diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14213 @@
+{
+  "best_metric": 0.15097101032733917,
+  "best_model_checkpoint": "saved_model/lop_sep_2024/checkpoint-13492",
+  "epoch": 2.999888827126181,
+  "eval_steps": 500,
+  "global_step": 20238,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": NaN,
+      "learning_rate": 0.0,
+      "loss": 70.4771,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 18.768062591552734,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 70.9315,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 19.307415008544922,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 70.6744,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 22.3438663482666,
+      "learning_rate": 1.2e-05,
+      "loss": 68.4448,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 27.788856506347656,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 64.7802,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 34.86759948730469,
+      "learning_rate": 2.15e-05,
+      "loss": 56.6582,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 43.42910385131836,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 42.3126,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 14.587575912475586,
+      "learning_rate": 3.15e-05,
+      "loss": 20.2735,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 12.488475799560547,
+      "learning_rate": 3.65e-05,
+      "loss": 6.4981,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 3.4391798973083496,
+      "learning_rate": 4.15e-05,
+      "loss": 0.9414,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.6826833486557007,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.4819,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.1236026287078857,
+      "learning_rate": 5.1500000000000005e-05,
+      "loss": 0.4228,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.420114278793335,
+      "learning_rate": 5.65e-05,
+      "loss": 0.4401,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.8946385383605957,
+      "learning_rate": 6.15e-05,
+      "loss": 0.6266,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 12.989470481872559,
+      "learning_rate": 6.65e-05,
+      "loss": 0.5741,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 2.2886247634887695,
+      "learning_rate": 7.15e-05,
+      "loss": 0.4404,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 3.232952833175659,
+      "learning_rate": 7.65e-05,
+      "loss": 0.4768,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.457054615020752,
+      "learning_rate": 8.15e-05,
+      "loss": 0.4633,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.454648733139038,
+      "learning_rate": 8.65e-05,
+      "loss": 0.4295,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 3.4959018230438232,
+      "learning_rate": 9.15e-05,
+      "loss": 0.4526,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8352223634719849,
+      "learning_rate": 9.65e-05,
+      "loss": 0.3898,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.780531406402588,
+      "learning_rate": 9.999553969669938e-05,
+      "loss": 0.3962,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 2.455747127532959,
+      "learning_rate": 9.998067201903063e-05,
+      "loss": 0.4071,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.8026726245880127,
+      "learning_rate": 9.996580434136188e-05,
+      "loss": 0.383,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.4635368585586548,
+      "learning_rate": 9.995093666369314e-05,
+      "loss": 0.4433,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.1103901863098145,
+      "learning_rate": 9.993606898602439e-05,
+      "loss": 0.3985,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 4.443718910217285,
+      "learning_rate": 9.992120130835564e-05,
+      "loss": 0.3978,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.011767387390137,
+      "learning_rate": 9.990633363068688e-05,
+      "loss": 0.4696,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 5.497318744659424,
+      "learning_rate": 9.989146595301815e-05,
+      "loss": 0.4204,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.5428732633590698,
+      "learning_rate": 9.98765982753494e-05,
+      "loss": 0.4107,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.399179220199585,
+      "learning_rate": 9.986173059768065e-05,
+      "loss": 0.3831,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3291690349578857,
+      "learning_rate": 9.98468629200119e-05,
+      "loss": 0.3904,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5392258167266846,
+      "learning_rate": 9.983199524234314e-05,
+      "loss": 0.3349,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 4.950677871704102,
+      "learning_rate": 9.981712756467441e-05,
+      "loss": 0.3969,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 5.227260112762451,
+      "learning_rate": 9.980225988700565e-05,
+      "loss": 0.4026,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.5111877918243408,
+      "learning_rate": 9.97873922093369e-05,
+      "loss": 0.3014,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 2.647972583770752,
+      "learning_rate": 9.977252453166816e-05,
+      "loss": 0.2976,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 3.595974922180176,
+      "learning_rate": 9.975765685399942e-05,
+      "loss": 0.3304,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.389302968978882,
+      "learning_rate": 9.974278917633067e-05,
+      "loss": 0.2689,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.902669906616211,
+      "learning_rate": 9.972792149866191e-05,
+      "loss": 0.2924,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.135059118270874,
+      "learning_rate": 9.971305382099317e-05,
+      "loss": 0.3061,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 3.143860101699829,
+      "learning_rate": 9.969818614332441e-05,
+      "loss": 0.2687,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 2.1663131713867188,
+      "learning_rate": 9.968331846565568e-05,
+      "loss": 0.2633,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 4.806302547454834,
+      "learning_rate": 9.966845078798692e-05,
+      "loss": 0.2626,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.9977257251739502,
+      "learning_rate": 9.965358311031817e-05,
+      "loss": 0.2876,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0454623699188232,
+      "learning_rate": 9.963871543264943e-05,
+      "loss": 0.2663,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 2.6900148391723633,
+      "learning_rate": 9.962384775498068e-05,
+      "loss": 0.2593,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.413218975067139,
+      "learning_rate": 9.960898007731194e-05,
+      "loss": 0.2523,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 4.643527984619141,
+      "learning_rate": 9.959411239964318e-05,
+      "loss": 0.272,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.9696474075317383,
+      "learning_rate": 9.957924472197443e-05,
+      "loss": 0.2435,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.4355976581573486,
+      "learning_rate": 9.956437704430567e-05,
+      "loss": 0.2267,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6311492919921875,
+      "learning_rate": 9.954950936663694e-05,
+      "loss": 0.2622,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.0157711505889893,
+      "learning_rate": 9.953464168896818e-05,
+      "loss": 0.2461,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.7786158323287964,
+      "learning_rate": 9.951977401129944e-05,
+      "loss": 0.2369,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.6747984886169434,
+      "learning_rate": 9.950490633363069e-05,
+      "loss": 0.2115,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9853987693786621,
+      "learning_rate": 9.949003865596195e-05,
+      "loss": 0.2139,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 2.4679114818573,
+      "learning_rate": 9.94751709782932e-05,
+      "loss": 0.262,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.7220086455345154,
+      "learning_rate": 9.946030330062444e-05,
+      "loss": 0.2262,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.3707178831100464,
+      "learning_rate": 9.94454356229557e-05,
+      "loss": 0.2501,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6298980712890625,
+      "learning_rate": 9.943056794528694e-05,
+      "loss": 0.2237,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 5.072420597076416,
+      "learning_rate": 9.941570026761821e-05,
+      "loss": 0.2358,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9326792359352112,
+      "learning_rate": 9.940083258994945e-05,
+      "loss": 0.2013,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6766129732131958,
+      "learning_rate": 9.93859649122807e-05,
+      "loss": 0.1931,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.741708755493164,
+      "learning_rate": 9.937109723461196e-05,
+      "loss": 0.2234,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 4.426298141479492,
+      "learning_rate": 9.935622955694321e-05,
+      "loss": 0.2425,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8037042021751404,
+      "learning_rate": 9.934136187927447e-05,
+      "loss": 0.2243,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 2.159052610397339,
+      "learning_rate": 9.932649420160571e-05,
+      "loss": 0.2101,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.6584820747375488,
+      "learning_rate": 9.931162652393696e-05,
+      "loss": 0.2234,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 4.1978840827941895,
+      "learning_rate": 9.929675884626822e-05,
+      "loss": 0.2182,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 6.110688209533691,
+      "learning_rate": 9.928189116859947e-05,
+      "loss": 0.2312,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 1.9383026361465454,
+      "learning_rate": 9.926702349093073e-05,
+      "loss": 0.2148,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8565307855606079,
+      "learning_rate": 9.925215581326197e-05,
+      "loss": 0.2262,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8206549286842346,
+      "learning_rate": 9.923728813559322e-05,
+      "loss": 0.2025,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.799793004989624,
+      "learning_rate": 9.922242045792448e-05,
+      "loss": 0.2057,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 2.572327136993408,
+      "learning_rate": 9.920755278025573e-05,
+      "loss": 0.1932,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 3.779280185699463,
+      "learning_rate": 9.919268510258697e-05,
+      "loss": 0.2238,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 4.892170429229736,
+      "learning_rate": 9.917781742491823e-05,
+      "loss": 0.2243,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.9165033102035522,
+      "learning_rate": 9.916294974724948e-05,
+      "loss": 0.2015,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.2273848056793213,
+      "learning_rate": 9.914808206958074e-05,
+      "loss": 0.1935,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9774864315986633,
+      "learning_rate": 9.9133214391912e-05,
+      "loss": 0.1876,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.4223579168319702,
+      "learning_rate": 9.911834671424323e-05,
+      "loss": 0.1963,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9810081124305725,
+      "learning_rate": 9.910347903657449e-05,
+      "loss": 0.1806,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.8502489328384399,
+      "learning_rate": 9.908861135890574e-05,
+      "loss": 0.2476,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.8983548879623413,
+      "learning_rate": 9.9073743681237e-05,
+      "loss": 0.1994,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.5797375440597534,
+      "learning_rate": 9.905887600356824e-05,
+      "loss": 0.1977,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.8001552224159241,
+      "learning_rate": 9.90440083258995e-05,
+      "loss": 0.2037,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.5331403017044067,
+      "learning_rate": 9.902914064823075e-05,
+      "loss": 0.197,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6793760061264038,
+      "learning_rate": 9.9014272970562e-05,
+      "loss": 0.2199,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.6174043416976929,
+      "learning_rate": 9.899940529289326e-05,
+      "loss": 0.2152,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.187853455543518,
+      "learning_rate": 9.89845376152245e-05,
+      "loss": 0.2005,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.541826009750366,
+      "learning_rate": 9.896966993755576e-05,
+      "loss": 0.2007,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.699213981628418,
+      "learning_rate": 9.895480225988701e-05,
+      "loss": 0.2046,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9006096124649048,
+      "learning_rate": 9.893993458221826e-05,
+      "loss": 0.196,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.516926646232605,
+      "learning_rate": 9.89250669045495e-05,
+      "loss": 0.1857,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.140449047088623,
+      "learning_rate": 9.891019922688076e-05,
+      "loss": 0.1988,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 1.6413800716400146,
+      "learning_rate": 9.889533154921203e-05,
+      "loss": 0.1958,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7474754452705383,
+      "learning_rate": 9.888046387154327e-05,
+      "loss": 0.1907,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.7224648594856262,
+      "learning_rate": 9.886559619387453e-05,
+      "loss": 0.1969,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9668368697166443,
+      "learning_rate": 9.885072851620577e-05,
+      "loss": 0.2143,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3923943042755127,
+      "learning_rate": 9.883586083853702e-05,
+      "loss": 0.1999,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.00828218460083,
+      "learning_rate": 9.882099316086828e-05,
+      "loss": 0.1885,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7937036156654358,
+      "learning_rate": 9.880612548319953e-05,
+      "loss": 0.1842,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9639225602149963,
+      "learning_rate": 9.879125780553079e-05,
+      "loss": 0.1884,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 2.1198477745056152,
+      "learning_rate": 9.877639012786203e-05,
+      "loss": 0.1958,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7752804756164551,
+      "learning_rate": 9.87615224501933e-05,
+      "loss": 0.1889,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.39463403820991516,
+      "learning_rate": 9.874665477252454e-05,
+      "loss": 0.1738,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.6980741024017334,
+      "learning_rate": 9.873178709485579e-05,
+      "loss": 0.1875,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7736172080039978,
+      "learning_rate": 9.871691941718703e-05,
+      "loss": 0.1893,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.968677282333374,
+      "learning_rate": 9.870205173951829e-05,
+      "loss": 0.1737,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1933799982070923,
+      "learning_rate": 9.868718406184954e-05,
+      "loss": 0.1706,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 2.9177980422973633,
+      "learning_rate": 9.86723163841808e-05,
+      "loss": 0.1942,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.6118002533912659,
+      "learning_rate": 9.865744870651205e-05,
+      "loss": 0.172,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.0720818042755127,
+      "learning_rate": 9.864258102884329e-05,
+      "loss": 0.1899,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.246082067489624,
+      "learning_rate": 9.862771335117456e-05,
+      "loss": 0.1944,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5863025188446045,
+      "learning_rate": 9.86128456735058e-05,
+      "loss": 0.1836,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.1113203763961792,
+      "learning_rate": 9.859797799583706e-05,
+      "loss": 0.1969,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.720690369606018,
+      "learning_rate": 9.85831103181683e-05,
+      "loss": 0.2078,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.5786492824554443,
+      "learning_rate": 9.856824264049957e-05,
+      "loss": 0.1941,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 2.1644389629364014,
+      "learning_rate": 9.855337496283081e-05,
+      "loss": 0.1853,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.7992768287658691,
+      "learning_rate": 9.853850728516206e-05,
+      "loss": 0.1937,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3800251483917236,
+      "learning_rate": 9.852363960749332e-05,
+      "loss": 0.1991,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 3.22444224357605,
+      "learning_rate": 9.850877192982456e-05,
+      "loss": 0.1876,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.721161961555481,
+      "learning_rate": 9.849390425215583e-05,
+      "loss": 0.1776,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 2.3023221492767334,
+      "learning_rate": 9.847903657448707e-05,
+      "loss": 0.1912,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.44242483377456665,
+      "learning_rate": 9.846416889681832e-05,
+      "loss": 0.1716,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.6946061849594116,
+      "learning_rate": 9.844930121914956e-05,
+      "loss": 0.1822,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 2.3526597023010254,
+      "learning_rate": 9.843443354148083e-05,
+      "loss": 0.1931,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.595988392829895,
+      "learning_rate": 9.841956586381207e-05,
+      "loss": 0.1764,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5539588928222656,
+      "learning_rate": 9.840469818614333e-05,
+      "loss": 0.181,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.45790284872055054,
+      "learning_rate": 9.838983050847458e-05,
+      "loss": 0.1831,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.5096701383590698,
+      "learning_rate": 9.837496283080582e-05,
+      "loss": 0.1648,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.4358065128326416,
+      "learning_rate": 9.836009515313709e-05,
+      "loss": 0.1892,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.45344677567481995,
+      "learning_rate": 9.834522747546833e-05,
+      "loss": 0.1882,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.9001879692077637,
+      "learning_rate": 9.833035979779959e-05,
+      "loss": 0.182,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 1.2151376008987427,
+      "learning_rate": 9.831549212013084e-05,
+      "loss": 0.2068,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.489504814147949,
+      "learning_rate": 9.83006244424621e-05,
+      "loss": 0.1874,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 2.392239809036255,
+      "learning_rate": 9.828575676479335e-05,
+      "loss": 0.2012,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8293458819389343,
+      "learning_rate": 9.82708890871246e-05,
+      "loss": 0.1756,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.4222114086151123,
+      "learning_rate": 9.825602140945585e-05,
+      "loss": 0.1781,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.6707155704498291,
+      "learning_rate": 9.824115373178709e-05,
+      "loss": 0.1687,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.2496354579925537,
+      "learning_rate": 9.822628605411836e-05,
+      "loss": 0.1698,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7204919457435608,
+      "learning_rate": 9.82114183764496e-05,
+      "loss": 0.1756,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.3436416387557983,
+      "learning_rate": 9.819655069878085e-05,
+      "loss": 0.1813,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.1300007104873657,
+      "learning_rate": 9.818168302111211e-05,
+      "loss": 0.1865,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.4229620695114136,
+      "learning_rate": 9.816681534344336e-05,
+      "loss": 0.1765,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 1.7664425373077393,
+      "learning_rate": 9.815194766577462e-05,
+      "loss": 0.19,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.6647424697875977,
+      "learning_rate": 9.813707998810586e-05,
+      "loss": 0.1759,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.3758875131607056,
+      "learning_rate": 9.812221231043711e-05,
+      "loss": 0.1771,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.0963157415390015,
+      "learning_rate": 9.810734463276836e-05,
+      "loss": 0.181,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.6835664510726929,
+      "learning_rate": 9.809247695509962e-05,
+      "loss": 0.1668,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 2.1593921184539795,
+      "learning_rate": 9.807760927743087e-05,
+      "loss": 0.1816,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.253033995628357,
+      "learning_rate": 9.806274159976212e-05,
+      "loss": 0.1721,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 2.998976945877075,
+      "learning_rate": 9.804787392209337e-05,
+      "loss": 0.1867,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.396804928779602,
+      "learning_rate": 9.803300624442463e-05,
+      "loss": 0.1721,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.6815312504768372,
+      "learning_rate": 9.801813856675588e-05,
+      "loss": 0.1826,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8488101959228516,
+      "learning_rate": 9.800327088908713e-05,
+      "loss": 0.1872,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0025837421417236,
+      "learning_rate": 9.798840321141838e-05,
+      "loss": 0.1683,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0327823162078857,
+      "learning_rate": 9.797353553374962e-05,
+      "loss": 0.1722,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.4573004245758057,
+      "learning_rate": 9.795866785608089e-05,
+      "loss": 0.1717,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.6835706233978271,
+      "learning_rate": 9.794380017841213e-05,
+      "loss": 0.1797,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.981534481048584,
+      "learning_rate": 9.792893250074339e-05,
+      "loss": 0.1733,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.49460187554359436,
+      "learning_rate": 9.791406482307464e-05,
+      "loss": 0.175,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.1679047346115112,
+      "learning_rate": 9.78991971454059e-05,
+      "loss": 0.1754,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.691370964050293,
+      "learning_rate": 9.788432946773715e-05,
+      "loss": 0.167,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.409994900226593,
+      "learning_rate": 9.786946179006839e-05,
+      "loss": 0.1688,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7671188116073608,
+      "learning_rate": 9.785459411239965e-05,
+      "loss": 0.1626,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.6950274109840393,
+      "learning_rate": 9.78397264347309e-05,
+      "loss": 0.1717,
+      "step": 1660
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.0336863994598389,
+      "learning_rate": 9.782485875706216e-05,
+      "loss": 0.1906,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 3.111928939819336,
+      "learning_rate": 9.780999107939341e-05,
+      "loss": 0.2204,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.817621648311615,
+      "learning_rate": 9.779512340172465e-05,
+      "loss": 0.192,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2715766429901123,
+      "learning_rate": 9.77802557240559e-05,
+      "loss": 0.1728,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.4694221019744873,
+      "learning_rate": 9.776538804638716e-05,
+      "loss": 0.1751,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 2.2308688163757324,
+      "learning_rate": 9.775052036871842e-05,
+      "loss": 0.1749,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.3650729656219482,
+      "learning_rate": 9.773565269104966e-05,
+      "loss": 0.1786,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.3981892466545105,
+      "learning_rate": 9.772078501338091e-05,
+      "loss": 0.1724,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.6411010026931763,
+      "learning_rate": 9.770591733571217e-05,
+      "loss": 0.1774,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.6380265951156616,
+      "learning_rate": 9.769104965804342e-05,
+      "loss": 0.1793,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.4115835130214691,
+      "learning_rate": 9.767618198037468e-05,
+      "loss": 0.1636,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.61713707447052,
+      "learning_rate": 9.766131430270592e-05,
+      "loss": 0.1769,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0060006380081177,
+      "learning_rate": 9.764644662503717e-05,
+      "loss": 0.1691,
+      "step": 1790
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.9878020286560059,
+      "learning_rate": 9.763157894736843e-05,
+      "loss": 0.1749,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 2.226569414138794,
+      "learning_rate": 9.761671126969968e-05,
+      "loss": 0.1671,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7667270302772522,
+      "learning_rate": 9.760184359203092e-05,
+      "loss": 0.1822,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.7759143114089966,
+      "learning_rate": 9.758697591436218e-05,
+      "loss": 0.175,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.309430718421936,
+      "learning_rate": 9.757210823669343e-05,
+      "loss": 0.1683,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.3665989339351654,
+      "learning_rate": 9.755724055902469e-05,
+      "loss": 0.1748,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5607463121414185,
+      "learning_rate": 9.754237288135594e-05,
+      "loss": 0.1646,
+      "step": 1860
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 2.0324952602386475,
+      "learning_rate": 9.752750520368718e-05,
+      "loss": 0.1818,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.5004349946975708,
+      "learning_rate": 9.751263752601844e-05,
+      "loss": 0.1795,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.5794762372970581,
+      "learning_rate": 9.749776984834969e-05,
+      "loss": 0.1877,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.6873047947883606,
+      "learning_rate": 9.748290217068095e-05,
