "epoch": 5.8690347071583515,
+      "grad_norm": 0.19421501457691193,
+      "learning_rate": 0.0001,
+      "loss": 0.1376,
+      "step": 4329
+    },
+    {
+      "epoch": 5.870390455531453,
+      "grad_norm": 0.2795417606830597,
+      "learning_rate": 0.0001,
+      "loss": 0.1363,
+      "step": 4330
+    },
+    {
+      "epoch": 5.871746203904555,
+      "grad_norm": 0.208097442984581,
+      "learning_rate": 0.0001,
+      "loss": 0.1486,
+      "step": 4331
+    },
+    {
+      "epoch": 5.873101952277658,
+      "grad_norm": 0.22109094262123108,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 4332
+    },
+    {
+      "epoch": 5.874457700650759,
+      "grad_norm": 0.1619035303592682,
+      "learning_rate": 0.0001,
+      "loss": 0.1193,
+      "step": 4333
+    },
+    {
+      "epoch": 5.875813449023861,
+      "grad_norm": 0.1579362154006958,
+      "learning_rate": 0.0001,
+      "loss": 0.1201,
+      "step": 4334
+    },
+    {
+      "epoch": 5.877169197396963,
+      "grad_norm": 0.18892550468444824,
+      "learning_rate": 0.0001,
+      "loss": 0.1411,
+      "step": 4335
+    },
+    {
+      "epoch": 5.8785249457700655,
+      "grad_norm": 0.24527977406978607,
+      "learning_rate": 0.0001,
+      "loss": 0.1391,
+      "step": 4336
+    },
+    {
+      "epoch": 5.879880694143167,
+      "grad_norm": 0.22025178372859955,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 4337
+    },
+    {
+      "epoch": 5.881236442516269,
+      "grad_norm": 0.19745710492134094,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 4338
+    },
+    {
+      "epoch": 5.882592190889371,
+      "grad_norm": 0.18283306062221527,
+      "learning_rate": 0.0001,
+      "loss": 0.1519,
+      "step": 4339
+    },
+    {
+      "epoch": 5.883947939262473,
+      "grad_norm": 0.1992870718240738,
+      "learning_rate": 0.0001,
+      "loss": 0.1421,
+      "step": 4340
+    },
+    {
+      "epoch": 5.885303687635575,
+      "grad_norm": 0.21646317839622498,
+      "learning_rate": 0.0001,
+      "loss": 0.1406,
+      "step": 4341
+    },
+    {
+      "epoch": 5.886659436008677,
+      "grad_norm": 0.1882300227880478,
+      "learning_rate": 0.0001,
+      "loss": 0.1393,
+      "step": 4342
+    },
+    {
+      "epoch": 5.8880151843817785,
+      "grad_norm": 0.21877427399158478,
+      "learning_rate": 0.0001,
+      "loss": 0.1454,
+      "step": 4343
+    },
+    {
+      "epoch": 5.889370932754881,
+      "grad_norm": 0.190181702375412,
+      "learning_rate": 0.0001,
+      "loss": 0.1575,
+      "step": 4344
+    },
+    {
+      "epoch": 5.890726681127982,
+      "grad_norm": 0.17933569848537445,
+      "learning_rate": 0.0001,
+      "loss": 0.1492,
+      "step": 4345
+    },
+    {
+      "epoch": 5.892082429501085,
+      "grad_norm": 0.17628858983516693,
+      "learning_rate": 0.0001,
+      "loss": 0.1369,
+      "step": 4346
+    },
+    {
+      "epoch": 5.893438177874186,
+      "grad_norm": 0.20829370617866516,
+      "learning_rate": 0.0001,
+      "loss": 0.1485,
+      "step": 4347
+    },
+    {
+      "epoch": 5.894793926247289,
+      "grad_norm": 0.19245053827762604,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 4348
+    },
+    {
+      "epoch": 5.89614967462039,
+      "grad_norm": 0.17646130919456482,
+      "learning_rate": 0.0001,
+      "loss": 0.1396,
+      "step": 4349
+    },
+    {
+      "epoch": 5.8975054229934925,
+      "grad_norm": 0.19283893704414368,
+      "learning_rate": 0.0001,
+      "loss": 0.1398,
+      "step": 4350
+    },
+    {
+      "epoch": 5.898861171366594,
+      "grad_norm": 0.22285659611225128,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 4351
+    },
+    {
+      "epoch": 5.900216919739696,
+      "grad_norm": 0.1792987585067749,
+      "learning_rate": 0.0001,
+      "loss": 0.1327,
+      "step": 4352
+    },
+    {
+      "epoch": 5.901572668112799,
+      "grad_norm": 0.19473497569561005,