+      "loss": 0.1872,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.3563670814037323,
+      "learning_rate": 9.746803449301219e-05,
+      "loss": 0.173,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.47983822226524353,
+      "learning_rate": 9.745316681534344e-05,
+      "loss": 0.1654,
+      "step": 1920
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.31588858366012573,
+      "learning_rate": 9.74382991376747e-05,
+      "loss": 0.1588,
+      "step": 1930
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.5894639492034912,
+      "learning_rate": 9.742343146000595e-05,
+      "loss": 0.1778,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5094777941703796,
+      "learning_rate": 9.740856378233721e-05,
+      "loss": 0.1869,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4982427656650543,
+      "learning_rate": 9.739369610466845e-05,
+      "loss": 0.182,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.5018313527107239,
+      "learning_rate": 9.73788284269997e-05,
+      "loss": 0.1621,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.4199548363685608,
+      "learning_rate": 9.736396074933096e-05,
+      "loss": 0.1654,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 1.6279269456863403,
+      "learning_rate": 9.734909307166221e-05,
+      "loss": 0.1802,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.3614368438720703,
+      "learning_rate": 9.733422539399347e-05,
+      "loss": 0.1636,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6040695905685425,
+      "learning_rate": 9.731935771632471e-05,
+      "loss": 0.1694,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.44964730739593506,
+      "learning_rate": 9.730449003865598e-05,
+      "loss": 0.164,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.9905171394348145,
+      "learning_rate": 9.728962236098722e-05,
+      "loss": 0.1692,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.676896572113037,
+      "learning_rate": 9.727475468331847e-05,
+      "loss": 0.1734,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.833280086517334,
+      "learning_rate": 9.725988700564971e-05,
+      "loss": 0.1842,
+      "step": 2050
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.3920565843582153,
+      "learning_rate": 9.724501932798098e-05,
+      "loss": 0.1834,
+      "step": 2060
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7751169800758362,
+      "learning_rate": 9.723015165031222e-05,
+      "loss": 0.1755,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.8627113103866577,
+      "learning_rate": 9.721528397264348e-05,
+      "loss": 0.1695,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.434849351644516,
+      "learning_rate": 9.720041629497473e-05,
+      "loss": 0.1675,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.202000617980957,
+      "learning_rate": 9.718554861730597e-05,
+      "loss": 0.1749,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7374542951583862,
+      "learning_rate": 9.717068093963724e-05,
+      "loss": 0.169,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.6856274604797363,
+      "learning_rate": 9.715581326196848e-05,
+      "loss": 0.1724,
+      "step": 2120
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8698721528053284,
+      "learning_rate": 9.714094558429974e-05,
+      "loss": 0.1705,
+      "step": 2130
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.2129219770431519,
+      "learning_rate": 9.712607790663098e-05,
+      "loss": 0.1704,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.3243730962276459,
+      "learning_rate": 9.711121022896225e-05,
+      "loss": 0.1753,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1494802236557007,
+      "learning_rate": 9.709634255129349e-05,
+      "loss": 0.1702,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.7077322602272034,
+      "learning_rate": 9.708147487362474e-05,
+      "loss": 0.176,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 3.5321290493011475,
+      "learning_rate": 9.7066607195956e-05,
+      "loss": 0.1799,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.40482622385025024,
+      "learning_rate": 9.705173951828724e-05,
+      "loss": 0.1817,
+      "step": 2190
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.4115407466888428,
+      "learning_rate": 9.703687184061851e-05,
+      "loss": 0.1863,
+      "step": 2200
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.6012375354766846,
+      "learning_rate": 9.702200416294975e-05,
+      "loss": 0.1697,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 3.5311667919158936,
+      "learning_rate": 9.7007136485281e-05,
+      "loss": 0.1756,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.0460686683654785,
+      "learning_rate": 9.699226880761225e-05,
+      "loss": 0.1733,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 2.7502028942108154,
+      "learning_rate": 9.697740112994351e-05,
+      "loss": 0.177,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.4361170530319214,
+      "learning_rate": 9.696253345227476e-05,
+      "loss": 0.1798,
+      "step": 2250
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 2.131711006164551,
+      "learning_rate": 9.694766577460601e-05,
+      "loss": 0.1678,
+      "step": 2260
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.6454190611839294,
+      "learning_rate": 9.693279809693727e-05,
+      "loss": 0.1575,
+      "step": 2270
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5098270773887634,
+      "learning_rate": 9.69179304192685e-05,
+      "loss": 0.1788,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5451599359512329,
+      "learning_rate": 9.690306274159977e-05,
+      "loss": 0.1641,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.5445219874382019,
+      "learning_rate": 9.688819506393102e-05,
+      "loss": 0.1667,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.9343221187591553,
+      "learning_rate": 9.687332738626227e-05,
+      "loss": 0.1676,
+      "step": 2310
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.7586581707000732,
+      "learning_rate": 9.685845970859351e-05,
+      "loss": 0.1726,
+      "step": 2320
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.4477677643299103,
+      "learning_rate": 9.684359203092478e-05,
+      "loss": 0.1703,
+      "step": 2330
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.32315364480018616,
+      "learning_rate": 9.682872435325604e-05,
+      "loss": 0.1541,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.7400575876235962,
+      "learning_rate": 9.681385667558728e-05,
+      "loss": 0.1718,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.6638018488883972,
+      "learning_rate": 9.679898899791853e-05,
+      "loss": 0.1673,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.43184348940849304,
+      "learning_rate": 9.678412132024977e-05,
+      "loss": 0.1853,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.706583023071289,
+      "learning_rate": 9.676925364258104e-05,
+      "loss": 0.169,
+      "step": 2380
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 2.9006285667419434,
+      "learning_rate": 9.675438596491228e-05,
+      "loss": 0.1705,
+      "step": 2390
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.9855265617370605,
+      "learning_rate": 9.673951828724354e-05,
+      "loss": 0.1694,
+      "step": 2400
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0507304668426514,
+      "learning_rate": 9.672465060957479e-05,
+      "loss": 0.1595,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.4149112701416016,
+      "learning_rate": 9.670978293190605e-05,
+      "loss": 0.1763,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.5968409180641174,
+      "learning_rate": 9.66949152542373e-05,
+      "loss": 0.1552,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.7032540440559387,
+      "learning_rate": 9.668004757656854e-05,
+      "loss": 0.159,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.025107979774475,
+      "learning_rate": 9.66651798988998e-05,
+      "loss": 0.1634,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.37000682950019836,
+      "learning_rate": 9.665031222123104e-05,
+      "loss": 0.1653,
+      "step": 2460
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8552184700965881,
+      "learning_rate": 9.66354445435623e-05,
+      "loss": 0.1566,
+      "step": 2470
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3340959846973419,
+      "learning_rate": 9.662057686589355e-05,
+      "loss": 0.1654,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7219660878181458,
+      "learning_rate": 9.66057091882248e-05,
+      "loss": 0.1668,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.3376657962799072,
+      "learning_rate": 9.659084151055606e-05,
+      "loss": 0.1631,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.40849828720092773,
+      "learning_rate": 9.657597383288731e-05,
+      "loss": 0.1714,
+      "step": 2510
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 4.0708909034729,
+      "learning_rate": 9.656110615521857e-05,
+      "loss": 0.1748,
+      "step": 2520
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8569905757904053,
+      "learning_rate": 9.654623847754981e-05,
+      "loss": 0.1759,
+      "step": 2530
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.6083622574806213,
+      "learning_rate": 9.653137079988106e-05,
+      "loss": 0.1771,
+      "step": 2540
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.9755212068557739,
+      "learning_rate": 9.651650312221232e-05,
+      "loss": 0.1653,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.2756294906139374,
+      "learning_rate": 9.650163544454357e-05,
+      "loss": 0.161,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.36292827129364014,
+      "learning_rate": 9.648676776687481e-05,
+      "loss": 0.1627,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.31777092814445496,
+      "learning_rate": 9.647190008920607e-05,
+      "loss": 0.1635,
+      "step": 2580
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.4227207601070404,
+      "learning_rate": 9.645703241153732e-05,
+      "loss": 0.1693,
+      "step": 2590
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.3526834547519684,
+      "learning_rate": 9.644216473386858e-05,
+      "loss": 0.1567,
+      "step": 2600
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.9331281185150146,
+      "learning_rate": 9.642729705619983e-05,
+      "loss": 0.1691,
+      "step": 2610
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.4683771133422852,
+      "learning_rate": 9.641242937853107e-05,
+      "loss": 0.1773,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 2.2021124362945557,
+      "learning_rate": 9.639756170086233e-05,
+      "loss": 0.1749,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.63923180103302,
+      "learning_rate": 9.638269402319358e-05,
+      "loss": 0.1755,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.34077224135398865,
+      "learning_rate": 9.636782634552484e-05,
+      "loss": 0.1595,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8533111810684204,
+      "learning_rate": 9.635295866785608e-05,
+      "loss": 0.1765,
+      "step": 2660
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 2.3589627742767334,
+      "learning_rate": 9.633809099018733e-05,
+      "loss": 0.1564,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4352911412715912,
+      "learning_rate": 9.632322331251859e-05,
+      "loss": 0.1645,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.36782145500183105,
+      "learning_rate": 9.630835563484984e-05,
+      "loss": 0.1651,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.4545479714870453,
+      "learning_rate": 9.62934879571811e-05,
+      "loss": 0.168,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.7421327829360962,
+      "learning_rate": 9.627862027951234e-05,
+      "loss": 0.1643,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7703805565834045,
+      "learning_rate": 9.62637526018436e-05,
+      "loss": 0.1613,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.7927514314651489,
+      "learning_rate": 9.624888492417485e-05,
+      "loss": 0.1542,
+      "step": 2730
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9497324824333191,
+      "learning_rate": 9.62340172465061e-05,
+      "loss": 0.169,
+      "step": 2740
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.6463251113891602,
+      "learning_rate": 9.621914956883736e-05,
+      "loss": 0.1545,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.5164654850959778,
+      "learning_rate": 9.62042818911686e-05,
+      "loss": 0.1746,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.879035472869873,
+      "learning_rate": 9.618941421349985e-05,
+      "loss": 0.1617,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.6450806856155396,
+      "learning_rate": 9.617454653583111e-05,
+      "loss": 0.1694,
+      "step": 2780
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.4430038630962372,
+      "learning_rate": 9.615967885816236e-05,
+      "loss": 0.1618,
+      "step": 2790
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.111541509628296,
+      "learning_rate": 9.61448111804936e-05,
+      "loss": 0.1537,
+      "step": 2800
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.177138090133667,
+      "learning_rate": 9.612994350282486e-05,
+      "loss": 0.171,
+      "step": 2810
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 2.1342763900756836,
+      "learning_rate": 9.611507582515611e-05,
+      "loss": 0.1701,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7735350131988525,
+      "learning_rate": 9.610020814748737e-05,
+      "loss": 0.1588,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.3419579267501831,
+      "learning_rate": 9.608534046981862e-05,
+      "loss": 0.1573,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.48323726654052734,
+      "learning_rate": 9.607047279214987e-05,
+      "loss": 0.1593,
+      "step": 2850
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.437145471572876,
+      "learning_rate": 9.605560511448112e-05,
+      "loss": 0.1691,
+      "step": 2860
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7413353323936462,
+      "learning_rate": 9.604073743681237e-05,
+      "loss": 0.1665,
+      "step": 2870
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7211183905601501,
+      "learning_rate": 9.602586975914363e-05,
+      "loss": 0.155,
+      "step": 2880
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7412411570549011,
+      "learning_rate": 9.601100208147487e-05,
+      "loss": 0.1627,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9689399003982544,
+      "learning_rate": 9.599613440380613e-05,
+      "loss": 0.1734,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9807812571525574,
+      "learning_rate": 9.598126672613738e-05,
+      "loss": 0.1668,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 2.6782937049865723,
+      "learning_rate": 9.596639904846864e-05,
+      "loss": 0.1802,
+      "step": 2920
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.54747074842453,
+      "learning_rate": 9.595153137079989e-05,
+      "loss": 0.1727,
+      "step": 2930
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.2045674324035645,
+      "learning_rate": 9.593666369313113e-05,
+      "loss": 0.1677,
+      "step": 2940
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.2668368816375732,
+      "learning_rate": 9.592179601546239e-05,
+      "loss": 0.1652,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7421100735664368,
+      "learning_rate": 9.590692833779364e-05,
+      "loss": 0.1552,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7410224676132202,
+      "learning_rate": 9.58920606601249e-05,
+      "loss": 0.1569,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.5380434989929199,
+      "learning_rate": 9.587719298245614e-05,
+      "loss": 0.1676,
+      "step": 2980
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.3664904832839966,
+      "learning_rate": 9.586232530478739e-05,
+      "loss": 0.1629,
+      "step": 2990
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.5946515798568726,
+      "learning_rate": 9.584745762711866e-05,
+      "loss": 0.1779,
+      "step": 3000
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.7002317905426025,
+      "learning_rate": 9.58325899494499e-05,
+      "loss": 0.1769,
+      "step": 3010
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.5792291164398193,
+      "learning_rate": 9.581772227178116e-05,
+      "loss": 0.173,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 2.637342691421509,
+      "learning_rate": 9.58028545941124e-05,
+      "loss": 0.184,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.315929651260376,
+      "learning_rate": 9.578798691644367e-05,
+      "loss": 0.1626,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.4628597497940063,
+      "learning_rate": 9.57731192387749e-05,
+      "loss": 0.1585,
+      "step": 3050
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.037359356880188,
+      "learning_rate": 9.575825156110616e-05,
+      "loss": 0.1659,
+      "step": 3060
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7535769939422607,
+      "learning_rate": 9.574338388343742e-05,
+      "loss": 0.1552,
+      "step": 3070
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.6997020244598389,
+      "learning_rate": 9.572851620576866e-05,
+      "loss": 0.161,
+      "step": 3080
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.6227778196334839,
+      "learning_rate": 9.571364852809993e-05,
+      "loss": 0.1664,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9602690935134888,
+      "learning_rate": 9.569878085043117e-05,
+      "loss": 0.1588,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9173592925071716,
+      "learning_rate": 9.568391317276242e-05,
+      "loss": 0.1628,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.1894303560256958,
+      "learning_rate": 9.566904549509366e-05,
+      "loss": 0.1629,
+      "step": 3120
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.7936124205589294,
+      "learning_rate": 9.565417781742493e-05,
+      "loss": 0.169,
+      "step": 3130
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7208412289619446,
+      "learning_rate": 9.563931013975617e-05,
+      "loss": 0.1766,
+      "step": 3140
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2436951398849487,
+      "learning_rate": 9.562444246208743e-05,
+      "loss": 0.1621,
+      "step": 3150
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0052576065063477,
+      "learning_rate": 9.560957478441868e-05,
+      "loss": 0.1607,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 2.342569589614868,
+      "learning_rate": 9.559470710674992e-05,
+      "loss": 0.1655,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5531949400901794,
+      "learning_rate": 9.557983942908119e-05,
+      "loss": 0.1684,
+      "step": 3180
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.5959020256996155,
+      "learning_rate": 9.556497175141243e-05,
+      "loss": 0.1592,
+      "step": 3190
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.7381748557090759,
+      "learning_rate": 9.555010407374369e-05,
+      "loss": 0.1689,
+      "step": 3200
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.3665748834609985,
+      "learning_rate": 9.553523639607493e-05,
+      "loss": 0.157,
+      "step": 3210
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.744511365890503,
+      "learning_rate": 9.55203687184062e-05,
+      "loss": 0.1715,
+      "step": 3220
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.2675743103027344,
+      "learning_rate": 9.550550104073744e-05,
+      "loss": 0.1627,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.7240175008773804,
+      "learning_rate": 9.549063336306869e-05,
+      "loss": 0.1652,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.61540687084198,
+      "learning_rate": 9.547576568539995e-05,
+      "loss": 0.182,
+      "step": 3250
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.7293829917907715,
+      "learning_rate": 9.546089800773119e-05,
+      "loss": 0.173,
+      "step": 3260
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.0945603847503662,
+      "learning_rate": 9.544603033006246e-05,
+      "loss": 0.1598,
+      "step": 3270
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.4362248182296753,
+      "learning_rate": 9.54311626523937e-05,
+      "loss": 0.1634,
+      "step": 3280
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.6548261642456055,
+      "learning_rate": 9.541629497472495e-05,
+      "loss": 0.1594,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3333025872707367,
+      "learning_rate": 9.54014272970562e-05,
+      "loss": 0.1636,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.0312073230743408,
+      "learning_rate": 9.538655961938746e-05,
+      "loss": 0.1623,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 1.9303184747695923,
+      "learning_rate": 9.53716919417187e-05,
+      "loss": 0.1667,
+      "step": 3320
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.3857884109020233,
+      "learning_rate": 9.535682426404996e-05,
+      "loss": 0.1604,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8410763144493103,
+      "learning_rate": 9.534195658638121e-05,
+      "loss": 0.1536,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9080554842948914,
+      "learning_rate": 9.532708890871245e-05,
+      "loss": 0.1625,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0813413858413696,
+      "learning_rate": 9.531222123104372e-05,
+      "loss": 0.1587,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.5331724286079407,
+      "learning_rate": 9.529735355337496e-05,
+      "loss": 0.1606,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 2.220775842666626,
+      "learning_rate": 9.528248587570622e-05,
+      "loss": 0.1642,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.435058355331421,
+      "learning_rate": 9.526761819803747e-05,
+      "loss": 0.1704,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.1016638278961182,
+      "learning_rate": 9.525275052036873e-05,
+      "loss": 0.1602,
+      "step": 3400
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0263643264770508,
+      "learning_rate": 9.523788284269998e-05,
+      "loss": 0.1734,
+      "step": 3410
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.1863274574279785,
+      "learning_rate": 9.522301516503122e-05,
+      "loss": 0.1804,
+      "step": 3420
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5040327310562134,
+      "learning_rate": 9.520814748736248e-05,
+      "loss": 0.1688,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4307524561882019,
+      "learning_rate": 9.519327980969372e-05,
+      "loss": 0.162,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.9712070226669312,
+      "learning_rate": 9.517841213202499e-05,
+      "loss": 0.171,
+      "step": 3450
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.4332945942878723,
+      "learning_rate": 9.516354445435623e-05,
+      "loss": 0.1603,
+      "step": 3460
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.5506505966186523,
+      "learning_rate": 9.514867677668748e-05,
+      "loss": 0.156,
+      "step": 3470
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.394182562828064,
+      "learning_rate": 9.513380909901874e-05,
+      "loss": 0.1645,
+      "step": 3480
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.7404809594154358,
+      "learning_rate": 9.511894142135e-05,
+      "loss": 0.1652,
+      "step": 3490
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 2.453972339630127,
+      "learning_rate": 9.510407374368125e-05,
+      "loss": 0.17,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.8291536569595337,
+      "learning_rate": 9.508920606601249e-05,
+      "loss": 0.1711,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.329593539237976,
+      "learning_rate": 9.507433838834374e-05,
+      "loss": 0.1689,
+      "step": 3520
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.166062593460083,
+      "learning_rate": 9.5059470710675e-05,
+      "loss": 0.1697,
+      "step": 3530
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.4099891185760498,
+      "learning_rate": 9.504460303300625e-05,
+      "loss": 0.1536,
+      "step": 3540
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6420068740844727,
+      "learning_rate": 9.50297353553375e-05,
+      "loss": 0.1588,
+      "step": 3550
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.9775470495224,
+      "learning_rate": 9.501486767766875e-05,
+      "loss": 0.1709,
+      "step": 3560
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.8352495431900024,
+      "learning_rate": 9.5e-05,
+      "loss": 0.1669,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.752253532409668,
+      "learning_rate": 9.498513232233126e-05,
+      "loss": 0.1556,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 1.41259765625,
+      "learning_rate": 9.497026464466251e-05,
+      "loss": 0.1606,
+      "step": 3590
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.6287984251976013,
+      "learning_rate": 9.495539696699376e-05,
+      "loss": 0.1553,
+      "step": 3600
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.47083550691604614,
+      "learning_rate": 9.494052928932501e-05,
+      "loss": 0.1577,
+      "step": 3610
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.32023587822914124,
+      "learning_rate": 9.492566161165627e-05,
+      "loss": 0.158,
+      "step": 3620
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 2.142852544784546,
+      "learning_rate": 9.491079393398752e-05,
+      "loss": 0.167,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.24139684438705444,
+      "learning_rate": 9.489592625631876e-05,
+      "loss": 0.1673,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.6434214115142822,
+      "learning_rate": 9.488105857865002e-05,
+      "loss": 0.1657,
+      "step": 3650
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.7096123695373535,
+      "learning_rate": 9.486619090098127e-05,
+      "loss": 0.1685,
+      "step": 3660
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.4420715868473053,
+      "learning_rate": 9.485132322331253e-05,
+      "loss": 0.1735,
+      "step": 3670
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.7800606489181519,
+      "learning_rate": 9.483645554564378e-05,
+      "loss": 0.1563,
+      "step": 3680
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.5434880256652832,
+      "learning_rate": 9.482158786797502e-05,
+      "loss": 0.1669,
+      "step": 3690
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.879857063293457,
+      "learning_rate": 9.480672019030628e-05,
+      "loss": 0.1715,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1706616878509521,
+      "learning_rate": 9.479185251263753e-05,
+      "loss": 0.168,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.0965744256973267,
+      "learning_rate": 9.477698483496879e-05,
+      "loss": 0.1684,
+      "step": 3720
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.4567830562591553,
+      "learning_rate": 9.476211715730004e-05,
+      "loss": 0.1566,
+      "step": 3730
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.6394124031066895,
+      "learning_rate": 9.474724947963128e-05,
+      "loss": 0.1629,
+      "step": 3740
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6630517244338989,
+      "learning_rate": 9.473238180196254e-05,
+      "loss": 0.1519,
+      "step": 3750
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.7115375995635986,
+      "learning_rate": 9.471751412429379e-05,
+      "loss": 0.1602,
+      "step": 3760
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.2985321581363678,
+      "learning_rate": 9.470264644662505e-05,
+      "loss": 0.154,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.459088921546936,
+      "learning_rate": 9.468777876895629e-05,
+      "loss": 0.1651,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6432485580444336,
+      "learning_rate": 9.467291109128754e-05,
+      "loss": 0.1618,
+      "step": 3790
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.6554367542266846,
+      "learning_rate": 9.46580434136188e-05,
+      "loss": 0.1554,
+      "step": 3800
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.821692705154419,
+      "learning_rate": 9.464317573595005e-05,
+      "loss": 0.1665,
+      "step": 3810
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.4133671522140503,
+      "learning_rate": 9.46283080582813e-05,
+      "loss": 0.1574,
+      "step": 3820
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.609708547592163,
+      "learning_rate": 9.461344038061255e-05,
+      "loss": 0.1658,
+      "step": 3830
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.29960155487060547,
+      "learning_rate": 9.45985727029438e-05,
+      "loss": 0.1638,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.37204432487487793,
+      "learning_rate": 9.458370502527506e-05,
+      "loss": 0.1502,
+      "step": 3850
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.4778790473937988,
+      "learning_rate": 9.456883734760631e-05,
+      "loss": 0.1585,
+      "step": 3860
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.30051159858703613,
+      "learning_rate": 9.455396966993755e-05,
+      "loss": 0.1595,
+      "step": 3870
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.3415975570678711,
+      "learning_rate": 9.453910199226881e-05,
+      "loss": 0.1556,
+      "step": 3880
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 2.3571910858154297,
+      "learning_rate": 9.452423431460006e-05,
+      "loss": 0.1657,
+      "step": 3890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 1.502201795578003,
+      "learning_rate": 9.450936663693132e-05,
+      "loss": 0.1525,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.4980350434780121,
+      "learning_rate": 9.449449895926257e-05,
+      "loss": 0.1575,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7586767077445984,
+      "learning_rate": 9.447963128159381e-05,
+      "loss": 0.1685,
+      "step": 3920
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.33721524477005005,
+      "learning_rate": 9.446476360392507e-05,
+      "loss": 0.1616,
+      "step": 3930
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.36988070607185364,
+      "learning_rate": 9.444989592625632e-05,
+      "loss": 0.1645,
+      "step": 3940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6627219915390015,
+      "learning_rate": 9.443502824858758e-05,
+      "loss": 0.1507,
+      "step": 3950
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4597422778606415,
+      "learning_rate": 9.442016057091882e-05,
+      "loss": 0.1492,
+      "step": 3960
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.6863605380058289,
+      "learning_rate": 9.440529289325007e-05,
+      "loss": 0.1649,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.924536943435669,
+      "learning_rate": 9.439042521558133e-05,
+      "loss": 0.1605,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.3096588850021362,
+      "learning_rate": 9.437555753791258e-05,
+      "loss": 0.1642,
+      "step": 3990
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.33845776319503784,
+      "learning_rate": 9.436068986024384e-05,
+      "loss": 0.1631,
+      "step": 4000
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.4413401782512665,
+      "learning_rate": 9.434582218257508e-05,
+      "loss": 0.1574,
+      "step": 4010
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 3.155444622039795,
+      "learning_rate": 9.433095450490635e-05,
+      "loss": 0.1685,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.26439428329467773,
+      "learning_rate": 9.431608682723759e-05,
+      "loss": 0.1585,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.3483470380306244,
+      "learning_rate": 9.430121914956884e-05,
+      "loss": 0.1636,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4697681665420532,
+      "learning_rate": 9.42863514719001e-05,
+      "loss": 0.1528,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.4681277871131897,
+      "learning_rate": 9.427148379423134e-05,
+      "loss": 0.1492,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.5307562351226807,
+      "learning_rate": 9.425661611656261e-05,
+      "loss": 0.1599,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.5165395736694336,
+      "learning_rate": 9.424174843889385e-05,
+      "loss": 0.1605,
+      "step": 4080
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.7521703243255615,
+      "learning_rate": 9.42268807612251e-05,
+      "loss": 0.1564,
+      "step": 4090
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.0230251550674438,
+      "learning_rate": 9.421201308355635e-05,
+      "loss": 0.1552,
+      "step": 4100
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3919179439544678,
+      "learning_rate": 9.419714540588761e-05,
+      "loss": 0.1598,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1220026016235352,
+      "learning_rate": 9.418227772821885e-05,
+      "loss": 0.1674,
+      "step": 4120
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.90169358253479,
+      "learning_rate": 9.416741005055011e-05,
+      "loss": 0.1635,
+      "step": 4130
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 2.0692789554595947,
+      "learning_rate": 9.415254237288136e-05,
+      "loss": 0.1698,
+      "step": 4140
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.22812828421592712,
+      "learning_rate": 9.41376746952126e-05,
+      "loss": 0.1542,
+      "step": 4150
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.1693663597106934,
+      "learning_rate": 9.412280701754387e-05,
+      "loss": 0.1615,
+      "step": 4160
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.3354921042919159,
+      "learning_rate": 9.410793933987511e-05,
+      "loss": 0.1671,
+      "step": 4170
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6962907314300537,
+      "learning_rate": 9.409307166220637e-05,
+      "loss": 0.1613,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.6421184539794922,
+      "learning_rate": 9.407820398453761e-05,
+      "loss": 0.1618,
+      "step": 4190
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7653113007545471,
+      "learning_rate": 9.406333630686888e-05,
+      "loss": 0.1654,
+      "step": 4200
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.4837770462036133,
+      "learning_rate": 9.404846862920012e-05,
+      "loss": 0.1732,
+      "step": 4210
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7182124853134155,
+      "learning_rate": 9.403360095153138e-05,
+      "loss": 0.1543,
+      "step": 4220
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.36238598823547363,
+      "learning_rate": 9.401873327386263e-05,
+      "loss": 0.1632,
+      "step": 4230
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.8226011991500854,
+      "learning_rate": 9.400386559619387e-05,
+      "loss": 0.1523,
+      "step": 4240
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.411160945892334,
+      "learning_rate": 9.398899791852514e-05,
+      "loss": 0.1557,
+      "step": 4250
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.1795895099639893,
+      "learning_rate": 9.397413024085638e-05,
+      "loss": 0.166,
+      "step": 4260
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.4530087113380432,
+      "learning_rate": 9.395926256318764e-05,
+      "loss": 0.1567,
+      "step": 4270
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 2.1539831161499023,
+      "learning_rate": 9.394439488551888e-05,
+      "loss": 0.1627,
+      "step": 4280
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.4007501006126404,
+      "learning_rate": 9.392952720785015e-05,
+      "loss": 0.1609,
+      "step": 4290
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.5847845673561096,
+      "learning_rate": 9.391465953018139e-05,
+      "loss": 0.16,
+      "step": 4300
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.2719428241252899,
+      "learning_rate": 9.389979185251264e-05,
+      "loss": 0.1712,
+      "step": 4310
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.8159836530685425,
+      "learning_rate": 9.38849241748439e-05,
+      "loss": 0.1618,
+      "step": 4320
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 2.801260232925415,
+      "learning_rate": 9.387005649717514e-05,
+      "loss": 0.1607,
+      "step": 4330
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.856196641921997,
+      "learning_rate": 9.38551888195064e-05,
+      "loss": 0.1585,
+      "step": 4340
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8878040909767151,
+      "learning_rate": 9.384032114183765e-05,
+      "loss": 0.1576,
+      "step": 4350
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.753246009349823,
+      "learning_rate": 9.38254534641689e-05,
+      "loss": 0.1705,
+      "step": 4360
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.5893975496292114,
+      "learning_rate": 9.381058578650014e-05,
+      "loss": 0.1597,
+      "step": 4370
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1142498254776,
+      "learning_rate": 9.379571810883141e-05,
+      "loss": 0.1548,
+      "step": 4380
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.9535636901855469,
+      "learning_rate": 9.378085043116267e-05,
+      "loss": 0.1747,
+      "step": 4390
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1514941453933716,
+      "learning_rate": 9.376598275349391e-05,
+      "loss": 0.1619,
+      "step": 4400
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8565515875816345,
+      "learning_rate": 9.375111507582516e-05,
+      "loss": 0.1619,
+      "step": 4410
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.49910810589790344,
+      "learning_rate": 9.37362473981564e-05,
+      "loss": 0.1742,
+      "step": 4420
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.327064573764801,
+      "learning_rate": 9.372137972048767e-05,
+      "loss": 0.15,
+      "step": 4430
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.32551905512809753,
+      "learning_rate": 9.370651204281891e-05,
+      "loss": 0.1617,
+      "step": 4440
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.39346516132354736,
+      "learning_rate": 9.369164436515017e-05,
+      "loss": 0.1597,
+      "step": 4450
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.9522442817687988,
+      "learning_rate": 9.367677668748142e-05,
+      "loss": 0.1538,
+      "step": 4460
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 1.2629470825195312,
+      "learning_rate": 9.366190900981268e-05,
+      "loss": 0.1596,
+      "step": 4470
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.3080795109272003,
+      "learning_rate": 9.364704133214393e-05,
+      "loss": 0.1606,
+      "step": 4480
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3222214877605438,
+      "learning_rate": 9.363217365447517e-05,
+      "loss": 0.1502,
+      "step": 4490
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.2670872211456299,
+      "learning_rate": 9.361730597680643e-05,
+      "loss": 0.1571,
+      "step": 4500
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.1200741529464722,
+      "learning_rate": 9.360243829913768e-05,
+      "loss": 0.1644,
+      "step": 4510
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.3835075795650482,
+      "learning_rate": 9.358757062146894e-05,
+      "loss": 0.1548,
+      "step": 4520
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.4065500497817993,
+      "learning_rate": 9.357270294380018e-05,
+      "loss": 0.1622,
+      "step": 4530
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0452622175216675,
+      "learning_rate": 9.355783526613143e-05,
+      "loss": 0.1646,
+      "step": 4540
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.7965924739837646,
+      "learning_rate": 9.354296758846269e-05,
+      "loss": 0.1689,
+      "step": 4550
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.2334436178207397,
+      "learning_rate": 9.352809991079394e-05,
+      "loss": 0.1754,
+      "step": 4560
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.1238471269607544,
+      "learning_rate": 9.35132322331252e-05,
+      "loss": 0.1593,
+      "step": 4570
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.39519649744033813,
+      "learning_rate": 9.349836455545644e-05,
+      "loss": 0.1664,
+      "step": 4580
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2505687177181244,
+      "learning_rate": 9.348349687778769e-05,
+      "loss": 0.1644,
+      "step": 4590
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.1792103052139282,
+      "learning_rate": 9.346862920011895e-05,
+      "loss": 0.1644,
+      "step": 4600
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.8339989185333252,
+      "learning_rate": 9.34537615224502e-05,
+      "loss": 0.158,
+      "step": 4610
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.2600146532058716,
+      "learning_rate": 9.343889384478144e-05,
+      "loss": 0.1657,
+      "step": 4620
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.7673107385635376,
+      "learning_rate": 9.34240261671127e-05,
+      "loss": 0.1624,
+      "step": 4630
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.3657977283000946,
+      "learning_rate": 9.340915848944395e-05,
+      "loss": 0.165,
+      "step": 4640
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.5369674563407898,
+      "learning_rate": 9.339429081177521e-05,
+      "loss": 0.1643,
+      "step": 4650
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0297385454177856,
+      "learning_rate": 9.337942313410646e-05,
+      "loss": 0.1551,
+      "step": 4660
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.2391917705535889,
+      "learning_rate": 9.33645554564377e-05,
+      "loss": 0.1541,
+      "step": 4670
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.3810855150222778,
+      "learning_rate": 9.334968777876896e-05,
+      "loss": 0.1597,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.7452163696289062,
+      "learning_rate": 9.333482010110021e-05,
+      "loss": 0.1672,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.4795212745666504,
+      "learning_rate": 9.331995242343147e-05,
+      "loss": 0.1686,
+      "step": 4700
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1324572563171387,
+      "learning_rate": 9.330508474576271e-05,
+      "loss": 0.1568,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.038382887840271,
+      "learning_rate": 9.329021706809396e-05,
+      "loss": 0.1605,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.45325565338134766,
+      "learning_rate": 9.327534939042522e-05,
+      "loss": 0.1497,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.37989452481269836,
+      "learning_rate": 9.326048171275647e-05,
+      "loss": 0.1581,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8576381802558899,
+      "learning_rate": 9.324561403508773e-05,
+      "loss": 0.1636,
+      "step": 4750
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.400477796792984,
+      "learning_rate": 9.323074635741897e-05,
+      "loss": 0.1526,
+      "step": 4760
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.37354516983032227,
+      "learning_rate": 9.321587867975022e-05,
+      "loss": 0.1575,
+      "step": 4770
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 2.289604425430298,
+      "learning_rate": 9.320101100208148e-05,
+      "loss": 0.1655,
+      "step": 4780
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.4728387594223022,
+      "learning_rate": 9.318614332441273e-05,
+      "loss": 0.1626,
+      "step": 4790
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.617341160774231,
+      "learning_rate": 9.317127564674399e-05,
+      "loss": 0.1641,
+      "step": 4800
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 1.5229002237319946,
+      "learning_rate": 9.315640796907523e-05,
+      "loss": 0.1575,
+      "step": 4810
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.7396848797798157,
+      "learning_rate": 9.314154029140648e-05,
+      "loss": 0.1692,
+      "step": 4820
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.8826942443847656,
+      "learning_rate": 9.312667261373774e-05,
+      "loss": 0.154,
+      "step": 4830
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8908949494361877,
+      "learning_rate": 9.3111804936069e-05,
+      "loss": 0.1576,
+      "step": 4840
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.829403281211853,
+      "learning_rate": 9.309693725840024e-05,
+      "loss": 0.1626,
+      "step": 4850
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.9040987491607666,
+      "learning_rate": 9.308206958073149e-05,
+      "loss": 0.1617,
+      "step": 4860
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.24965029954910278,
+      "learning_rate": 9.306720190306275e-05,
+      "loss": 0.1677,
+      "step": 4870
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.3442169427871704,
+      "learning_rate": 9.3052334225394e-05,
+      "loss": 0.1563,
+      "step": 4880
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.4237022399902344,
+      "learning_rate": 9.303746654772525e-05,
+      "loss": 0.1645,
+      "step": 4890
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.734087347984314,
+      "learning_rate": 9.30225988700565e-05,
+      "loss": 0.1523,
+      "step": 4900
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.896193265914917,
+      "learning_rate": 9.300773119238775e-05,
+      "loss": 0.167,
+      "step": 4910
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.47176918387413025,
+      "learning_rate": 9.2992863514719e-05,
+      "loss": 0.1584,
+      "step": 4920
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.4008644819259644,
+      "learning_rate": 9.297799583705026e-05,
+      "loss": 0.1584,
+      "step": 4930
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.161672830581665,
+      "learning_rate": 9.29631281593815e-05,
+      "loss": 0.1653,
+      "step": 4940
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.0139836072921753,
+      "learning_rate": 9.294826048171276e-05,
+      "loss": 0.1574,
+      "step": 4950
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.755262017250061,
+      "learning_rate": 9.293339280404401e-05,
+      "loss": 0.1549,
+      "step": 4960
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.763871431350708,
+      "learning_rate": 9.291852512637527e-05,
+      "loss": 0.1598,
+      "step": 4970
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4599681496620178,
+      "learning_rate": 9.290365744870652e-05,
+      "loss": 0.1581,
+      "step": 4980
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6009973287582397,
+      "learning_rate": 9.288878977103776e-05,
+      "loss": 0.1627,
+      "step": 4990
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.40760254859924316,
+      "learning_rate": 9.287392209336903e-05,
+      "loss": 0.1633,
+      "step": 5000
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.41892722249031067,
+      "learning_rate": 9.285905441570027e-05,
+      "loss": 0.1606,
+      "step": 5010
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.4405099153518677,
+      "learning_rate": 9.284418673803153e-05,
+      "loss": 0.1565,
+      "step": 5020
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.49205201864242554,
+      "learning_rate": 9.282931906036277e-05,
+      "loss": 0.1631,
+      "step": 5030
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.3461616635322571,
+      "learning_rate": 9.281445138269402e-05,
+      "loss": 0.1629,
+      "step": 5040
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.33874744176864624,
+      "learning_rate": 9.279958370502529e-05,
+      "loss": 0.1497,
+      "step": 5050
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9239627122879028,
+      "learning_rate": 9.278471602735653e-05,
+      "loss": 0.1568,
+      "step": 5060
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.629723310470581,
+      "learning_rate": 9.276984834968779e-05,
+      "loss": 0.1583,
+      "step": 5070
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.38796573877334595,
+      "learning_rate": 9.275498067201903e-05,
+      "loss": 0.1595,
+      "step": 5080
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.2880231142044067,
+      "learning_rate": 9.27401129943503e-05,
+      "loss": 0.1649,
+      "step": 5090
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.4222949147224426,
+      "learning_rate": 9.272524531668154e-05,
+      "loss": 0.1614,
+      "step": 5100
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.258013367652893,
+      "learning_rate": 9.271037763901279e-05,
+      "loss": 0.1645,
+      "step": 5110
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.2159521579742432,
+      "learning_rate": 9.269550996134405e-05,
+      "loss": 0.1536,
+      "step": 5120
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.3227091133594513,
+      "learning_rate": 9.268064228367529e-05,
+      "loss": 0.1621,
+      "step": 5130
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.7758958339691162,
+      "learning_rate": 9.266577460600656e-05,
+      "loss": 0.1545,
+      "step": 5140
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.6320562362670898,
+      "learning_rate": 9.26509069283378e-05,
+      "loss": 0.1676,
+      "step": 5150
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.9210827350616455,
+      "learning_rate": 9.263603925066905e-05,
+      "loss": 0.157,
+      "step": 5160
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6037706732749939,
+      "learning_rate": 9.26211715730003e-05,
+      "loss": 0.1535,
+      "step": 5170
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 2.3255116939544678,
+      "learning_rate": 9.260630389533156e-05,
+      "loss": 0.1681,
+      "step": 5180
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8426361680030823,
+      "learning_rate": 9.25914362176628e-05,
+      "loss": 0.1633,
+      "step": 5190
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.29783931374549866,
+      "learning_rate": 9.257656853999406e-05,
+      "loss": 0.1557,
+      "step": 5200
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.6732050776481628,
+      "learning_rate": 9.256170086232531e-05,
+      "loss": 0.1604,
+      "step": 5210
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9958044290542603,
+      "learning_rate": 9.254683318465655e-05,
+      "loss": 0.1654,
+      "step": 5220
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.254502534866333,
+      "learning_rate": 9.253196550698782e-05,
+      "loss": 0.1559,
+      "step": 5230
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.2834548056125641,
+      "learning_rate": 9.251709782931906e-05,
+      "loss": 0.1625,
+      "step": 5240
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.3441438376903534,
+      "learning_rate": 9.250223015165032e-05,
+      "loss": 0.1511,
+      "step": 5250
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.31368038058280945,
+      "learning_rate": 9.248736247398156e-05,
+      "loss": 0.1587,
+      "step": 5260
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.5394402146339417,
+      "learning_rate": 9.247249479631283e-05,
+      "loss": 0.1582,
+      "step": 5270
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.29884475469589233,
+      "learning_rate": 9.245762711864407e-05,
+      "loss": 0.1552,
+      "step": 5280
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.530449628829956,
+      "learning_rate": 9.244275944097532e-05,
+      "loss": 0.1703,
+      "step": 5290
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 2.9983556270599365,
+      "learning_rate": 9.242789176330658e-05,
+      "loss": 0.1549,
+      "step": 5300
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.7032681703567505,
+      "learning_rate": 9.241302408563782e-05,
+      "loss": 0.1609,
+      "step": 5310
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.7556791305541992,
+      "learning_rate": 9.239815640796909e-05,
+      "loss": 0.1576,
+      "step": 5320
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.46143102645874023,
+      "learning_rate": 9.238328873030033e-05,
+      "loss": 0.1725,
+      "step": 5330
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.6132158041000366,
+      "learning_rate": 9.236842105263158e-05,
+      "loss": 0.1578,
+      "step": 5340
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.317751407623291,
+      "learning_rate": 9.235355337496282e-05,
+      "loss": 0.1548,
+      "step": 5350
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.5321698188781738,
+      "learning_rate": 9.23386856972941e-05,
+      "loss": 0.1616,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.509246826171875,
+      "learning_rate": 9.232381801962533e-05,
+      "loss": 0.1582,
+      "step": 5370
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.4150841534137726,
+      "learning_rate": 9.230895034195659e-05,
+      "loss": 0.1588,
+      "step": 5380
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.5779004693031311,
+      "learning_rate": 9.229408266428784e-05,