+      "learning_rate": 0.0001,
+      "loss": 0.1423,
+      "step": 4353
+    },
+    {
+      "epoch": 5.9029284164859,
+      "grad_norm": 0.215098574757576,
+      "learning_rate": 0.0001,
+      "loss": 0.1261,
+      "step": 4354
+    },
+    {
+      "epoch": 5.9042841648590025,
+      "grad_norm": 0.1684647500514984,
+      "learning_rate": 0.0001,
+      "loss": 0.1274,
+      "step": 4355
+    },
+    {
+      "epoch": 5.905639913232104,
+      "grad_norm": 0.19122451543807983,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 4356
+    },
+    {
+      "epoch": 5.906995661605206,
+      "grad_norm": 0.19239547848701477,
+      "learning_rate": 0.0001,
+      "loss": 0.1415,
+      "step": 4357
+    },
+    {
+      "epoch": 5.908351409978308,
+      "grad_norm": 0.17797715961933136,
+      "learning_rate": 0.0001,
+      "loss": 0.1232,
+      "step": 4358
+    },
+    {
+      "epoch": 5.90970715835141,
+      "grad_norm": 0.2211216241121292,
+      "learning_rate": 0.0001,
+      "loss": 0.1474,
+      "step": 4359
+    },
+    {
+      "epoch": 5.911062906724512,
+      "grad_norm": 0.18178202211856842,
+      "learning_rate": 0.0001,
+      "loss": 0.1211,
+      "step": 4360
+    },
+    {
+      "epoch": 5.912418655097614,
+      "grad_norm": 0.1982683688402176,
+      "learning_rate": 0.0001,
+      "loss": 0.1437,
+      "step": 4361
+    },
+    {
+      "epoch": 5.913774403470716,
+      "grad_norm": 0.20967188477516174,
+      "learning_rate": 0.0001,
+      "loss": 0.1514,
+      "step": 4362
+    },
+    {
+      "epoch": 5.915130151843818,
+      "grad_norm": 0.20189055800437927,
+      "learning_rate": 0.0001,
+      "loss": 0.1533,
+      "step": 4363
+    },
+    {
+      "epoch": 5.9164859002169194,
+      "grad_norm": 0.19186367094516754,
+      "learning_rate": 0.0001,
+      "loss": 0.1441,
+      "step": 4364
+    },
+    {
+      "epoch": 5.917841648590022,
+      "grad_norm": 0.23633533716201782,
+      "learning_rate": 0.0001,
+      "loss": 0.1512,
+      "step": 4365
+    },
+    {
+      "epoch": 5.919197396963123,
+      "grad_norm": 0.19711485505104065,
+      "learning_rate": 0.0001,
+      "loss": 0.1507,
+      "step": 4366
+    },
+    {
+      "epoch": 5.920553145336226,
+      "grad_norm": 0.1767686903476715,
+      "learning_rate": 0.0001,
+      "loss": 0.1374,
+      "step": 4367
+    },
+    {
+      "epoch": 5.921908893709327,
+      "grad_norm": 0.21915297210216522,
+      "learning_rate": 0.0001,
+      "loss": 0.1539,
+      "step": 4368
+    },
+    {
+      "epoch": 5.9232646420824295,
+      "grad_norm": 0.17034420371055603,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 4369
+    },
+    {
+      "epoch": 5.924620390455532,
+      "grad_norm": 0.21115270256996155,
+      "learning_rate": 0.0001,
+      "loss": 0.1538,
+      "step": 4370
+    },
+    {
+      "epoch": 5.925976138828633,
+      "grad_norm": 0.1877342164516449,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 4371
+    },
+    {
+      "epoch": 5.927331887201735,
+      "grad_norm": 0.17378148436546326,
+      "learning_rate": 0.0001,
+      "loss": 0.1269,
+      "step": 4372
+    },
+    {
+      "epoch": 5.928687635574837,
+      "grad_norm": 0.19502784311771393,
+      "learning_rate": 0.0001,
+      "loss": 0.1413,
+      "step": 4373
+    },
+    {
+      "epoch": 5.93004338394794,
+      "grad_norm": 0.18004482984542847,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 4374
+    },
+    {
+      "epoch": 5.931399132321041,
+      "grad_norm": 0.19593027234077454,
+      "learning_rate": 0.0001,
+      "loss": 0.156,
+      "step": 4375
+    },
+    {
+      "epoch": 5.9327548806941435,
+      "grad_norm": 0.19377541542053223,
+      "learning_rate": 0.0001,
+      "loss": 0.1334,
+      "step": 4376
+    },
+    {
+      "epoch": 5.934110629067245,
+      "grad_norm": 0.2138359695672989,
+      "learning_rate": 0.0001,
+      "loss": 0.1387,
+      "step": 4377
+    },
+    {
+      "epoch": 5.935466377440347,
+      "grad_norm": 0.19009967148303986,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 4378
+    },
+    {
+      "epoch": 5.936822125813449,
+      "grad_norm": 0.21776951849460602,
+      "learning_rate": 0.0001,
+      "loss": 0.1531,
+      "step": 4379
+    },
+    {
+      "epoch": 5.938177874186551,
+      "grad_norm": 0.236973375082016,
+      "learning_rate": 0.0001,
+      "loss": 0.1312,
+      "step": 4380
+    },
+    {
+      "epoch": 5.939533622559653,
+      "grad_norm": 0.20591019093990326,
+      "learning_rate": 0.0001,
+      "loss": 0.1456,
+      "step": 4381
+    },
+    {
+      "epoch": 5.940889370932755,
+      "grad_norm": 0.1998893767595291,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 4382
+    },
+    {
+      "epoch": 5.9422451193058565,
+      "grad_norm": 0.21007481217384338,
+      "learning_rate": 0.0001,
+      "loss": 0.1167,
+      "step": 4383
+    },
+    {
+      "epoch": 5.943600867678959,
+      "grad_norm": 0.2162458449602127,
+      "learning_rate": 0.0001,
+      "loss": 0.1417,
+      "step": 4384
+    },
+    {
+      "epoch": 5.94495661605206,
+      "grad_norm": 0.2039053589105606,
+      "learning_rate": 0.0001,
+      "loss": 0.1419,
+      "step": 4385
+    },
+    {
+      "epoch": 5.946312364425163,
+      "grad_norm": 0.2083650380373001,
+      "learning_rate": 0.0001,
+      "loss": 0.1604,
+      "step": 4386
+    },
+    {
+      "epoch": 5.947668112798265,
+      "grad_norm": 0.17899303138256073,
+      "learning_rate": 0.0001,
+      "loss": 0.1323,
+      "step": 4387
+    },
+    {
+      "epoch": 5.949023861171367,
+      "grad_norm": 0.21787215769290924,
+      "learning_rate": 0.0001,
+      "loss": 0.1395,
+      "step": 4388
+    },
+    {
+      "epoch": 5.950379609544468,
+      "grad_norm": 0.18161660432815552,
+      "learning_rate": 0.0001,
+      "loss": 0.143,
+      "step": 4389
+    },
+    {
+      "epoch": 5.9517353579175705,
+      "grad_norm": 0.24860841035842896,
+      "learning_rate": 0.0001,
+      "loss": 0.1425,
+      "step": 4390
+    },
+    {
+      "epoch": 5.953091106290673,
+      "grad_norm": 0.18302762508392334,
+      "learning_rate": 0.0001,
+      "loss": 0.1249,
+      "step": 4391
+    },
+    {
+      "epoch": 5.954446854663774,
+      "grad_norm": 0.19403283298015594,
+      "learning_rate": 0.0001,
+      "loss": 0.1438,
+      "step": 4392
+    },
+    {
+      "epoch": 5.955802603036877,
+      "grad_norm": 0.17736317217350006,
+      "learning_rate": 0.0001,
+      "loss": 0.1534,
+      "step": 4393
+    },
+    {
+      "epoch": 5.957158351409978,
+      "grad_norm": 0.18197286128997803,
+      "learning_rate": 0.0001,
+      "loss": 0.1347,
+      "step": 4394
+    },
+    {
+      "epoch": 5.9585140997830806,
+      "grad_norm": 0.2025502771139145,
+      "learning_rate": 0.0001,
+      "loss": 0.1354,
+      "step": 4395
+    },
+    {
+      "epoch": 5.959869848156182,
+      "grad_norm": 0.21466895937919617,
+      "learning_rate": 0.0001,
+      "loss": 0.1544,
+      "step": 4396
+    },
+    {
+      "epoch": 5.961225596529284,
+      "grad_norm": 0.2115817368030548,
+      "learning_rate": 0.0001,
+      "loss": 0.1496,
+      "step": 4397
+    },
+    {
+      "epoch": 5.962581344902386,
+      "grad_norm": 0.2186838835477829,
+      "learning_rate": 0.0001,
+      "loss": 0.1396,
+      "step": 4398
+    },
+    {
+      "epoch": 5.963937093275488,
+      "grad_norm": 0.2008133977651596,
+      "learning_rate": 0.0001,
+      "loss": 0.1455,
+      "step": 4399
+    },
+    {
+      "epoch": 5.96529284164859,
+      "grad_norm": 0.19514040648937225,
+      "learning_rate": 0.0001,
+      "loss": 0.1311,
+      "step": 4400
+    },
+    {
+      "epoch": 5.966648590021692,
+      "grad_norm": 0.20468895137310028,
+      "learning_rate": 0.0001,
+      "loss": 0.1257,
+      "step": 4401