+      "loss": 0.1571,
+      "step": 5390
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8876129984855652,
+      "learning_rate": 9.227921498661909e-05,
+      "loss": 0.1528,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.6326704025268555,
+      "learning_rate": 9.226434730895035e-05,
+      "loss": 0.1632,
+      "step": 5410
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.0380185842514038,
+      "learning_rate": 9.22494796312816e-05,
+      "loss": 0.1577,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.268033504486084,
+      "learning_rate": 9.223461195361285e-05,
+      "loss": 0.1608,
+      "step": 5430
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5740419626235962,
+      "learning_rate": 9.22197442759441e-05,
+      "loss": 0.1648,
+      "step": 5440
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.36971914768218994,
+      "learning_rate": 9.220487659827536e-05,
+      "loss": 0.1601,
+      "step": 5450
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.39549487829208374,
+      "learning_rate": 9.219000892060661e-05,
+      "loss": 0.154,
+      "step": 5460
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9269174933433533,
+      "learning_rate": 9.217514124293785e-05,
+      "loss": 0.1581,
+      "step": 5470
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.122969388961792,
+      "learning_rate": 9.216027356526911e-05,
+      "loss": 0.154,
+      "step": 5480
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.5427556037902832,
+      "learning_rate": 9.214540588760036e-05,
+      "loss": 0.1555,
+      "step": 5490
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5179612636566162,
+      "learning_rate": 9.213053820993162e-05,
+      "loss": 0.1569,
+      "step": 5500
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7659315466880798,
+      "learning_rate": 9.211567053226286e-05,
+      "loss": 0.1667,
+      "step": 5510
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.04986572265625,
+      "learning_rate": 9.210080285459412e-05,
+      "loss": 0.1646,
+      "step": 5520
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.455209493637085,
+      "learning_rate": 9.208593517692537e-05,
+      "loss": 0.1579,
+      "step": 5530
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.3870110511779785,
+      "learning_rate": 9.207106749925662e-05,
+      "loss": 0.1526,
+      "step": 5540
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.7180333733558655,
+      "learning_rate": 9.205619982158788e-05,
+      "loss": 0.16,
+      "step": 5550
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.5366026759147644,
+      "learning_rate": 9.204133214391912e-05,
+      "loss": 0.1571,
+      "step": 5560
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.4958914518356323,
+      "learning_rate": 9.202646446625038e-05,
+      "loss": 0.155,
+      "step": 5570
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.368592768907547,
+      "learning_rate": 9.201159678858163e-05,
+      "loss": 0.1559,
+      "step": 5580
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8154517412185669,
+      "learning_rate": 9.199672911091289e-05,
+      "loss": 0.1526,
+      "step": 5590
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.32349810004234314,
+      "learning_rate": 9.198186143324413e-05,
+      "loss": 0.1689,
+      "step": 5600
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8520846366882324,
+      "learning_rate": 9.196699375557538e-05,
+      "loss": 0.1599,
+      "step": 5610
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.8153768181800842,
+      "learning_rate": 9.195212607790664e-05,
+      "loss": 0.1583,
+      "step": 5620
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0507680177688599,
+      "learning_rate": 9.193725840023789e-05,
+      "loss": 0.1553,
+      "step": 5630
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.34485527873039246,
+      "learning_rate": 9.192239072256915e-05,
+      "loss": 0.1521,
+      "step": 5640
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3874395787715912,
+      "learning_rate": 9.190752304490039e-05,
+      "loss": 0.159,
+      "step": 5650
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.9768149852752686,
+      "learning_rate": 9.189265536723164e-05,
+      "loss": 0.1572,
+      "step": 5660
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.3996683359146118,
+      "learning_rate": 9.18777876895629e-05,
+      "loss": 0.1579,
+      "step": 5670
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.6045333743095398,
+      "learning_rate": 9.186292001189415e-05,
+      "loss": 0.1532,
+      "step": 5680
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.318505883216858,
+      "learning_rate": 9.184805233422539e-05,
+      "loss": 0.1589,
+      "step": 5690
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7289694547653198,
+      "learning_rate": 9.183318465655665e-05,
+      "loss": 0.168,
+      "step": 5700
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2598559856414795,
+      "learning_rate": 9.18183169788879e-05,
+      "loss": 0.1595,
+      "step": 5710
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.0090540647506714,
+      "learning_rate": 9.180344930121916e-05,
+      "loss": 0.1579,
+      "step": 5720
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.2339668273925781,
+      "learning_rate": 9.178858162355041e-05,
+      "loss": 0.1568,
+      "step": 5730
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.9758799076080322,
+      "learning_rate": 9.177371394588165e-05,
+      "loss": 0.1568,
+      "step": 5740
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.6045335531234741,
+      "learning_rate": 9.175884626821291e-05,
+      "loss": 0.1564,
+      "step": 5750
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.40253713726997375,
+      "learning_rate": 9.174397859054416e-05,
+      "loss": 0.1584,
+      "step": 5760
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.332417756319046,
+      "learning_rate": 9.172911091287542e-05,
+      "loss": 0.1492,
+      "step": 5770
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.3337450623512268,
+      "learning_rate": 9.171424323520667e-05,
+      "loss": 0.1496,
+      "step": 5780
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.6006415486335754,
+      "learning_rate": 9.169937555753791e-05,
+      "loss": 0.1664,
+      "step": 5790
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.1703321933746338,
+      "learning_rate": 9.168450787986917e-05,
+      "loss": 0.1515,
+      "step": 5800
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.246114730834961,
+      "learning_rate": 9.166964020220042e-05,
+      "loss": 0.1569,
+      "step": 5810
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.7468287944793701,
+      "learning_rate": 9.165477252453168e-05,
+      "loss": 0.1523,
+      "step": 5820
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8291600942611694,
+      "learning_rate": 9.163990484686292e-05,
+      "loss": 0.1642,
+      "step": 5830
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.5658857226371765,
+      "learning_rate": 9.162503716919417e-05,
+      "loss": 0.151,
+      "step": 5840
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.1321245431900024,
+      "learning_rate": 9.161016949152543e-05,
+      "loss": 0.1544,
+      "step": 5850
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9204747080802917,
+      "learning_rate": 9.159530181385668e-05,
+      "loss": 0.1565,
+      "step": 5860
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.616126298904419,
+      "learning_rate": 9.158043413618794e-05,
+      "loss": 0.1665,
+      "step": 5870
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.37067389488220215,
+      "learning_rate": 9.156556645851918e-05,
+      "loss": 0.1609,
+      "step": 5880
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9909270405769348,
+      "learning_rate": 9.155069878085043e-05,
+      "loss": 0.1481,
+      "step": 5890
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.0309427976608276,
+      "learning_rate": 9.153583110318169e-05,
+      "loss": 0.1523,
+      "step": 5900
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.7706396579742432,
+      "learning_rate": 9.152096342551294e-05,
+      "loss": 0.1608,
+      "step": 5910
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.229520082473755,
+      "learning_rate": 9.150609574784418e-05,
+      "loss": 0.152,
+      "step": 5920
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.4807567298412323,
+      "learning_rate": 9.149122807017544e-05,
+      "loss": 0.156,
+      "step": 5930
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.37514257431030273,
+      "learning_rate": 9.14763603925067e-05,
+      "loss": 0.1587,
+      "step": 5940
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.0821585655212402,
+      "learning_rate": 9.146149271483795e-05,
+      "loss": 0.151,
+      "step": 5950
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8426055908203125,
+      "learning_rate": 9.14466250371692e-05,
+      "loss": 0.1617,
+      "step": 5960
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.3103870451450348,
+      "learning_rate": 9.143175735950044e-05,
+      "loss": 0.1565,
+      "step": 5970
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8296404480934143,
+      "learning_rate": 9.141688968183171e-05,
+      "loss": 0.1533,
+      "step": 5980
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.0698027610778809,
+      "learning_rate": 9.140202200416295e-05,
+      "loss": 0.1537,
+      "step": 5990
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.0414214134216309,
+      "learning_rate": 9.138715432649421e-05,
+      "loss": 0.1532,
+      "step": 6000
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.6497471332550049,
+      "learning_rate": 9.137228664882545e-05,
+      "loss": 0.1606,
+      "step": 6010
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.3416892886161804,
+      "learning_rate": 9.13574189711567e-05,
+      "loss": 0.1499,
+      "step": 6020
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.4196206033229828,
+      "learning_rate": 9.134255129348796e-05,
+      "loss": 0.1449,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3455292582511902,
+      "learning_rate": 9.132768361581921e-05,
+      "loss": 0.149,
+      "step": 6040
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.3500816822052002,
+      "learning_rate": 9.131281593815047e-05,
+      "loss": 0.1543,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.5100831389427185,
+      "learning_rate": 9.129794826048171e-05,
+      "loss": 0.1535,
+      "step": 6060
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.426044225692749,
+      "learning_rate": 9.128308058281298e-05,
+      "loss": 0.1532,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.7652605772018433,
+      "learning_rate": 9.126821290514422e-05,
+      "loss": 0.1612,
+      "step": 6080
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6746863126754761,
+      "learning_rate": 9.125334522747547e-05,
+      "loss": 0.1595,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.6860318779945374,
+      "learning_rate": 9.123847754980673e-05,
+      "loss": 0.156,
+      "step": 6100
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8772507905960083,
+      "learning_rate": 9.122360987213797e-05,
+      "loss": 0.1549,
+      "step": 6110
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.5276684761047363,
+      "learning_rate": 9.120874219446924e-05,
+      "loss": 0.1692,
+      "step": 6120
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.8537567853927612,
+      "learning_rate": 9.119387451680048e-05,
+      "loss": 0.1646,
+      "step": 6130
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.389715552330017,
+      "learning_rate": 9.117900683913173e-05,
+      "loss": 0.1606,
+      "step": 6140
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.4574753940105438,
+      "learning_rate": 9.116413916146298e-05,
+      "loss": 0.1483,
+      "step": 6150
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.41414326429367065,
+      "learning_rate": 9.114927148379424e-05,
+      "loss": 0.1575,
+      "step": 6160
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.36833006143569946,
+      "learning_rate": 9.113440380612549e-05,
+      "loss": 0.1583,
+      "step": 6170
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7688261866569519,
+      "learning_rate": 9.111953612845674e-05,
+      "loss": 0.1553,
+      "step": 6180
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.27714991569519043,
+      "learning_rate": 9.1104668450788e-05,
+      "loss": 0.1607,
+      "step": 6190
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.7245810627937317,
+      "learning_rate": 9.108980077311924e-05,
+      "loss": 0.1495,
+      "step": 6200
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 2.5172832012176514,
+      "learning_rate": 9.10749330954505e-05,
+      "loss": 0.17,
+      "step": 6210
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.39452874660491943,
+      "learning_rate": 9.106006541778175e-05,
+      "loss": 0.152,
+      "step": 6220
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.6295764446258545,
+      "learning_rate": 9.1045197740113e-05,
+      "loss": 0.155,
+      "step": 6230
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8026036620140076,
+      "learning_rate": 9.103033006244424e-05,
+      "loss": 0.149,
+      "step": 6240
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2626301050186157,
+      "learning_rate": 9.101546238477551e-05,
+      "loss": 0.1527,
+      "step": 6250
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.3312925398349762,
+      "learning_rate": 9.100059470710675e-05,
+      "loss": 0.147,
+      "step": 6260
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.411736249923706,
+      "learning_rate": 9.0985727029438e-05,
+      "loss": 0.1511,
+      "step": 6270
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.4057709872722626,
+      "learning_rate": 9.097085935176926e-05,
+      "loss": 0.1557,
+      "step": 6280
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.2932243347167969,
+      "learning_rate": 9.09559916741005e-05,
+      "loss": 0.1552,
+      "step": 6290
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.286341667175293,
+      "learning_rate": 9.094112399643177e-05,
+      "loss": 0.1472,
+      "step": 6300
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.9994767904281616,
+      "learning_rate": 9.092625631876301e-05,
+      "loss": 0.1651,
+      "step": 6310
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.4747790098190308,
+      "learning_rate": 9.091138864109427e-05,
+      "loss": 0.1657,
+      "step": 6320
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8761974573135376,
+      "learning_rate": 9.089652096342551e-05,
+      "loss": 0.1469,
+      "step": 6330
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.5455211997032166,
+      "learning_rate": 9.088165328575678e-05,
+      "loss": 0.1611,
+      "step": 6340
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.3673850297927856,
+      "learning_rate": 9.086678560808802e-05,
+      "loss": 0.1655,
+      "step": 6350
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8073334693908691,
+      "learning_rate": 9.085191793041927e-05,
+      "loss": 0.1555,
+      "step": 6360
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.519607424736023,
+      "learning_rate": 9.083705025275053e-05,
+      "loss": 0.1597,
+      "step": 6370
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8752426505088806,
+      "learning_rate": 9.082218257508177e-05,
+      "loss": 0.1519,
+      "step": 6380
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3730453550815582,
+      "learning_rate": 9.080731489741304e-05,
+      "loss": 0.1535,
+      "step": 6390
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.3529815673828125,
+      "learning_rate": 9.079244721974428e-05,
+      "loss": 0.1618,
+      "step": 6400
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.1450254917144775,
+      "learning_rate": 9.077757954207553e-05,
+      "loss": 0.1618,
+      "step": 6410
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.6045798659324646,
+      "learning_rate": 9.076271186440677e-05,
+      "loss": 0.1594,
+      "step": 6420
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7981701493263245,
+      "learning_rate": 9.074784418673804e-05,
+      "loss": 0.1568,
+      "step": 6430
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.3604912757873535,
+      "learning_rate": 9.07329765090693e-05,
+      "loss": 0.1493,
+      "step": 6440
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8800011873245239,
+      "learning_rate": 9.071810883140054e-05,
+      "loss": 0.149,
+      "step": 6450
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.3312394618988037,
+      "learning_rate": 9.070324115373179e-05,
+      "loss": 0.1507,
+      "step": 6460
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0276165008544922,
+      "learning_rate": 9.068837347606305e-05,
+      "loss": 0.1539,
+      "step": 6470
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.2565714120864868,
+      "learning_rate": 9.06735057983943e-05,
+      "loss": 0.1615,
+      "step": 6480
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5447033643722534,
+      "learning_rate": 9.065863812072554e-05,
+      "loss": 0.1467,
+      "step": 6490
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.4500904083251953,
+      "learning_rate": 9.06437704430568e-05,
+      "loss": 0.1514,
+      "step": 6500
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.5841881036758423,
+      "learning_rate": 9.062890276538805e-05,
+      "loss": 0.1576,
+      "step": 6510
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9153670072555542,
+      "learning_rate": 9.061403508771931e-05,
+      "loss": 0.15,
+      "step": 6520
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.758307695388794,
+      "learning_rate": 9.059916741005056e-05,
+      "loss": 0.1564,
+      "step": 6530
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8557407855987549,
+      "learning_rate": 9.05842997323818e-05,
+      "loss": 0.1545,
+      "step": 6540
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.1173782348632812,
+      "learning_rate": 9.056943205471306e-05,
+      "loss": 0.1519,
+      "step": 6550
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.6467628479003906,
+      "learning_rate": 9.055456437704431e-05,
+      "loss": 0.1522,
+      "step": 6560
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.2692776620388031,
+      "learning_rate": 9.053969669937557e-05,
+      "loss": 0.1558,
+      "step": 6570
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.31524816155433655,
+      "learning_rate": 9.052482902170681e-05,
+      "loss": 0.1534,
+      "step": 6580
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.44416341185569763,
+      "learning_rate": 9.050996134403806e-05,
+      "loss": 0.1574,
+      "step": 6590
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.8063576221466064,
+      "learning_rate": 9.049509366636932e-05,
+      "loss": 0.1636,
+      "step": 6600
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8080287575721741,
+      "learning_rate": 9.048022598870057e-05,
+      "loss": 0.1518,
+      "step": 6610
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.5311675071716309,
+      "learning_rate": 9.046535831103183e-05,
+      "loss": 0.1546,
+      "step": 6620
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.27562418580055237,
+      "learning_rate": 9.045049063336307e-05,
+      "loss": 0.1472,
+      "step": 6630
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.1780297756195068,
+      "learning_rate": 9.043562295569432e-05,
+      "loss": 0.1484,
+      "step": 6640
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2727899551391602,
+      "learning_rate": 9.042075527802558e-05,
+      "loss": 0.153,
+      "step": 6650
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 2.0948357582092285,
+      "learning_rate": 9.040588760035683e-05,
+      "loss": 0.1465,
+      "step": 6660
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.42854198813438416,
+      "learning_rate": 9.039101992268807e-05,
+      "loss": 0.1606,
+      "step": 6670
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.9154927730560303,
+      "learning_rate": 9.037615224501933e-05,
+      "loss": 0.1568,
+      "step": 6680
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.49313873052597046,
+      "learning_rate": 9.036128456735058e-05,
+      "loss": 0.15,
+      "step": 6690
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.1896381378173828,
+      "learning_rate": 9.034641688968184e-05,
+      "loss": 0.1519,
+      "step": 6700
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.2026231288909912,
+      "learning_rate": 9.03315492120131e-05,
+      "loss": 0.1514,
+      "step": 6710
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7389522194862366,
+      "learning_rate": 9.031668153434433e-05,
+      "loss": 0.1506,
+      "step": 6720
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.0212032794952393,
+      "learning_rate": 9.030181385667559e-05,
+      "loss": 0.1541,
+      "step": 6730
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.937395453453064,
+      "learning_rate": 9.028694617900684e-05,
+      "loss": 0.1568,
+      "step": 6740
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.15145954489707947,
+      "eval_runtime": 2482.4757,
+      "eval_samples_per_second": 235.119,
+      "eval_steps_per_second": 3.674,
+      "step": 6746
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 1.5865614414215088,
+      "learning_rate": 9.02720785013381e-05,
+      "loss": 0.1487,
+      "step": 6750
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 2.532686471939087,
+      "learning_rate": 9.025721082366934e-05,
+      "loss": 0.1525,
+      "step": 6760
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.690150797367096,
+      "learning_rate": 9.02423431460006e-05,
+      "loss": 0.1543,
+      "step": 6770
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.32673001289367676,
+      "learning_rate": 9.022747546833185e-05,
+      "loss": 0.1401,
+      "step": 6780
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.432534396648407,
+      "learning_rate": 9.02126077906631e-05,
+      "loss": 0.1482,
+      "step": 6790
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.469577431678772,
+      "learning_rate": 9.019774011299436e-05,
+      "loss": 0.1517,
+      "step": 6800
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.8778477907180786,
+      "learning_rate": 9.01828724353256e-05,
+      "loss": 0.143,
+      "step": 6810
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.7702934741973877,
+      "learning_rate": 9.016800475765686e-05,
+      "loss": 0.1547,
+      "step": 6820
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.1953880786895752,
+      "learning_rate": 9.015313707998811e-05,
+      "loss": 0.1537,
+      "step": 6830
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.4178327322006226,
+      "learning_rate": 9.013826940231936e-05,
+      "loss": 0.1463,
+      "step": 6840
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.5760785341262817,
+      "learning_rate": 9.012340172465062e-05,
+      "loss": 0.1661,
+      "step": 6850
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.34104621410369873,
+      "learning_rate": 9.010853404698186e-05,
+      "loss": 0.156,
+      "step": 6860
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4194226562976837,
+      "learning_rate": 9.009366636931312e-05,
+      "loss": 0.1533,
+      "step": 6870
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.504073143005371,
+      "learning_rate": 9.007879869164437e-05,
+      "loss": 0.1447,
+      "step": 6880
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.8167688846588135,
+      "learning_rate": 9.006393101397563e-05,
+      "loss": 0.1501,
+      "step": 6890
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4795800447463989,
+      "learning_rate": 9.004906333630687e-05,
+      "loss": 0.1444,
+      "step": 6900
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4590786397457123,
+      "learning_rate": 9.003419565863812e-05,
+      "loss": 0.1471,
+      "step": 6910
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7949062585830688,
+      "learning_rate": 9.001932798096938e-05,
+      "loss": 0.149,
+      "step": 6920
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5231783390045166,
+      "learning_rate": 9.000446030330063e-05,
+      "loss": 0.1459,
+      "step": 6930
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.8959499597549438,
+      "learning_rate": 8.998959262563189e-05,
+      "loss": 0.1585,
+      "step": 6940
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.5581046342849731,
+      "learning_rate": 8.997472494796313e-05,
+      "loss": 0.1516,
+      "step": 6950
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 2.33567476272583,
+      "learning_rate": 8.99598572702944e-05,
+      "loss": 0.1512,
+      "step": 6960
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.262761652469635,
+      "learning_rate": 8.994498959262564e-05,
+      "loss": 0.1537,
+      "step": 6970
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.34269747138023376,
+      "learning_rate": 8.993012191495689e-05,
+      "loss": 0.1539,
+      "step": 6980
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.5534690618515015,
+      "learning_rate": 8.991525423728813e-05,
+      "loss": 0.1462,
+      "step": 6990
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.38825082778930664,
+      "learning_rate": 8.990038655961939e-05,
+      "loss": 0.1413,
+      "step": 7000
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9814270734786987,
+      "learning_rate": 8.988551888195064e-05,
+      "loss": 0.1545,
+      "step": 7010
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.6726862192153931,
+      "learning_rate": 8.98706512042819e-05,
+      "loss": 0.1528,
+      "step": 7020
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0980238914489746,
+      "learning_rate": 8.985578352661315e-05,
+      "loss": 0.1505,
+      "step": 7030
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.7269726991653442,
+      "learning_rate": 8.984091584894439e-05,
+      "loss": 0.1519,
+      "step": 7040
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8081973195075989,
+      "learning_rate": 8.982604817127566e-05,
+      "loss": 0.1487,
+      "step": 7050