+    },
+    {
+      "epoch": 5.968004338394794,
+      "grad_norm": 0.20378762483596802,
+      "learning_rate": 0.0001,
+      "loss": 0.1499,
+      "step": 4402
+    },
+    {
+      "epoch": 5.969360086767896,
+      "grad_norm": 0.2656407356262207,
+      "learning_rate": 0.0001,
+      "loss": 0.1362,
+      "step": 4403
+    },
+    {
+      "epoch": 5.9707158351409975,
+      "grad_norm": 0.21926485002040863,
+      "learning_rate": 0.0001,
+      "loss": 0.1381,
+      "step": 4404
+    },
+    {
+      "epoch": 5.9720715835141,
+      "grad_norm": 0.21793991327285767,
+      "learning_rate": 0.0001,
+      "loss": 0.1517,
+      "step": 4405
+    },
+    {
+      "epoch": 5.973427331887201,
+      "grad_norm": 0.21080057322978973,
+      "learning_rate": 0.0001,
+      "loss": 0.1408,
+      "step": 4406
+    },
+    {
+      "epoch": 5.974783080260304,
+      "grad_norm": 0.2037307173013687,
+      "learning_rate": 0.0001,
+      "loss": 0.1451,
+      "step": 4407
+    },
+    {
+      "epoch": 5.976138828633406,
+      "grad_norm": 0.2911374270915985,
+      "learning_rate": 0.0001,
+      "loss": 0.146,
+      "step": 4408
+    },
+    {
+      "epoch": 5.9774945770065075,
+      "grad_norm": 0.20008008182048798,
+      "learning_rate": 0.0001,
+      "loss": 0.1477,
+      "step": 4409
+    },
+    {
+      "epoch": 5.97885032537961,
+      "grad_norm": 0.18661649525165558,
+      "learning_rate": 0.0001,
+      "loss": 0.1352,
+      "step": 4410
+    },
+    {
+      "epoch": 5.980206073752711,
+      "grad_norm": 0.16551697254180908,
+      "learning_rate": 0.0001,
+      "loss": 0.141,
+      "step": 4411
+    },
+    {
+      "epoch": 5.981561822125814,
+      "grad_norm": 0.17837435007095337,
+      "learning_rate": 0.0001,
+      "loss": 0.1306,
+      "step": 4412
+    },
+    {
+      "epoch": 5.982917570498915,
+      "grad_norm": 0.15049603581428528,
+      "learning_rate": 0.0001,
+      "loss": 0.136,
+      "step": 4413
+    },
+    {
+      "epoch": 5.984273318872018,
+      "grad_norm": 0.24113358557224274,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 4414
+    },
+    {
+      "epoch": 5.985629067245119,
+      "grad_norm": 0.19876691699028015,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 4415
+    },
+    {
+      "epoch": 5.9869848156182215,
+      "grad_norm": 0.18655920028686523,
+      "learning_rate": 0.0001,
+      "loss": 0.1336,
+      "step": 4416
+    },
+    {
+      "epoch": 5.988340563991323,
+      "grad_norm": 0.18066854774951935,
+      "learning_rate": 0.0001,
+      "loss": 0.1511,
+      "step": 4417
+    },
+    {
+      "epoch": 5.989696312364425,
+      "grad_norm": 0.17450904846191406,
+      "learning_rate": 0.0001,
+      "loss": 0.1414,
+      "step": 4418
+    },
+    {
+      "epoch": 5.991052060737527,
+      "grad_norm": 0.18048705160617828,
+      "learning_rate": 0.0001,
+      "loss": 0.1386,
+      "step": 4419
+    },
+    {
+      "epoch": 5.992407809110629,
+      "grad_norm": 0.17317599058151245,
+      "learning_rate": 0.0001,
+      "loss": 0.1294,
+      "step": 4420
+    },
+    {
+      "epoch": 5.993763557483731,
+      "grad_norm": 0.18497884273529053,
+      "learning_rate": 0.0001,
+      "loss": 0.1385,
+      "step": 4421
+    },
+    {
+      "epoch": 5.995119305856833,
+      "grad_norm": 0.21262528002262115,
+      "learning_rate": 0.0001,
+      "loss": 0.144,
+      "step": 4422
+    },
+    {
+      "epoch": 5.995119305856833,
+      "step": 4422,
+      "total_flos": 8.901590069421277e+18,
+      "train_loss": 0.17196221762365854,
+      "train_runtime": 80854.8024,
+      "train_samples_per_second": 1.642,
+      "train_steps_per_second": 0.055
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4422,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 6,
+  "save_steps": 500,
+  "total_flos": 8.901590069421277e+18,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}