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.884931206703186,
+      "learning_rate": 8.98111804936069e-05,
+      "loss": 0.1451,
+      "step": 7060
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8839534521102905,
+      "learning_rate": 8.979631281593816e-05,
+      "loss": 0.1534,
+      "step": 7070
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.30532771348953247,
+      "learning_rate": 8.97814451382694e-05,
+      "loss": 0.1534,
+      "step": 7080
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0127397775650024,
+      "learning_rate": 8.976657746060065e-05,
+      "loss": 0.1516,
+      "step": 7090
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.871941328048706,
+      "learning_rate": 8.975170978293192e-05,
+      "loss": 0.1472,
+      "step": 7100
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.0577542781829834,
+      "learning_rate": 8.973684210526316e-05,
+      "loss": 0.1484,
+      "step": 7110
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9246624112129211,
+      "learning_rate": 8.972197442759442e-05,
+      "loss": 0.1467,
+      "step": 7120
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.204312801361084,
+      "learning_rate": 8.970710674992566e-05,
+      "loss": 0.1523,
+      "step": 7130
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.6742764115333557,
+      "learning_rate": 8.969223907225693e-05,
+      "loss": 0.1497,
+      "step": 7140
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.5196964740753174,
+      "learning_rate": 8.967737139458817e-05,
+      "loss": 0.1512,
+      "step": 7150
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.3582424521446228,
+      "learning_rate": 8.966250371691942e-05,
+      "loss": 0.1618,
+      "step": 7160
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.34651896357536316,
+      "learning_rate": 8.964763603925068e-05,
+      "loss": 0.1511,
+      "step": 7170
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.343338966369629,
+      "learning_rate": 8.963276836158192e-05,
+      "loss": 0.1431,
+      "step": 7180
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.5551012754440308,
+      "learning_rate": 8.961790068391319e-05,
+      "loss": 0.1451,
+      "step": 7190
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4299207627773285,
+      "learning_rate": 8.960303300624443e-05,
+      "loss": 0.1464,
+      "step": 7200
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.44814541935920715,
+      "learning_rate": 8.958816532857568e-05,
+      "loss": 0.1508,
+      "step": 7210
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.5933278799057007,
+      "learning_rate": 8.957329765090692e-05,
+      "loss": 0.1558,
+      "step": 7220
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.4118923842906952,
+      "learning_rate": 8.955842997323819e-05,
+      "loss": 0.1492,
+      "step": 7230
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.303074836730957,
+      "learning_rate": 8.954356229556943e-05,
+      "loss": 0.1451,
+      "step": 7240
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.5382298231124878,
+      "learning_rate": 8.952869461790069e-05,
+      "loss": 0.154,
+      "step": 7250
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.7457761764526367,
+      "learning_rate": 8.951382694023194e-05,
+      "loss": 0.1494,
+      "step": 7260
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.7700926661491394,
+      "learning_rate": 8.949895926256318e-05,
+      "loss": 0.1629,
+      "step": 7270
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.2998206615447998,
+      "learning_rate": 8.948409158489445e-05,
+      "loss": 0.1459,
+      "step": 7280
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.41362789273262024,
+      "learning_rate": 8.94692239072257e-05,
+      "loss": 0.1468,
+      "step": 7290
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.507246494293213,
+      "learning_rate": 8.945435622955695e-05,
+      "loss": 0.1526,
+      "step": 7300
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.7437586784362793,
+      "learning_rate": 8.943948855188819e-05,
+      "loss": 0.1632,
+      "step": 7310
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.2855551242828369,
+      "learning_rate": 8.942462087421946e-05,
+      "loss": 0.1485,
+      "step": 7320
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.5423638820648193,
+      "learning_rate": 8.94097531965507e-05,
+      "loss": 0.1572,
+      "step": 7330
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9757463932037354,
+      "learning_rate": 8.939488551888195e-05,
+      "loss": 0.1523,
+      "step": 7340
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.25639691948890686,
+      "learning_rate": 8.938001784121321e-05,
+      "loss": 0.1576,
+      "step": 7350
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.7795055508613586,
+      "learning_rate": 8.936515016354445e-05,
+      "loss": 0.1516,
+      "step": 7360
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.48449787497520447,
+      "learning_rate": 8.935028248587572e-05,
+      "loss": 0.158,
+      "step": 7370
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0721737146377563,
+      "learning_rate": 8.933541480820696e-05,
+      "loss": 0.1523,
+      "step": 7380
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.2508677244186401,
+      "learning_rate": 8.932054713053821e-05,
+      "loss": 0.1454,
+      "step": 7390
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.39935556054115295,
+      "learning_rate": 8.930567945286946e-05,
+      "loss": 0.1543,
+      "step": 7400
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.5586414337158203,
+      "learning_rate": 8.929081177520072e-05,
+      "loss": 0.1596,
+      "step": 7410
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.2489172220230103,
+      "learning_rate": 8.927594409753197e-05,
+      "loss": 0.1548,
+      "step": 7420
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 2.5162367820739746,
+      "learning_rate": 8.926107641986322e-05,
+      "loss": 0.15,
+      "step": 7430
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1505943536758423,
+      "learning_rate": 8.924620874219447e-05,
+      "loss": 0.1472,
+      "step": 7440
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.3584840297698975,
+      "learning_rate": 8.923134106452573e-05,
+      "loss": 0.1535,
+      "step": 7450
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.40715011954307556,
+      "learning_rate": 8.921647338685698e-05,
+      "loss": 0.1556,
+      "step": 7460
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.03001070022583,
+      "learning_rate": 8.920160570918823e-05,
+      "loss": 0.1571,
+      "step": 7470
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.2230408489704132,
+      "learning_rate": 8.918673803151948e-05,
+      "loss": 0.147,
+      "step": 7480
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.4231066703796387,
+      "learning_rate": 8.917187035385073e-05,
+      "loss": 0.157,
+      "step": 7490
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.9302521347999573,
+      "learning_rate": 8.915700267618199e-05,
+      "loss": 0.1538,
+      "step": 7500
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.3577553033828735,
+      "learning_rate": 8.914213499851324e-05,
+      "loss": 0.1447,
+      "step": 7510
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 2.897783041000366,
+      "learning_rate": 8.912726732084449e-05,
+      "loss": 0.1476,
+      "step": 7520
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.3453262448310852,
+      "learning_rate": 8.911239964317574e-05,
+      "loss": 0.1603,
+      "step": 7530
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.2519605159759521,
+      "learning_rate": 8.9097531965507e-05,
+      "loss": 0.1551,
+      "step": 7540
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0790461301803589,
+      "learning_rate": 8.908266428783825e-05,
+      "loss": 0.1472,
+      "step": 7550
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6116156578063965,
+      "learning_rate": 8.906779661016949e-05,
+      "loss": 0.1483,
+      "step": 7560
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.32002952694892883,
+      "learning_rate": 8.905292893250075e-05,
+      "loss": 0.1455,
+      "step": 7570
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 2.1142666339874268,
+      "learning_rate": 8.9038061254832e-05,
+      "loss": 0.1514,
+      "step": 7580
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.7548339366912842,
+      "learning_rate": 8.902319357716326e-05,
+      "loss": 0.1503,
+      "step": 7590
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0314886569976807,
+      "learning_rate": 8.900832589949451e-05,
+      "loss": 0.1484,
+      "step": 7600
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0514923334121704,
+      "learning_rate": 8.899345822182575e-05,
+      "loss": 0.1597,
+      "step": 7610
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.3399512767791748,
+      "learning_rate": 8.8978590544157e-05,
+      "loss": 0.1531,
+      "step": 7620
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.079697608947754,
+      "learning_rate": 8.896372286648826e-05,
+      "loss": 0.1472,
+      "step": 7630
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.370299369096756,
+      "learning_rate": 8.894885518881952e-05,
+      "loss": 0.1448,
+      "step": 7640
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 2.146461248397827,
+      "learning_rate": 8.893398751115076e-05,
+      "loss": 0.1494,
+      "step": 7650
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.532644271850586,
+      "learning_rate": 8.891911983348201e-05,
+      "loss": 0.1442,
+      "step": 7660
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0400909185409546,
+      "learning_rate": 8.890425215581327e-05,
+      "loss": 0.155,
+      "step": 7670
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.5614340901374817,
+      "learning_rate": 8.888938447814452e-05,
+      "loss": 0.149,
+      "step": 7680
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.8623751401901245,
+      "learning_rate": 8.887451680047578e-05,
+      "loss": 0.1646,
+      "step": 7690
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2477229833602905,
+      "learning_rate": 8.885964912280702e-05,
+      "loss": 0.1502,
+      "step": 7700
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.0643510818481445,
+      "learning_rate": 8.884478144513827e-05,
+      "loss": 0.1558,
+      "step": 7710
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 0.7074918746948242,
+      "learning_rate": 8.882991376746953e-05,
+      "loss": 0.1465,
+      "step": 7720
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.3349978923797607,
+      "learning_rate": 8.881504608980078e-05,
+      "loss": 0.1538,
+      "step": 7730
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.33089113235473633,
+      "learning_rate": 8.880017841213202e-05,
+      "loss": 0.1466,
+      "step": 7740
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.4978218078613281,
+      "learning_rate": 8.878531073446328e-05,
+      "loss": 0.155,
+      "step": 7750
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.7654755711555481,
+      "learning_rate": 8.877044305679453e-05,
+      "loss": 0.1467,
+      "step": 7760
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.5333597660064697,
+      "learning_rate": 8.875557537912579e-05,
+      "loss": 0.1523,
+      "step": 7770
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.6722001433372498,
+      "learning_rate": 8.874070770145704e-05,
+      "loss": 0.1503,
+      "step": 7780
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1389368772506714,
+      "learning_rate": 8.872584002378828e-05,
+      "loss": 0.1491,
+      "step": 7790
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.3832188546657562,
+      "learning_rate": 8.871097234611954e-05,
+      "loss": 0.1483,
+      "step": 7800
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1785801649093628,
+      "learning_rate": 8.869610466845079e-05,
+      "loss": 0.1556,
+      "step": 7810
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.5937321186065674,
+      "learning_rate": 8.868123699078205e-05,
+      "loss": 0.1515,
+      "step": 7820
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.9288078546524048,
+      "learning_rate": 8.86663693131133e-05,
+      "loss": 0.1524,
+      "step": 7830
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.9835410714149475,
+      "learning_rate": 8.865150163544454e-05,
+      "loss": 0.1453,
+      "step": 7840
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.201693534851074,
+      "learning_rate": 8.86366339577758e-05,
+      "loss": 0.1516,
+      "step": 7850
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.7419443130493164,
+      "learning_rate": 8.862176628010705e-05,
+      "loss": 0.153,
+      "step": 7860
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.239664077758789,
+      "learning_rate": 8.860689860243831e-05,
+      "loss": 0.1514,
+      "step": 7870
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.9454985857009888,
+      "learning_rate": 8.859203092476955e-05,
+      "loss": 0.1453,
+      "step": 7880
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.5655785202980042,
+      "learning_rate": 8.85771632471008e-05,
+      "loss": 0.1442,
+      "step": 7890
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2254222631454468,
+      "learning_rate": 8.856229556943206e-05,
+      "loss": 0.1555,
+      "step": 7900
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.7310748100280762,
+      "learning_rate": 8.854742789176331e-05,
+      "loss": 0.1514,
+      "step": 7910
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 2.103505849838257,
+      "learning_rate": 8.853256021409457e-05,
+      "loss": 0.1504,
+      "step": 7920
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.45951688289642334,
+      "learning_rate": 8.851769253642581e-05,
+      "loss": 0.1507,
+      "step": 7930
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5322825312614441,
+      "learning_rate": 8.850282485875708e-05,
+      "loss": 0.1455,
+      "step": 7940
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.80711430311203,
+      "learning_rate": 8.848795718108832e-05,
+      "loss": 0.1489,
+      "step": 7950
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.1508015394210815,
+      "learning_rate": 8.847308950341957e-05,
+      "loss": 0.1504,
+      "step": 7960
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.5405661463737488,
+      "learning_rate": 8.845822182575081e-05,
+      "loss": 0.155,
+      "step": 7970
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.4886877238750458,
+      "learning_rate": 8.844335414808207e-05,
+      "loss": 0.1471,
+      "step": 7980
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.6233669519424438,
+      "learning_rate": 8.842848647041332e-05,
+      "loss": 0.1481,
+      "step": 7990
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.8412140011787415,
+      "learning_rate": 8.841361879274458e-05,
+      "loss": 0.1476,
+      "step": 8000
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.6617283821105957,
+      "learning_rate": 8.839875111507583e-05,
+      "loss": 0.1457,
+      "step": 8010
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.756656289100647,
+      "learning_rate": 8.838388343740707e-05,
+      "loss": 0.1436,
+      "step": 8020
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.0248194932937622,
+      "learning_rate": 8.836901575973834e-05,
+      "loss": 0.1483,
+      "step": 8030
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9655269980430603,
+      "learning_rate": 8.835414808206958e-05,
+      "loss": 0.1455,
+      "step": 8040
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.9099483489990234,
+      "learning_rate": 8.833928040440084e-05,
+      "loss": 0.1449,
+      "step": 8050
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.3274220824241638,
+      "learning_rate": 8.832441272673208e-05,
+      "loss": 0.1586,
+      "step": 8060
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.5055317878723145,
+      "learning_rate": 8.830954504906334e-05,
+      "loss": 0.1482,
+      "step": 8070
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.36009564995765686,
+      "learning_rate": 8.829467737139459e-05,
+      "loss": 0.1536,
+      "step": 8080
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.616855263710022,
+      "learning_rate": 8.827980969372584e-05,
+      "loss": 0.1432,
+      "step": 8090
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.257025957107544,
+      "learning_rate": 8.82649420160571e-05,
+      "loss": 0.1491,
+      "step": 8100
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.2415412664413452,
+      "learning_rate": 8.825007433838834e-05,
+      "loss": 0.159,
+      "step": 8110
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.49086448550224304,
+      "learning_rate": 8.823520666071961e-05,
+      "loss": 0.1477,
+      "step": 8120
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.015421748161316,
+      "learning_rate": 8.822033898305085e-05,
+      "loss": 0.1456,
+      "step": 8130
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.44600141048431396,
+      "learning_rate": 8.82054713053821e-05,
+      "loss": 0.1567,
+      "step": 8140
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.0561884641647339,
+      "learning_rate": 8.819060362771336e-05,
+      "loss": 0.1583,
+      "step": 8150
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.7750810384750366,
+      "learning_rate": 8.81757359500446e-05,
+      "loss": 0.1497,
+      "step": 8160
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.3530331552028656,
+      "learning_rate": 8.816086827237587e-05,
+      "loss": 0.1529,
+      "step": 8170
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.564988613128662,
+      "learning_rate": 8.814600059470711e-05,
+      "loss": 0.157,
+      "step": 8180
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 0.599676251411438,
+      "learning_rate": 8.813113291703837e-05,
+      "loss": 0.15,
+      "step": 8190
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.738433837890625,
+      "learning_rate": 8.81162652393696e-05,
+      "loss": 0.1481,
+      "step": 8200
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.467214822769165,
+      "learning_rate": 8.810139756170087e-05,
+      "loss": 0.1456,
+      "step": 8210
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8870519399642944,
+      "learning_rate": 8.808652988403212e-05,
+      "loss": 0.1467,
+      "step": 8220
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.8673567175865173,
+      "learning_rate": 8.807166220636337e-05,
+      "loss": 0.1511,
+      "step": 8230
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.7647720575332642,
+      "learning_rate": 8.805679452869463e-05,
+      "loss": 0.1409,
+      "step": 8240
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.1034643650054932,
+      "learning_rate": 8.804192685102587e-05,
+      "loss": 0.1377,
+      "step": 8250
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.3493852615356445,
+      "learning_rate": 8.802705917335713e-05,
+      "loss": 0.1511,
+      "step": 8260
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.0202926397323608,
+      "learning_rate": 8.801219149568838e-05,
+      "loss": 0.1443,
+      "step": 8270
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.9459919929504395,
+      "learning_rate": 8.799732381801963e-05,
+      "loss": 0.1476,
+      "step": 8280
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.475825309753418,
+      "learning_rate": 8.798245614035087e-05,
+      "loss": 0.1431,
+      "step": 8290
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.3227538466453552,
+      "learning_rate": 8.796758846268214e-05,
+      "loss": 0.151,
+      "step": 8300
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 2.327113151550293,
+      "learning_rate": 8.795272078501338e-05,
+      "loss": 0.1465,
+      "step": 8310
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.28288453817367554,
+      "learning_rate": 8.793785310734464e-05,
+      "loss": 0.1451,
+      "step": 8320
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.3570489883422852,
+      "learning_rate": 8.792298542967589e-05,
+      "loss": 0.1544,
+      "step": 8330
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.3985998630523682,
+      "learning_rate": 8.790811775200713e-05,
+      "loss": 0.1453,
+      "step": 8340
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.7871092557907104,
+      "learning_rate": 8.78932500743384e-05,
+      "loss": 0.1485,
+      "step": 8350
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.5759921669960022,
+      "learning_rate": 8.787838239666964e-05,
+      "loss": 0.1578,
+      "step": 8360
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.36994293332099915,
+      "learning_rate": 8.78635147190009e-05,
+      "loss": 0.1553,
+      "step": 8370
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.2848123610019684,
+      "learning_rate": 8.784864704133214e-05,
+      "loss": 0.1488,
+      "step": 8380
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.373757928609848,
+      "learning_rate": 8.78337793636634e-05,
+      "loss": 0.1434,
+      "step": 8390
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.3947400450706482,
+      "learning_rate": 8.781891168599465e-05,
+      "loss": 0.1433,
+      "step": 8400
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.7297700643539429,
+      "learning_rate": 8.78040440083259e-05,
+      "loss": 0.1489,
+      "step": 8410
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.4975549280643463,
+      "learning_rate": 8.778917633065716e-05,
+      "loss": 0.1556,
+      "step": 8420
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.2790517508983612,
+      "learning_rate": 8.777430865298841e-05,
+      "loss": 0.1478,
+      "step": 8430
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.5845293998718262,
+      "learning_rate": 8.775944097531967e-05,
+      "loss": 0.1446,
+      "step": 8440
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.2284702062606812,
+      "learning_rate": 8.774457329765091e-05,
+      "loss": 0.1506,
+      "step": 8450
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.9352693557739258,
+      "learning_rate": 8.772970561998216e-05,
+      "loss": 0.1447,
+      "step": 8460
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5690743327140808,
+      "learning_rate": 8.771632471008029e-05,
+      "loss": 0.1677,
+      "step": 8470
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.257373571395874,
+      "learning_rate": 8.770145703241154e-05,
+      "loss": 0.1507,
+      "step": 8480
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0186293125152588,
+      "learning_rate": 8.76865893547428e-05,
+      "loss": 0.1531,
+      "step": 8490
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0547682046890259,
+      "learning_rate": 8.767172167707405e-05,
+      "loss": 0.1578,
+      "step": 8500
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7997061014175415,
+      "learning_rate": 8.765685399940529e-05,
+      "loss": 0.1546,
+      "step": 8510
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.921222448348999,
+      "learning_rate": 8.764198632173655e-05,
+      "loss": 0.149,
+      "step": 8520
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.5580063462257385,
+      "learning_rate": 8.76271186440678e-05,
+      "loss": 0.153,
+      "step": 8530
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.184913992881775,
+      "learning_rate": 8.761225096639906e-05,
+      "loss": 0.1519,
+      "step": 8540
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.3542340993881226,
+      "learning_rate": 8.759738328873031e-05,
+      "loss": 0.1558,
+      "step": 8550
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.39081811904907227,
+      "learning_rate": 8.758251561106155e-05,
+      "loss": 0.1499,
+      "step": 8560
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.175268530845642,
+      "learning_rate": 8.756764793339281e-05,
+      "loss": 0.1463,
+      "step": 8570
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.516953706741333,
+      "learning_rate": 8.755278025572406e-05,
+      "loss": 0.15,
+      "step": 8580
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.45542412996292114,
+      "learning_rate": 8.753791257805532e-05,
+      "loss": 0.1443,
+      "step": 8590
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.25929051637649536,
+      "learning_rate": 8.752304490038656e-05,
+      "loss": 0.145,
+      "step": 8600
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.7502154111862183,
+      "learning_rate": 8.750817722271781e-05,
+      "loss": 0.1529,
+      "step": 8610
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.36616241931915283,
+      "learning_rate": 8.749330954504907e-05,
+      "loss": 0.1484,
+      "step": 8620
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.555590808391571,
+      "learning_rate": 8.747844186738032e-05,
+      "loss": 0.1502,
+      "step": 8630
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2765053510665894,
+      "learning_rate": 8.746357418971158e-05,
+      "loss": 0.1497,
+      "step": 8640
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.160529613494873,
+      "learning_rate": 8.744870651204282e-05,
+      "loss": 0.1488,
+      "step": 8650
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.5419323444366455,
+      "learning_rate": 8.743383883437407e-05,
+      "loss": 0.1531,
+      "step": 8660
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.5298939943313599,
+      "learning_rate": 8.741897115670533e-05,
+      "loss": 0.1486,
+      "step": 8670
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.2307096719741821,
+      "learning_rate": 8.740410347903658e-05,
+      "loss": 0.1525,
+      "step": 8680
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.4540654122829437,
+      "learning_rate": 8.738923580136782e-05,
+      "loss": 0.1535,
+      "step": 8690
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.8886024951934814,
+      "learning_rate": 8.737436812369908e-05,
+      "loss": 0.1541,
+      "step": 8700
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.232670545578003,
+      "learning_rate": 8.735950044603033e-05,
+      "loss": 0.1596,
+      "step": 8710
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.6752057075500488,
+      "learning_rate": 8.734463276836159e-05,
+      "loss": 0.1579,
+      "step": 8720
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.6931501030921936,
+      "learning_rate": 8.732976509069284e-05,
+      "loss": 0.1492,
+      "step": 8730
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.9172624349594116,
+      "learning_rate": 8.731489741302409e-05,
+      "loss": 0.1492,
+      "step": 8740
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.35857778787612915,
+      "learning_rate": 8.730002973535534e-05,
+      "loss": 0.1436,
+      "step": 8750
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.6047963500022888,
+      "learning_rate": 8.72851620576866e-05,
+      "loss": 0.1434,
+      "step": 8760
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0199538469314575,
+      "learning_rate": 8.727029438001785e-05,
+      "loss": 0.1467,
+      "step": 8770
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5496332049369812,
+      "learning_rate": 8.725542670234909e-05,
+      "loss": 0.146,
+      "step": 8780
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.5620203614234924,
+      "learning_rate": 8.724055902468035e-05,
+      "loss": 0.1483,
+      "step": 8790
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.36885154247283936,
+      "learning_rate": 8.72256913470116e-05,
+      "loss": 0.1366,
+      "step": 8800
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.235111117362976,
+      "learning_rate": 8.721082366934285e-05,
+      "loss": 0.145,
+      "step": 8810
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.3093607723712921,
+      "learning_rate": 8.719595599167411e-05,
+      "loss": 0.1625,
+      "step": 8820
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.4602314233779907,
+      "learning_rate": 8.718108831400535e-05,
+      "loss": 0.1433,
+      "step": 8830
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.46107497811317444,
+      "learning_rate": 8.71662206363366e-05,
+      "loss": 0.1451,
+      "step": 8840
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.23074732720851898,
+      "learning_rate": 8.715135295866786e-05,
+      "loss": 0.1498,
+      "step": 8850
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.0006585121154785,
+      "learning_rate": 8.713648528099912e-05,
+      "loss": 0.1585,
+      "step": 8860
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.6358426213264465,
+      "learning_rate": 8.712161760333037e-05,
+      "loss": 0.1522,
+      "step": 8870
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.39934954047203064,
+      "learning_rate": 8.710674992566161e-05,
+      "loss": 0.1449,
+      "step": 8880
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5131770372390747,
+      "learning_rate": 8.709188224799288e-05,
+      "loss": 0.1424,
+      "step": 8890
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 3.1670219898223877,
+      "learning_rate": 8.707701457032412e-05,
+      "loss": 0.1516,
+      "step": 8900
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.6028851270675659,
+      "learning_rate": 8.706214689265538e-05,
+      "loss": 0.1485,
+      "step": 8910
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.5696167945861816,
+      "learning_rate": 8.704727921498662e-05,
+      "loss": 0.1451,
+      "step": 8920
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 0.7680513262748718,
+      "learning_rate": 8.703241153731787e-05,
+      "loss": 0.1431,
+      "step": 8930
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.506998896598816,
+      "learning_rate": 8.701754385964913e-05,
+      "loss": 0.1562,
+      "step": 8940
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.24090486764907837,
+      "learning_rate": 8.700267618198038e-05,
+      "loss": 0.1519,
+      "step": 8950
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.36027562618255615,
+      "learning_rate": 8.698780850431164e-05,
+      "loss": 0.1425,
+      "step": 8960
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.2053183317184448,
+      "learning_rate": 8.697294082664288e-05,
+      "loss": 0.1426,
+      "step": 8970
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.4818166494369507,
+      "learning_rate": 8.695807314897415e-05,
+      "loss": 0.1549,
+      "step": 8980
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.4848952293395996,
+      "learning_rate": 8.694320547130539e-05,
+      "loss": 0.1505,
+      "step": 8990
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1278949975967407,
+      "learning_rate": 8.692833779363664e-05,
+      "loss": 0.1514,
+      "step": 9000
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.55078125,
+      "learning_rate": 8.691347011596788e-05,
+      "loss": 0.1522,
+      "step": 9010
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6898670196533203,
+      "learning_rate": 8.689860243829914e-05,
+      "loss": 0.1439,
+      "step": 9020
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.9387264251708984,
+      "learning_rate": 8.688373476063039e-05,
+      "loss": 0.1492,
+      "step": 9030
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6217661499977112,
+      "learning_rate": 8.686886708296165e-05,
+      "loss": 0.1448,
+      "step": 9040
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.3894174098968506,
+      "learning_rate": 8.68539994052929e-05,
+      "loss": 0.1473,
+      "step": 9050
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.6577085852622986,
+      "learning_rate": 8.683913172762414e-05,
+      "loss": 0.1522,
+      "step": 9060
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.36721089482307434,
+      "learning_rate": 8.682426404995541e-05,
+      "loss": 0.1518,
+      "step": 9070
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.7687819004058838,
+      "learning_rate": 8.680939637228665e-05,
+      "loss": 0.1527,
+      "step": 9080
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 2.106863260269165,
+      "learning_rate": 8.679452869461791e-05,
+      "loss": 0.1526,
+      "step": 9090
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.8379847407341003,
+      "learning_rate": 8.677966101694915e-05,
+      "loss": 0.1525,
+      "step": 9100
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.9839054346084595,
+      "learning_rate": 8.67647933392804e-05,
+      "loss": 0.1476,
+      "step": 9110
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.1237411499023438,
+      "learning_rate": 8.674992566161166e-05,
+      "loss": 0.1516,
+      "step": 9120
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2351957559585571,
+      "learning_rate": 8.673505798394291e-05,
+      "loss": 0.1553,
+      "step": 9130
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.330193281173706,
+      "learning_rate": 8.672019030627417e-05,
+      "loss": 0.1507,
+      "step": 9140
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5106310844421387,
+      "learning_rate": 8.670532262860541e-05,
+      "loss": 0.1509,
+      "step": 9150
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.2819235026836395,
+      "learning_rate": 8.669045495093668e-05,
+      "loss": 0.1496,
+      "step": 9160
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3204076290130615,
+      "learning_rate": 8.667558727326792e-05,
+      "loss": 0.1541,
+      "step": 9170
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.3066985309123993,
+      "learning_rate": 8.666071959559917e-05,
+      "loss": 0.1432,
+      "step": 9180
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.5264164805412292,
+      "learning_rate": 8.664585191793041e-05,
+      "loss": 0.1501,
+      "step": 9190
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.0511165857315063,
+      "learning_rate": 8.663098424026167e-05,
+      "loss": 0.1512,
+      "step": 9200
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3407282531261444,
+      "learning_rate": 8.661611656259294e-05,
+      "loss": 0.1483,
+      "step": 9210
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.6586543321609497,
+      "learning_rate": 8.660124888492418e-05,
+      "loss": 0.1528,
+      "step": 9220
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.29302939772605896,
+      "learning_rate": 8.658638120725543e-05,
+      "loss": 0.146,
+      "step": 9230
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.0969610214233398,
+      "learning_rate": 8.657151352958667e-05,
+      "loss": 0.1515,
+      "step": 9240
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.3798691928386688,
+      "learning_rate": 8.655664585191794e-05,
+      "loss": 0.1502,
+      "step": 9250
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5284072756767273,
+      "learning_rate": 8.654177817424918e-05,
+      "loss": 0.1438,
+      "step": 9260
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.5143082141876221,
+      "learning_rate": 8.652691049658044e-05,
+      "loss": 0.1548,
+      "step": 9270
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 2.502887725830078,
+      "learning_rate": 8.65120428189117e-05,
+      "loss": 0.154,
+      "step": 9280
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.34845390915870667,
+      "learning_rate": 8.649717514124293e-05,
+      "loss": 0.1461,
+      "step": 9290
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.576566219329834,
+      "learning_rate": 8.64823074635742e-05,
+      "loss": 0.157,
+      "step": 9300
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.34153446555137634,
+      "learning_rate": 8.646743978590544e-05,
+      "loss": 0.1529,
+      "step": 9310
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.3493391275405884,
+      "learning_rate": 8.64525721082367e-05,
+      "loss": 0.1418,
+      "step": 9320
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.4563460648059845,
+      "learning_rate": 8.643770443056794e-05,
+      "loss": 0.1525,
+      "step": 9330
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.9894744157791138,
+      "learning_rate": 8.642283675289921e-05,
+      "loss": 0.1508,
+      "step": 9340
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.21944831311702728,
+      "learning_rate": 8.640796907523045e-05,
+      "loss": 0.1478,
+      "step": 9350
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.688185691833496,
+      "learning_rate": 8.63931013975617e-05,
+      "loss": 0.1545,
+      "step": 9360
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.3348112106323242,
+      "learning_rate": 8.637823371989296e-05,
+      "loss": 0.1498,
+      "step": 9370
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.32698962092399597,
+      "learning_rate": 8.636336604222421e-05,
+      "loss": 0.1485,
+      "step": 9380
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4172670543193817,
+      "learning_rate": 8.634849836455547e-05,
+      "loss": 0.157,
+      "step": 9390
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.5992726683616638,
+      "learning_rate": 8.633363068688671e-05,
+      "loss": 0.1473,
+      "step": 9400
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 0.4641229510307312,
+      "learning_rate": 8.631876300921796e-05,
+      "loss": 0.1454,
+      "step": 9410
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.2482893466949463,
+      "learning_rate": 8.63038953315492e-05,
+      "loss": 0.1438,
+      "step": 9420
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3719273805618286,
+      "learning_rate": 8.628902765388047e-05,
+      "loss": 0.1467,
+      "step": 9430
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5880813002586365,
+      "learning_rate": 8.627415997621172e-05,
+      "loss": 0.1496,
+      "step": 9440
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5161025524139404,
+      "learning_rate": 8.625929229854297e-05,
+      "loss": 0.1449,
+      "step": 9450
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.3996943533420563,
+      "learning_rate": 8.624442462087422e-05,
+      "loss": 0.1499,
+      "step": 9460
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.5722046494483948,
+      "learning_rate": 8.622955694320548e-05,
+      "loss": 0.1474,
+      "step": 9470
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.8009772300720215,
+      "learning_rate": 8.621617603330359e-05,
+      "loss": 0.1635,
+      "step": 9480
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.959286630153656,
+      "learning_rate": 8.620130835563486e-05,
+      "loss": 0.153,
+      "step": 9490
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.4346505403518677,
+      "learning_rate": 8.61864406779661e-05,
+      "loss": 0.1599,
+      "step": 9500
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.3234134018421173,
+      "learning_rate": 8.617157300029736e-05,
+      "loss": 0.1445,
+      "step": 9510
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.4339183270931244,
+      "learning_rate": 8.615670532262861e-05,
+      "loss": 0.151,
+      "step": 9520
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.9209549427032471,
+      "learning_rate": 8.614183764495987e-05,
+      "loss": 0.1498,
+      "step": 9530
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.8989786505699158,
+      "learning_rate": 8.612696996729112e-05,
+      "loss": 0.1423,
+      "step": 9540
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9294031858444214,
+      "learning_rate": 8.611210228962236e-05,
+      "loss": 0.1435,
+      "step": 9550
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4785991907119751,
+      "learning_rate": 8.609723461195362e-05,
+      "loss": 0.143,
+      "step": 9560
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.7767314314842224,
+      "learning_rate": 8.608236693428486e-05,
+      "loss": 0.1424,
+      "step": 9570
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.2666493654251099,
+      "learning_rate": 8.606749925661613e-05,
+      "loss": 0.1527,
+      "step": 9580
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.4553685188293457,
+      "learning_rate": 8.605263157894738e-05,
+      "loss": 0.1511,
+      "step": 9590
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.3553789556026459,
+      "learning_rate": 8.603776390127862e-05,
+      "loss": 0.1475,
+      "step": 9600
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.8654747009277344,
+      "learning_rate": 8.602289622360988e-05,
+      "loss": 0.1389,
+      "step": 9610
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7433649301528931,
+      "learning_rate": 8.600802854594113e-05,
+      "loss": 0.1477,
+      "step": 9620
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.25381016731262207,
+      "learning_rate": 8.599316086827239e-05,
+      "loss": 0.1494,
+      "step": 9630
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.8853144645690918,
+      "learning_rate": 8.597829319060363e-05,
+      "loss": 0.1529,
+      "step": 9640
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.3927299380302429,
+      "learning_rate": 8.596342551293488e-05,
+      "loss": 0.1487,
+      "step": 9650
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.46614259481430054,
+      "learning_rate": 8.594855783526614e-05,
+      "loss": 0.1521,
+      "step": 9660
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0246176719665527,
+      "learning_rate": 8.593369015759739e-05,
+      "loss": 0.1439,
+      "step": 9670
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.7353618144989014,
+      "learning_rate": 8.591882247992865e-05,
+      "loss": 0.1433,
+      "step": 9680
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.6362217664718628,
+      "learning_rate": 8.590395480225989e-05,
+      "loss": 0.1451,
+      "step": 9690
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.9315203428268433,
+      "learning_rate": 8.588908712459114e-05,
+      "loss": 0.1453,
+      "step": 9700
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.2445534467697144,
+      "learning_rate": 8.58742194469224e-05,
+      "loss": 0.1356,
+      "step": 9710
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.40973347425460815,
+      "learning_rate": 8.585935176925365e-05,
+      "loss": 0.1481,
+      "step": 9720
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.397741436958313,
+      "learning_rate": 8.584448409158489e-05,
+      "loss": 0.1449,
+      "step": 9730
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 3.0569045543670654,
+      "learning_rate": 8.582961641391615e-05,
+      "loss": 0.1561,
+      "step": 9740
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.37079283595085144,
+      "learning_rate": 8.58147487362474e-05,
+      "loss": 0.1542,
+      "step": 9750
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.637252688407898,
+      "learning_rate": 8.579988105857866e-05,
+      "loss": 0.1543,
+      "step": 9760
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.6145210862159729,
+      "learning_rate": 8.578501338090991e-05,
+      "loss": 0.1461,
+      "step": 9770
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.3545942306518555,
+      "learning_rate": 8.577014570324115e-05,
+      "loss": 0.1531,
+      "step": 9780
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.720514714717865,
+      "learning_rate": 8.575527802557241e-05,
+      "loss": 0.1441,
+      "step": 9790
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.3378904461860657,
+      "learning_rate": 8.574041034790366e-05,
+      "loss": 0.1464,
+      "step": 9800
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.7078978419303894,
+      "learning_rate": 8.572554267023492e-05,
+      "loss": 0.1544,
+      "step": 9810
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.368641972541809,
+      "learning_rate": 8.571067499256616e-05,
+      "loss": 0.1477,
+      "step": 9820
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.42889463901519775,
+      "learning_rate": 8.569580731489741e-05,
+      "loss": 0.1487,
+      "step": 9830
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.435215950012207,
+      "learning_rate": 8.568093963722867e-05,
+      "loss": 0.1474,
+      "step": 9840
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.9903163313865662,
+      "learning_rate": 8.566607195955992e-05,
+      "loss": 0.1433,
+      "step": 9850
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.3809950351715088,
+      "learning_rate": 8.565120428189118e-05,
+      "loss": 0.1447,
+      "step": 9860
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.8493334650993347,
+      "learning_rate": 8.563633660422242e-05,
+      "loss": 0.148,
+      "step": 9870
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.33634793758392334,
+      "learning_rate": 8.562146892655367e-05,
+      "loss": 0.1451,
+      "step": 9880
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.403535395860672,
+      "learning_rate": 8.560660124888493e-05,
+      "loss": 0.1467,
+      "step": 9890
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7498649954795837,
+      "learning_rate": 8.559173357121618e-05,
+      "loss": 0.1442,
+      "step": 9900
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.8162224888801575,
+      "learning_rate": 8.557686589354744e-05,
+      "loss": 0.1475,
+      "step": 9910
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.3795585632324219,
+      "learning_rate": 8.556199821587868e-05,
+      "loss": 0.1471,
+      "step": 9920
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.37970471382141113,
+      "learning_rate": 8.554713053820995e-05,
+      "loss": 0.1542,
+      "step": 9930
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.756047248840332,
+      "learning_rate": 8.553226286054119e-05,
+      "loss": 0.1424,
+      "step": 9940
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.36348363757133484,
+      "learning_rate": 8.551739518287244e-05,
+      "loss": 0.141,
+      "step": 9950
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.6128315329551697,
+      "learning_rate": 8.550252750520368e-05,
+      "loss": 0.1555,
+      "step": 9960
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.5416569113731384,
+      "learning_rate": 8.548765982753494e-05,
+      "loss": 0.1541,
+      "step": 9970
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9355424642562866,
+      "learning_rate": 8.54727921498662e-05,
+      "loss": 0.1475,
+      "step": 9980
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.1831605434417725,
+      "learning_rate": 8.545792447219745e-05,
+      "loss": 0.1533,
+      "step": 9990
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0193519592285156,
+      "learning_rate": 8.54430567945287e-05,
+      "loss": 0.1401,
+      "step": 10000
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.3186480700969696,
+      "learning_rate": 8.542818911685994e-05,
+      "loss": 0.1584,
+      "step": 10010
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.3131497800350189,
+      "learning_rate": 8.541332143919121e-05,
+      "loss": 0.1461,
+      "step": 10020
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.8620591759681702,
+      "learning_rate": 8.539845376152245e-05,
+      "loss": 0.1408,
+      "step": 10030
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.469287633895874,
+      "learning_rate": 8.538358608385371e-05,
+      "loss": 0.1462,
+      "step": 10040
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.40487322211265564,
+      "learning_rate": 8.536871840618495e-05,
+      "loss": 0.1419,
+      "step": 10050
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.30698254704475403,
+      "learning_rate": 8.53538507285162e-05,
+      "loss": 0.1473,
+      "step": 10060
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.4092752933502197,
+      "learning_rate": 8.533898305084746e-05,
+      "loss": 0.1605,
+      "step": 10070
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.45507368445396423,
+      "learning_rate": 8.532411537317871e-05,
+      "loss": 0.1482,
+      "step": 10080
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.43545153737068176,
+      "learning_rate": 8.530924769550997e-05,
+      "loss": 0.1506,
+      "step": 10090
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.4408663809299469,
+      "learning_rate": 8.529438001784121e-05,
+      "loss": 0.1448,
+      "step": 10100
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.2926168441772461,
+      "learning_rate": 8.527951234017248e-05,
+      "loss": 0.1538,
+      "step": 10110
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.5207601189613342,
+      "learning_rate": 8.526464466250372e-05,
+      "loss": 0.144,
+      "step": 10120
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4128458499908447,
+      "learning_rate": 8.524977698483497e-05,
+      "loss": 0.1407,
+      "step": 10130
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.6333006620407104,
+      "learning_rate": 8.523490930716622e-05,
+      "loss": 0.1476,
+      "step": 10140
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.300640344619751,
+      "learning_rate": 8.522004162949747e-05,
+      "loss": 0.145,
+      "step": 10150
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.1418519020080566,
+      "learning_rate": 8.520517395182873e-05,
+      "loss": 0.138,
+      "step": 10160
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.2554391324520111,
+      "learning_rate": 8.519030627415998e-05,
+      "loss": 0.1514,
+      "step": 10170
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.49666744470596313,
+      "learning_rate": 8.517543859649124e-05,
+      "loss": 0.1406,
+      "step": 10180
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.7986241579055786,
+      "learning_rate": 8.516057091882248e-05,
+      "loss": 0.1476,
+      "step": 10190
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.6229901313781738,
+      "learning_rate": 8.514570324115374e-05,
+      "loss": 0.1539,
+      "step": 10200
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.0691964626312256,
+      "learning_rate": 8.513083556348499e-05,
+      "loss": 0.1462,
+      "step": 10210
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 0.3225235342979431,
+      "learning_rate": 8.511596788581624e-05,
+      "loss": 0.1574,
+      "step": 10220
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.46615543961524963,
+      "learning_rate": 8.510110020814748e-05,
+      "loss": 0.1395,
+      "step": 10230
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2598484754562378,
+      "learning_rate": 8.508623253047874e-05,
+      "loss": 0.1358,
+      "step": 10240
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3854273855686188,
+      "learning_rate": 8.507136485281e-05,
+      "loss": 0.1458,
+      "step": 10250
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.6581094264984131,
+      "learning_rate": 8.505649717514125e-05,
+      "loss": 0.1506,
+      "step": 10260
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.3710658550262451,
+      "learning_rate": 8.50416294974725e-05,
+      "loss": 0.1439,
+      "step": 10270
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.9312505722045898,
+      "learning_rate": 8.502676181980374e-05,
+      "loss": 0.1486,
+      "step": 10280
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8688971996307373,
+      "learning_rate": 8.501189414213501e-05,
+      "loss": 0.1434,
+      "step": 10290
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.39325767755508423,
+      "learning_rate": 8.499702646446625e-05,
+      "loss": 0.1432,
+      "step": 10300
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.48646825551986694,
+      "learning_rate": 8.49821587867975e-05,
+      "loss": 0.1522,
+      "step": 10310
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.8287198543548584,
+      "learning_rate": 8.496729110912876e-05,
+      "loss": 0.1463,
+      "step": 10320
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.5924571752548218,
+      "learning_rate": 8.495242343146002e-05,
+      "loss": 0.1455,
+      "step": 10330
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 0.2828935980796814,
+      "learning_rate": 8.493755575379127e-05,
+      "loss": 0.1436,
+      "step": 10340
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.714647889137268,
+      "learning_rate": 8.492268807612251e-05,
+      "loss": 0.1476,
+      "step": 10350
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1875629425048828,
+      "learning_rate": 8.490782039845377e-05,
+      "loss": 0.1486,
+      "step": 10360
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.7616733312606812,
+      "learning_rate": 8.489295272078501e-05,
+      "loss": 0.1464,
+      "step": 10370
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.29138264060020447,
+      "learning_rate": 8.487808504311628e-05,
+      "loss": 0.1407,
+      "step": 10380
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.49664175510406494,
+      "learning_rate": 8.486321736544752e-05,
+      "loss": 0.1515,
+      "step": 10390
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.4667070806026459,
+      "learning_rate": 8.484834968777877e-05,
+      "loss": 0.15,
+      "step": 10400
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.3020360469818115,
+      "learning_rate": 8.483348201011003e-05,
+      "loss": 0.145,
+      "step": 10410
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.41862982511520386,
+      "learning_rate": 8.481861433244128e-05,
+      "loss": 0.1495,
+      "step": 10420
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.3439256250858307,
+      "learning_rate": 8.480374665477254e-05,
+      "loss": 0.1509,
+      "step": 10430
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.6848132610321045,
+      "learning_rate": 8.478887897710378e-05,
+      "loss": 0.1456,
+      "step": 10440
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.34469151496887207,
+      "learning_rate": 8.477401129943503e-05,
+      "loss": 0.147,
+      "step": 10450
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 2.268392324447632,
+      "learning_rate": 8.475914362176627e-05,
+      "loss": 0.1561,
+      "step": 10460
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.23007071018219,
+      "learning_rate": 8.474427594409754e-05,
+      "loss": 0.159,
+      "step": 10470
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5129968523979187,
+      "learning_rate": 8.472940826642878e-05,
+      "loss": 0.1468,
+      "step": 10480
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 0.5921646952629089,
+      "learning_rate": 8.471454058876004e-05,
+      "loss": 0.1428,
+      "step": 10490
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.3901660740375519,
+      "learning_rate": 8.469967291109129e-05,
+      "loss": 0.1483,
+      "step": 10500
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.8023537993431091,
+      "learning_rate": 8.468480523342255e-05,
+      "loss": 0.1503,
+      "step": 10510
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.708949625492096,
+      "learning_rate": 8.46699375557538e-05,
+      "loss": 0.1437,
+      "step": 10520
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.8118116855621338,
+      "learning_rate": 8.465506987808504e-05,
+      "loss": 0.1391,
+      "step": 10530
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.32893306016921997,
+      "learning_rate": 8.46402022004163e-05,
+      "loss": 0.1416,
+      "step": 10540
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.5859858393669128,
+      "learning_rate": 8.462533452274754e-05,
+      "loss": 0.1415,
+      "step": 10550
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.7075644135475159,
+      "learning_rate": 8.461046684507881e-05,
+      "loss": 0.1485,
+      "step": 10560
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.966854453086853,
+      "learning_rate": 8.459559916741005e-05,
+      "loss": 0.1433,
+      "step": 10570
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.5331383347511292,
+      "learning_rate": 8.45807314897413e-05,
+      "loss": 0.1508,
+      "step": 10580
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.2724835872650146,
+      "learning_rate": 8.456586381207256e-05,
+      "loss": 0.1507,
+      "step": 10590
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.28108617663383484,
+      "learning_rate": 8.455099613440381e-05,
+      "loss": 0.1453,
+      "step": 10600
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.3450464904308319,
+      "learning_rate": 8.453612845673507e-05,
+      "loss": 0.1478,
+      "step": 10610
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 0.8239097595214844,
+      "learning_rate": 8.452126077906631e-05,
+      "loss": 0.1438,
+      "step": 10620
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.9062354564666748,
+      "learning_rate": 8.450639310139756e-05,
+      "loss": 0.1467,
+      "step": 10630
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6238219141960144,
+      "learning_rate": 8.449152542372882e-05,
+      "loss": 0.1454,
+      "step": 10640
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.089106798171997,
+      "learning_rate": 8.447665774606007e-05,
+      "loss": 0.1393,
+      "step": 10650
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.6828879714012146,
+      "learning_rate": 8.446179006839133e-05,
+      "loss": 0.1471,
+      "step": 10660
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.3843400478363037,
+      "learning_rate": 8.444692239072257e-05,
+      "loss": 0.1414,
+      "step": 10670
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.483210802078247,
+      "learning_rate": 8.443205471305382e-05,
+      "loss": 0.146,
+      "step": 10680
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3872867822647095,
+      "learning_rate": 8.441718703538508e-05,
+      "loss": 0.1386,
+      "step": 10690
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7687630653381348,
+      "learning_rate": 8.440231935771633e-05,
+      "loss": 0.1524,
+      "step": 10700
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3632531762123108,
+      "learning_rate": 8.438745168004758e-05,
+      "loss": 0.1457,
+      "step": 10710
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.7489669919013977,
+      "learning_rate": 8.437258400237883e-05,
+      "loss": 0.149,
+      "step": 10720
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3110584318637848,
+      "learning_rate": 8.435771632471008e-05,
+      "loss": 0.1464,
+      "step": 10730
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.9831928014755249,
+      "learning_rate": 8.434284864704134e-05,
+      "loss": 0.1494,
+      "step": 10740
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.27060386538505554,
+      "learning_rate": 8.43279809693726e-05,
+      "loss": 0.1492,
+      "step": 10750
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 0.3173629939556122,
+      "learning_rate": 8.431311329170384e-05,
+      "loss": 0.1501,
+      "step": 10760
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.2598687410354614,
+      "learning_rate": 8.429824561403509e-05,
+      "loss": 0.1528,
+      "step": 10770
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.29677924513816833,
+      "learning_rate": 8.428337793636634e-05,
+      "loss": 0.1419,
+      "step": 10780
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.3866405487060547,
+      "learning_rate": 8.42685102586976e-05,
+      "loss": 0.1359,
+      "step": 10790
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.707709014415741,
+      "learning_rate": 8.425364258102884e-05,
+      "loss": 0.145,
+      "step": 10800
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.9584312438964844,
+      "learning_rate": 8.42387749033601e-05,
+      "loss": 0.1472,
+      "step": 10810
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.9559441208839417,
+      "learning_rate": 8.422390722569135e-05,
+      "loss": 0.1432,
+      "step": 10820
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.803369402885437,
+      "learning_rate": 8.42090395480226e-05,
+      "loss": 0.1508,
+      "step": 10830
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.5966110229492188,
+      "learning_rate": 8.419417187035386e-05,
+      "loss": 0.1502,
+      "step": 10840
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.5593287944793701,
+      "learning_rate": 8.41793041926851e-05,
+      "loss": 0.145,
+      "step": 10850
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 2.490039348602295,
+      "learning_rate": 8.416443651501636e-05,
+      "loss": 0.1561,
+      "step": 10860
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.7378647327423096,
+      "learning_rate": 8.414956883734761e-05,
+      "loss": 0.1475,
+      "step": 10870
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.7256468534469604,
+      "learning_rate": 8.413470115967887e-05,
+      "loss": 0.1471,
+      "step": 10880
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 0.516965389251709,
+      "learning_rate": 8.41198334820101e-05,
+      "loss": 0.1427,
+      "step": 10890
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.44696128368377686,
+      "learning_rate": 8.410496580434136e-05,
+      "loss": 0.1506,
+      "step": 10900
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.663616418838501,
+      "learning_rate": 8.409009812667262e-05,
+      "loss": 0.1451,
+      "step": 10910
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.6834636330604553,
+      "learning_rate": 8.407523044900387e-05,
+      "loss": 0.1464,
+      "step": 10920
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.3726295232772827,
+      "learning_rate": 8.406036277133513e-05,
+      "loss": 0.1496,
+      "step": 10930
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.9244528412818909,
+      "learning_rate": 8.404549509366637e-05,
+      "loss": 0.1445,
+      "step": 10940
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.3497867584228516,
+      "learning_rate": 8.403062741599762e-05,
+      "loss": 0.1486,
+      "step": 10950
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.5133809447288513,
+      "learning_rate": 8.401575973832888e-05,
+      "loss": 0.1521,
+      "step": 10960
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.28419455885887146,
+      "learning_rate": 8.400089206066013e-05,
+      "loss": 0.1466,
+      "step": 10970
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.7684698104858398,
+      "learning_rate": 8.398602438299139e-05,
+      "loss": 0.15,
+      "step": 10980
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.1589276790618896,
+      "learning_rate": 8.397115670532263e-05,
+      "loss": 0.1472,
+      "step": 10990
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.120748519897461,
+      "learning_rate": 8.39562890276539e-05,
+      "loss": 0.1433,
+      "step": 11000
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4150292873382568,
+      "learning_rate": 8.394142134998514e-05,
+      "loss": 0.1431,
+      "step": 11010
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4702545404434204,
+      "learning_rate": 8.392655367231639e-05,
+      "loss": 0.1456,
+      "step": 11020
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.38210058212280273,
+      "learning_rate": 8.391168599464763e-05,
+      "loss": 0.1436,
+      "step": 11030
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.4304143190383911,
+      "learning_rate": 8.389681831697889e-05,
+      "loss": 0.1487,
+      "step": 11040
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.3502871096134186,
+      "learning_rate": 8.388195063931014e-05,
+      "loss": 0.1469,
+      "step": 11050
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.36102864146232605,
+      "learning_rate": 8.38670829616414e-05,
+      "loss": 0.1552,
+      "step": 11060
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.8794988989830017,
+      "learning_rate": 8.385221528397265e-05,
+      "loss": 0.1436,
+      "step": 11070
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.30111417174339294,
+      "learning_rate": 8.383734760630389e-05,
+      "loss": 0.1411,
+      "step": 11080
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 0.3722517788410187,
+      "learning_rate": 8.382247992863516e-05,
+      "loss": 0.1416,
+      "step": 11090
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.2735258340835571,
+      "learning_rate": 8.38076122509664e-05,
+      "loss": 0.1492,
+      "step": 11100
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.6874326467514038,
+      "learning_rate": 8.379274457329766e-05,
+      "loss": 0.1472,
+      "step": 11110
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.6189324855804443,
+      "learning_rate": 8.37778768956289e-05,
+      "loss": 0.1492,
+      "step": 11120
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.7874268293380737,
+      "learning_rate": 8.376300921796015e-05,
+      "loss": 0.1489,
+      "step": 11130
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.9058316946029663,
+      "learning_rate": 8.374814154029141e-05,
+      "loss": 0.1466,
+      "step": 11140
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.27435657382011414,
+      "learning_rate": 8.373327386262266e-05,
+      "loss": 0.1548,
+      "step": 11150
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.48889249563217163,
+      "learning_rate": 8.371840618495392e-05,
+      "loss": 0.1531,
+      "step": 11160
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.809974730014801,
+      "learning_rate": 8.370353850728516e-05,
+      "loss": 0.1492,
+      "step": 11170
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.6859524846076965,
+      "learning_rate": 8.368867082961643e-05,
+      "loss": 0.1462,
+      "step": 11180
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.093574047088623,
+      "learning_rate": 8.367380315194767e-05,
+      "loss": 0.1498,
+      "step": 11190
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.384391188621521,
+      "learning_rate": 8.365893547427892e-05,
+      "loss": 0.1529,
+      "step": 11200
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.227850317955017,
+      "learning_rate": 8.364406779661016e-05,
+      "loss": 0.1448,
+      "step": 11210
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.0657614469528198,
+      "learning_rate": 8.362920011894142e-05,
+      "loss": 0.1444,
+      "step": 11220
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 0.9369791746139526,
+      "learning_rate": 8.361433244127267e-05,
+      "loss": 0.1484,
+      "step": 11230
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.27957215905189514,
+      "learning_rate": 8.359946476360393e-05,
+      "loss": 0.142,
+      "step": 11240
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.39011701941490173,
+      "learning_rate": 8.358459708593518e-05,
+      "loss": 0.1485,
+      "step": 11250
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4581025242805481,
+      "learning_rate": 8.356972940826642e-05,
+      "loss": 0.1567,
+      "step": 11260
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7829630970954895,
+      "learning_rate": 8.355486173059769e-05,
+      "loss": 0.1439,
+      "step": 11270
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.7222351431846619,
+      "learning_rate": 8.353999405292893e-05,
+      "loss": 0.1411,
+      "step": 11280
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.4510668218135834,
+      "learning_rate": 8.352512637526019e-05,
+      "loss": 0.14,
+      "step": 11290
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.3421786427497864,
+      "learning_rate": 8.351025869759144e-05,
+      "loss": 0.1495,
+      "step": 11300
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.24841587245464325,
+      "learning_rate": 8.34953910199227e-05,
+      "loss": 0.1395,
+      "step": 11310
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.25271904468536377,
+      "learning_rate": 8.348052334225395e-05,
+      "loss": 0.1427,
+      "step": 11320
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.5229553580284119,
+      "learning_rate": 8.34656556645852e-05,
+      "loss": 0.1425,
+      "step": 11330
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 0.3751966953277588,
+      "learning_rate": 8.345078798691645e-05,
+      "loss": 0.1435,
+      "step": 11340
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.0860705375671387,
+      "learning_rate": 8.343592030924769e-05,
+      "loss": 0.1482,
+      "step": 11350
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.6879124641418457,
+      "learning_rate": 8.342105263157896e-05,
+      "loss": 0.1495,
+      "step": 11360
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5659767985343933,
+      "learning_rate": 8.34061849539102e-05,
+      "loss": 0.1421,
+      "step": 11370
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.4345433712005615,
+      "learning_rate": 8.339131727624145e-05,
+      "loss": 0.1499,
+      "step": 11380
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.5097520351409912,
+      "learning_rate": 8.337644959857271e-05,
+      "loss": 0.1554,
+      "step": 11390
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.5569794178009033,
+      "learning_rate": 8.336158192090396e-05,
+      "loss": 0.1438,
+      "step": 11400
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.062078833580017,
+      "learning_rate": 8.334671424323522e-05,
+      "loss": 0.1532,
+      "step": 11410
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.6521728038787842,
+      "learning_rate": 8.333184656556646e-05,
+      "loss": 0.146,
+      "step": 11420
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.404455542564392,
+      "learning_rate": 8.331697888789771e-05,
+      "loss": 0.1506,
+      "step": 11430
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.4847467243671417,
+      "learning_rate": 8.330211121022896e-05,
+      "loss": 0.1416,
+      "step": 11440
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5724780559539795,
+      "learning_rate": 8.328724353256022e-05,
+      "loss": 0.1499,
+      "step": 11450
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.3448980152606964,
+      "learning_rate": 8.327237585489147e-05,
+      "loss": 0.1492,
+      "step": 11460
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.49433252215385437,
+      "learning_rate": 8.325750817722272e-05,
+      "loss": 0.1497,
+      "step": 11470
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2433007955551147,
+      "learning_rate": 8.324264049955398e-05,
+      "loss": 0.1426,
+      "step": 11480
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.5715564489364624,
+      "learning_rate": 8.322777282188523e-05,
+      "loss": 0.1462,
+      "step": 11490
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.7039369940757751,
+      "learning_rate": 8.321290514421648e-05,
+      "loss": 0.1468,
+      "step": 11500
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.6003488898277283,
+      "learning_rate": 8.319803746654773e-05,
+      "loss": 0.1423,
+      "step": 11510
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.4361822009086609,
+      "learning_rate": 8.318316978887898e-05,
+      "loss": 0.1389,
+      "step": 11520
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3091180622577667,
+      "learning_rate": 8.316830211121022e-05,
+      "loss": 0.1503,
+      "step": 11530
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.3191719651222229,
+      "learning_rate": 8.315343443354149e-05,
+      "loss": 0.1499,
+      "step": 11540
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.5728769898414612,
+      "learning_rate": 8.313856675587273e-05,
+      "loss": 0.1539,
+      "step": 11550
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.094186782836914,
+      "learning_rate": 8.312369907820399e-05,
+      "loss": 0.1457,
+      "step": 11560
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.9020346999168396,
+      "learning_rate": 8.310883140053524e-05,
+      "loss": 0.1417,
+      "step": 11570
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.3199603855609894,
+      "learning_rate": 8.30939637228665e-05,
+      "loss": 0.1461,
+      "step": 11580
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.0080779790878296,
+      "learning_rate": 8.307909604519775e-05,
+      "loss": 0.1436,
+      "step": 11590
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.37147411704063416,
+      "learning_rate": 8.306422836752899e-05,
+      "loss": 0.1354,
+      "step": 11600
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.7766250371932983,
+      "learning_rate": 8.304936068986025e-05,
+      "loss": 0.1465,
+      "step": 11610
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.32852333784103394,
+      "learning_rate": 8.303449301219149e-05,
+      "loss": 0.1422,
+      "step": 11620
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.5026690363883972,
+      "learning_rate": 8.301962533452276e-05,
+      "loss": 0.1493,
+      "step": 11630
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.4522212743759155,
+      "learning_rate": 8.300475765685401e-05,
+      "loss": 0.1381,
+      "step": 11640
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.3424636125564575,
+      "learning_rate": 8.298988997918525e-05,
+      "loss": 0.1468,
+      "step": 11650
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.5932589769363403,
+      "learning_rate": 8.29750223015165e-05,
+      "loss": 0.1533,
+      "step": 11660
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.817040741443634,
+      "learning_rate": 8.296015462384776e-05,
+      "loss": 0.1396,
+      "step": 11670
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.25373414158821106,
+      "learning_rate": 8.294528694617902e-05,
+      "loss": 0.147,
+      "step": 11680
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.36040133237838745,
+      "learning_rate": 8.293041926851026e-05,
+      "loss": 0.1508,
+      "step": 11690
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.032599925994873,
+      "learning_rate": 8.291555159084151e-05,
+      "loss": 0.1394,
+      "step": 11700
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.42168989777565,
+      "learning_rate": 8.290068391317277e-05,
+      "loss": 0.1495,
+      "step": 11710
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.49878957867622375,
+      "learning_rate": 8.288581623550402e-05,
+      "loss": 0.1383,
+      "step": 11720
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.8637439012527466,
+      "learning_rate": 8.287094855783528e-05,
+      "loss": 0.1388,
+      "step": 11730
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.34474968910217285,
+      "learning_rate": 8.285608088016652e-05,
+      "loss": 0.1506,
+      "step": 11740
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.9870109558105469,
+      "learning_rate": 8.284121320249777e-05,
+      "loss": 0.1516,
+      "step": 11750
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.502234160900116,
+      "learning_rate": 8.282634552482903e-05,
+      "loss": 0.149,
+      "step": 11760
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.7274667024612427,
+      "learning_rate": 8.281147784716028e-05,
+      "loss": 0.1487,
+      "step": 11770
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.27660071849823,
+      "learning_rate": 8.279661016949152e-05,
+      "loss": 0.1398,
+      "step": 11780
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.35568374395370483,
+      "learning_rate": 8.278174249182278e-05,
+      "loss": 0.1505,
+      "step": 11790
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.317064881324768,
+      "learning_rate": 8.276687481415403e-05,
+      "loss": 0.1522,
+      "step": 11800
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.6406526565551758,
+      "learning_rate": 8.275200713648529e-05,
+      "loss": 0.1518,
+      "step": 11810
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.24043872952461243,
+      "learning_rate": 8.273713945881654e-05,
+      "loss": 0.1514,
+      "step": 11820
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 2.5549800395965576,
+      "learning_rate": 8.272227178114778e-05,
+      "loss": 0.1501,
+      "step": 11830
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.8846415877342224,
+      "learning_rate": 8.270740410347904e-05,
+      "loss": 0.1491,
+      "step": 11840
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.6220967173576355,
+      "learning_rate": 8.269253642581029e-05,
+      "loss": 0.1497,
+      "step": 11850
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.24483416974544525,
+      "learning_rate": 8.267766874814155e-05,
+      "loss": 0.1385,
+      "step": 11860
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.46550866961479187,
+      "learning_rate": 8.266280107047279e-05,
+      "loss": 0.1484,
+      "step": 11870
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.3104701340198517,
+      "learning_rate": 8.264793339280404e-05,
+      "loss": 0.1435,
+      "step": 11880
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 2.2419028282165527,
+      "learning_rate": 8.26330657151353e-05,
+      "loss": 0.1523,
+      "step": 11890
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.3503584563732147,
+      "learning_rate": 8.261819803746655e-05,
+      "loss": 0.1461,
+      "step": 11900
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.2698685824871063,
+      "learning_rate": 8.260333035979781e-05,
+      "loss": 0.148,
+      "step": 11910
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.36493173241615295,
+      "learning_rate": 8.258846268212905e-05,
+      "loss": 0.1447,
+      "step": 11920
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.378431111574173,
+      "learning_rate": 8.25735950044603e-05,
+      "loss": 0.1528,
+      "step": 11930
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.6286464333534241,
+      "learning_rate": 8.255872732679156e-05,
+      "loss": 0.1419,
+      "step": 11940
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.7695555686950684,
+      "learning_rate": 8.254385964912281e-05,
+      "loss": 0.1496,
+      "step": 11950
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.46831145882606506,
+      "learning_rate": 8.252899197145407e-05,
+      "loss": 0.1418,
+      "step": 11960
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.4352415204048157,
+      "learning_rate": 8.251412429378531e-05,
+      "loss": 0.1506,
+      "step": 11970
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2404990941286087,
+      "learning_rate": 8.249925661611658e-05,
+      "loss": 0.1527,
+      "step": 11980
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.8183870315551758,
+      "learning_rate": 8.248438893844782e-05,
+      "loss": 0.1429,
+      "step": 11990
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.2836654484272003,
+      "learning_rate": 8.246952126077907e-05,
+      "loss": 0.1403,
+      "step": 12000
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.32020139694213867,
+      "learning_rate": 8.245465358311032e-05,
+      "loss": 0.1513,
+      "step": 12010
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.3946151733398438,
+      "learning_rate": 8.243978590544157e-05,
+      "loss": 0.1385,
+      "step": 12020
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.3150290250778198,
+      "learning_rate": 8.242491822777282e-05,
+      "loss": 0.1456,
+      "step": 12030
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.21857360005378723,
+      "learning_rate": 8.241005055010408e-05,
+      "loss": 0.1452,
+      "step": 12040
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.4940718412399292,
+      "learning_rate": 8.239518287243533e-05,
+      "loss": 0.1406,
+      "step": 12050
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.5689001083374023,
+      "learning_rate": 8.238031519476658e-05,
+      "loss": 0.1566,
+      "step": 12060
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.8649786710739136,
+      "learning_rate": 8.236544751709784e-05,
+      "loss": 0.1498,
+      "step": 12070
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.0453568696975708,
+      "learning_rate": 8.235057983942908e-05,
+      "loss": 0.1461,
+      "step": 12080
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.147850513458252,
+      "learning_rate": 8.233571216176034e-05,
+      "loss": 0.1438,
+      "step": 12090
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.5856927037239075,
+      "learning_rate": 8.232084448409158e-05,
+      "loss": 0.1477,
+      "step": 12100
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.24364539980888367,
+      "learning_rate": 8.230597680642284e-05,
+      "loss": 0.1489,
+      "step": 12110
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.500666618347168,
+      "learning_rate": 8.229110912875409e-05,
+      "loss": 0.1473,
+      "step": 12120
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.2935793697834015,
+      "learning_rate": 8.227624145108535e-05,
+      "loss": 0.1497,
+      "step": 12130
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3563259541988373,
+      "learning_rate": 8.22613737734166e-05,
+      "loss": 0.1418,
+      "step": 12140
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.3523156940937042,
+      "learning_rate": 8.224650609574784e-05,
+      "loss": 0.1453,
+      "step": 12150
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.0185507535934448,
+      "learning_rate": 8.223163841807911e-05,
+      "loss": 0.1399,
+      "step": 12160
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.5404852628707886,
+      "learning_rate": 8.221677074041035e-05,
+      "loss": 0.1463,
+      "step": 12170
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.4389984607696533,
+      "learning_rate": 8.22019030627416e-05,
+      "loss": 0.1524,
+      "step": 12180
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.39658161997795105,
+      "learning_rate": 8.218703538507285e-05,
+      "loss": 0.1424,
+      "step": 12190
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.3328675329685211,
+      "learning_rate": 8.217216770740411e-05,
+      "loss": 0.1478,
+      "step": 12200
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.8993117213249207,
+      "learning_rate": 8.215730002973536e-05,
+      "loss": 0.1431,
+      "step": 12210
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.31393373012542725,
+      "learning_rate": 8.214243235206661e-05,
+      "loss": 0.153,
+      "step": 12220
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.40346676111221313,
+      "learning_rate": 8.212756467439787e-05,
+      "loss": 0.1552,
+      "step": 12230
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 0.2991549074649811,
+      "learning_rate": 8.211269699672911e-05,
+      "loss": 0.1443,
+      "step": 12240
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.729020357131958,
+      "learning_rate": 8.209782931906038e-05,
+      "loss": 0.1419,
+      "step": 12250
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.1650629043579102,
+      "learning_rate": 8.208296164139162e-05,
+      "loss": 0.1547,
+      "step": 12260
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.2502027750015259,
+      "learning_rate": 8.206809396372287e-05,
+      "loss": 0.1468,
+      "step": 12270
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.664297342300415,
+      "learning_rate": 8.205322628605411e-05,
+      "loss": 0.1468,
+      "step": 12280
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.310942143201828,
+      "learning_rate": 8.203835860838538e-05,
+      "loss": 0.1441,
+      "step": 12290
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.28401151299476624,
+      "learning_rate": 8.202349093071664e-05,
+      "loss": 0.1436,
+      "step": 12300
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 0.32295674085617065,
+      "learning_rate": 8.200862325304788e-05,
+      "loss": 0.1445,
+      "step": 12310
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.49534422159194946,
+      "learning_rate": 8.199375557537913e-05,
+      "loss": 0.1505,
+      "step": 12320
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6425632834434509,
+      "learning_rate": 8.197888789771037e-05,
+      "loss": 0.1489,
+      "step": 12330
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.6739435195922852,
+      "learning_rate": 8.196402022004164e-05,
+      "loss": 0.1453,
+      "step": 12340
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.706912636756897,
+      "learning_rate": 8.194915254237288e-05,
+      "loss": 0.141,
+      "step": 12350
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.9807479381561279,
+      "learning_rate": 8.193428486470414e-05,
+      "loss": 0.1542,
+      "step": 12360
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.7933177947998047,
+      "learning_rate": 8.191941718703539e-05,
+      "loss": 0.1478,
+      "step": 12370
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.555246889591217,
+      "learning_rate": 8.190454950936665e-05,
+      "loss": 0.1593,
+      "step": 12380
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.279439777135849,
+      "learning_rate": 8.18896818316979e-05,
+      "loss": 0.1494,
+      "step": 12390
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.6358618140220642,
+      "learning_rate": 8.187481415402914e-05,
+      "loss": 0.1567,
+      "step": 12400
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.4418742060661316,
+      "learning_rate": 8.18599464763604e-05,
+      "loss": 0.1452,
+      "step": 12410
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.34239551424980164,
+      "learning_rate": 8.184507879869164e-05,
+      "loss": 0.1405,
+      "step": 12420
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.068682074546814,
+      "learning_rate": 8.183021112102291e-05,
+      "loss": 0.1454,
+      "step": 12430
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 0.748917281627655,
+      "learning_rate": 8.181534344335415e-05,
+      "loss": 0.1468,
+      "step": 12440
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.5174105167388916,
+      "learning_rate": 8.18004757656854e-05,
+      "loss": 0.137,
+      "step": 12450
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.29713815450668335,
+      "learning_rate": 8.178560808801666e-05,
+      "loss": 0.1443,
+      "step": 12460
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.3194617033004761,
+      "learning_rate": 8.177074041034791e-05,
+      "loss": 0.1463,
+      "step": 12470
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.9451056718826294,
+      "learning_rate": 8.175587273267917e-05,
+      "loss": 0.1488,
+      "step": 12480
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.5475955009460449,
+      "learning_rate": 8.174100505501041e-05,
+      "loss": 0.1487,
+      "step": 12490
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.4979711174964905,
+      "learning_rate": 8.172613737734166e-05,
+      "loss": 0.1379,
+      "step": 12500
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.6958137154579163,
+      "learning_rate": 8.17112696996729e-05,
+      "loss": 0.1395,
+      "step": 12510
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.6111878156661987,
+      "learning_rate": 8.169640202200417e-05,
+      "loss": 0.1396,
+      "step": 12520
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.024361252784729,
+      "learning_rate": 8.168153434433541e-05,
+      "loss": 0.1453,
+      "step": 12530
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.42374831438064575,
+      "learning_rate": 8.166666666666667e-05,
+      "loss": 0.1441,
+      "step": 12540
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.4444429278373718,
+      "learning_rate": 8.165179898899792e-05,
+      "loss": 0.1515,
+      "step": 12550
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.30336788296699524,
+      "learning_rate": 8.163693131132918e-05,
+      "loss": 0.1548,
+      "step": 12560
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.3255070447921753,
+      "learning_rate": 8.162206363366043e-05,
+      "loss": 0.1498,
+      "step": 12570
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 0.9393750429153442,
+      "learning_rate": 8.160719595599167e-05,
+      "loss": 0.1439,
+      "step": 12580
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.9588826298713684,
+      "learning_rate": 8.159232827832293e-05,
+      "loss": 0.1427,
+      "step": 12590
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.0595064163208008,
+      "learning_rate": 8.157746060065417e-05,
+      "loss": 0.1481,
+      "step": 12600
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.584743082523346,
+      "learning_rate": 8.156259292298544e-05,
+      "loss": 0.1322,
+      "step": 12610
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.6917292475700378,
+      "learning_rate": 8.154772524531668e-05,
+      "loss": 0.1406,
+      "step": 12620
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.39056283235549927,
+      "learning_rate": 8.153285756764793e-05,
+      "loss": 0.1464,
+      "step": 12630
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 2.2323410511016846,
+      "learning_rate": 8.151798988997919e-05,
+      "loss": 0.1497,
+      "step": 12640
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.2180223912000656,
+      "learning_rate": 8.150312221231044e-05,
+      "loss": 0.147,
+      "step": 12650
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.1179107427597046,
+      "learning_rate": 8.14882545346417e-05,
+      "loss": 0.1433,
+      "step": 12660
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.2884316146373749,
+      "learning_rate": 8.147338685697294e-05,
+      "loss": 0.1545,
+      "step": 12670
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.5363054275512695,
+      "learning_rate": 8.14585191793042e-05,
+      "loss": 0.152,
+      "step": 12680
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.2892308831214905,
+      "learning_rate": 8.144365150163545e-05,
+      "loss": 0.1472,
+      "step": 12690
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.8672362565994263,
+      "learning_rate": 8.14287838239667e-05,
+      "loss": 0.1566,
+      "step": 12700
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.8178413510322571,
+      "learning_rate": 8.141391614629796e-05,
+      "loss": 0.1459,
+      "step": 12710
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.3414991497993469,
+      "learning_rate": 8.13990484686292e-05,
+      "loss": 0.1454,
+      "step": 12720
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.8329755067825317,
+      "learning_rate": 8.138418079096045e-05,
+      "loss": 0.1429,
+      "step": 12730
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6989759802818298,
+      "learning_rate": 8.136931311329171e-05,
+      "loss": 0.1533,
+      "step": 12740
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.6167986392974854,
+      "learning_rate": 8.135444543562296e-05,
+      "loss": 0.1415,
+      "step": 12750
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.7545574903488159,
+      "learning_rate": 8.13395777579542e-05,
+      "loss": 0.1451,
+      "step": 12760
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.7932333946228027,
+      "learning_rate": 8.132471008028546e-05,
+      "loss": 0.1473,
+      "step": 12770
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 0.26402798295021057,
+      "learning_rate": 8.130984240261672e-05,
+      "loss": 0.1394,
+      "step": 12780
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5565439462661743,
+      "learning_rate": 8.129497472494797e-05,
+      "loss": 0.14,
+      "step": 12790
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.40671250224113464,
+      "learning_rate": 8.128010704727922e-05,
+      "loss": 0.1454,
+      "step": 12800
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.8522148728370667,
+      "learning_rate": 8.126523936961047e-05,
+      "loss": 0.1558,
+      "step": 12810
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5363799929618835,
+      "learning_rate": 8.125037169194172e-05,
+      "loss": 0.1404,
+      "step": 12820
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.44085416197776794,
+      "learning_rate": 8.123550401427298e-05,
+      "loss": 0.1512,
+      "step": 12830
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.3973836600780487,
+      "learning_rate": 8.122063633660423e-05,
+      "loss": 0.1456,
+      "step": 12840
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.5485923886299133,
+      "learning_rate": 8.120576865893547e-05,
+      "loss": 0.1424,
+      "step": 12850
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.4346306622028351,
+      "learning_rate": 8.119090098126673e-05,
+      "loss": 0.1508,
+      "step": 12860
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.2852851450443268,
+      "learning_rate": 8.117603330359798e-05,
+      "loss": 0.1459,
+      "step": 12870
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.7352854013442993,
+      "learning_rate": 8.116116562592924e-05,
+      "loss": 0.1475,
+      "step": 12880
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.3881273567676544,
+      "learning_rate": 8.114629794826049e-05,
+      "loss": 0.1429,
+      "step": 12890
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.1615941524505615,
+      "learning_rate": 8.113143027059173e-05,
+      "loss": 0.1473,
+      "step": 12900
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.435965359210968,
+      "learning_rate": 8.111656259292299e-05,
+      "loss": 0.1461,
+      "step": 12910
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6154923439025879,
+      "learning_rate": 8.110169491525424e-05,
+      "loss": 0.1438,
+      "step": 12920
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.5796480774879456,
+      "learning_rate": 8.10868272375855e-05,
+      "loss": 0.145,
+      "step": 12930
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.1157699823379517,
+      "learning_rate": 8.107195955991674e-05,
+      "loss": 0.15,
+      "step": 12940
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.9312446117401123,
+      "learning_rate": 8.105709188224799e-05,
+      "loss": 0.1419,
+      "step": 12950
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6173706650733948,
+      "learning_rate": 8.104222420457925e-05,
+      "loss": 0.1464,
+      "step": 12960
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.7330856919288635,
+      "learning_rate": 8.10273565269105e-05,
+      "loss": 0.1507,
+      "step": 12970
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.6786613464355469,
+      "learning_rate": 8.101248884924176e-05,
+      "loss": 0.1486,
+      "step": 12980
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.6271902322769165,
+      "learning_rate": 8.0997621171573e-05,
+      "loss": 0.1497,
+      "step": 12990
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.4697366952896118,
+      "learning_rate": 8.098275349390425e-05,
+      "loss": 0.1436,
+      "step": 13000
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4403237104415894,
+      "learning_rate": 8.096788581623551e-05,
+      "loss": 0.142,
+      "step": 13010
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.9504378437995911,
+      "learning_rate": 8.095301813856676e-05,
+      "loss": 0.1441,
+      "step": 13020
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.271784245967865,
+      "learning_rate": 8.093815046089802e-05,
+      "loss": 0.1419,
+      "step": 13030
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.47144797444343567,
+      "learning_rate": 8.092328278322926e-05,
+      "loss": 0.1471,
+      "step": 13040
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 0.2984504997730255,
+      "learning_rate": 8.090841510556053e-05,
+      "loss": 0.1453,
+      "step": 13050
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.020990252494812,
+      "learning_rate": 8.089354742789177e-05,
+      "loss": 0.1454,
+      "step": 13060
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.3087640702724457,
+      "learning_rate": 8.087867975022302e-05,
+      "loss": 0.1497,
+      "step": 13070
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.4391845464706421,
+      "learning_rate": 8.086381207255426e-05,
+      "loss": 0.1506,
+      "step": 13080
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 2.1297712326049805,
+      "learning_rate": 8.084894439488552e-05,
+      "loss": 0.1502,
+      "step": 13090
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.29619669914245605,
+      "learning_rate": 8.083407671721677e-05,
+      "loss": 0.1547,
+      "step": 13100
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.5125505924224854,
+      "learning_rate": 8.081920903954803e-05,
+      "loss": 0.145,
+      "step": 13110
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.0014768838882446,
+      "learning_rate": 8.080434136187928e-05,
+      "loss": 0.1471,
+      "step": 13120
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2931712865829468,
+      "learning_rate": 8.078947368421052e-05,
+      "loss": 0.1463,
+      "step": 13130
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.7871705293655396,
+      "learning_rate": 8.077460600654179e-05,
+      "loss": 0.1438,
+      "step": 13140
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.4664849042892456,
+      "learning_rate": 8.075973832887303e-05,
+      "loss": 0.1432,
+      "step": 13150
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.5097386837005615,
+      "learning_rate": 8.074487065120429e-05,
+      "loss": 0.1459,
+      "step": 13160
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.2903710603713989,
+      "learning_rate": 8.073000297353553e-05,
+      "loss": 0.1455,
+      "step": 13170
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 0.6520920991897583,
+      "learning_rate": 8.07151352958668e-05,
+      "loss": 0.1491,
+      "step": 13180
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.3731312155723572,
+      "learning_rate": 8.070026761819804e-05,
+      "loss": 0.1372,
+      "step": 13190
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.653975248336792,
+      "learning_rate": 8.06853999405293e-05,
+      "loss": 0.1439,
+      "step": 13200
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.7446883320808411,
+      "learning_rate": 8.067053226286055e-05,
+      "loss": 0.148,
+      "step": 13210
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.1628508567810059,
+      "learning_rate": 8.065566458519179e-05,
+      "loss": 0.1482,
+      "step": 13220
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.871608018875122,
+      "learning_rate": 8.064079690752306e-05,
+      "loss": 0.1473,
+      "step": 13230
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.35242322087287903,
+      "learning_rate": 8.06259292298543e-05,
+      "loss": 0.1481,
+      "step": 13240
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 0.6800448894500732,
+      "learning_rate": 8.061106155218555e-05,
+      "loss": 0.1469,
+      "step": 13250
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6389768719673157,
+      "learning_rate": 8.05961938745168e-05,
+      "loss": 0.1433,
+      "step": 13260
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.30566221475601196,
+      "learning_rate": 8.058132619684806e-05,
+      "loss": 0.1475,
+      "step": 13270
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.6695141196250916,
+      "learning_rate": 8.05664585191793e-05,
+      "loss": 0.1494,
+      "step": 13280
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.8827765583992004,
+      "learning_rate": 8.055159084151056e-05,
+      "loss": 0.1525,
+      "step": 13290
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.7884755730628967,
+      "learning_rate": 8.053672316384181e-05,
+      "loss": 0.1347,
+      "step": 13300
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.29021501541137695,
+      "learning_rate": 8.052185548617306e-05,
+      "loss": 0.1514,
+      "step": 13310
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 3.1000945568084717,
+      "learning_rate": 8.050698780850432e-05,
+      "loss": 0.1576,
+      "step": 13320
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 1.0783339738845825,
+      "learning_rate": 8.049212013083556e-05,
+      "loss": 0.1487,
+      "step": 13330
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.3350972533226013,
+      "learning_rate": 8.047725245316682e-05,
+      "loss": 0.1393,
+      "step": 13340
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.26007136702537537,
+      "learning_rate": 8.046238477549807e-05,
+      "loss": 0.138,
+      "step": 13350
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.37395599484443665,
+      "learning_rate": 8.044751709782933e-05,
+      "loss": 0.1444,
+      "step": 13360
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.48090776801109314,
+      "learning_rate": 8.043264942016058e-05,
+      "loss": 0.146,
+      "step": 13370
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.6154760718345642,
+      "learning_rate": 8.041778174249182e-05,
+      "loss": 0.1423,
+      "step": 13380
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.34295591711997986,
+      "learning_rate": 8.040291406482308e-05,
+      "loss": 0.1515,
+      "step": 13390
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.2805001735687256,
+      "learning_rate": 8.038804638715432e-05,
+      "loss": 0.1539,
+      "step": 13400
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.4166673719882965,
+      "learning_rate": 8.037317870948559e-05,
+      "loss": 0.1498,
+      "step": 13410
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5549523830413818,
+      "learning_rate": 8.035831103181683e-05,
+      "loss": 0.1401,
+      "step": 13420
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.3639054298400879,
+      "learning_rate": 8.034344335414809e-05,
+      "loss": 0.1479,
+      "step": 13430
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.5058225393295288,
+      "learning_rate": 8.032857567647934e-05,
+      "loss": 0.1437,
+      "step": 13440
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.7204506397247314,
+      "learning_rate": 8.03137079988106e-05,
+      "loss": 0.147,
+      "step": 13450
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.34979936480522156,
+      "learning_rate": 8.029884032114185e-05,
+      "loss": 0.1409,
+      "step": 13460
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.71246337890625,
+      "learning_rate": 8.028397264347309e-05,
+      "loss": 0.1442,
+      "step": 13470
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.47565925121307373,
+      "learning_rate": 8.026910496580435e-05,
+      "loss": 0.145,
+      "step": 13480
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.7164893746376038,
+      "learning_rate": 8.025423728813559e-05,
+      "loss": 0.1368,
+      "step": 13490
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.15097101032733917,
+      "eval_runtime": 2479.6392,
+      "eval_samples_per_second": 235.388,
+      "eval_steps_per_second": 3.678,
+      "step": 13492
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3602011501789093,
+      "learning_rate": 8.023936961046685e-05,
+      "loss": 0.1409,
+      "step": 13500
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 1.1467143297195435,
+      "learning_rate": 8.02245019327981e-05,
+      "loss": 0.1445,
+      "step": 13510
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.553377091884613,
+      "learning_rate": 8.020963425512935e-05,
+      "loss": 0.1326,
+      "step": 13520
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7122890949249268,
+      "learning_rate": 8.01947665774606e-05,
+      "loss": 0.1414,
+      "step": 13530
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.6272186040878296,
+      "learning_rate": 8.017989889979186e-05,
+      "loss": 0.1428,
+      "step": 13540
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.5186712741851807,
+      "learning_rate": 8.016503122212312e-05,
+      "loss": 0.1473,
+      "step": 13550
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.7185566425323486,
+      "learning_rate": 8.015016354445436e-05,
+      "loss": 0.1427,
+      "step": 13560
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.32846587896347046,
+      "learning_rate": 8.013529586678561e-05,
+      "loss": 0.143,
+      "step": 13570
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 1.1040384769439697,
+      "learning_rate": 8.012042818911685e-05,
+      "loss": 0.1442,
+      "step": 13580
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 0.3372102379798889,
+      "learning_rate": 8.010556051144812e-05,
+      "loss": 0.1373,
+      "step": 13590
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.5751693248748779,
+      "learning_rate": 8.009069283377936e-05,
+      "loss": 0.1372,
+      "step": 13600
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.8382410407066345,
+      "learning_rate": 8.007582515611062e-05,
+      "loss": 0.1393,
+      "step": 13610
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.2772016227245331,
+      "learning_rate": 8.006095747844187e-05,
+      "loss": 0.136,
+      "step": 13620
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.9653019309043884,
+      "learning_rate": 8.004608980077313e-05,
+      "loss": 0.1389,
+      "step": 13630
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.6972894668579102,
+      "learning_rate": 8.003122212310438e-05,
+      "loss": 0.1497,
+      "step": 13640
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.28461864590644836,
+      "learning_rate": 8.001635444543562e-05,
+      "loss": 0.1389,
+      "step": 13650
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 1.647381067276001,
+      "learning_rate": 8.000148676776688e-05,
+      "loss": 0.1411,
+      "step": 13660
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.6920177340507507,
+      "learning_rate": 7.998661909009813e-05,
+      "loss": 0.1464,
+      "step": 13670
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.29943400621414185,
+      "learning_rate": 7.997175141242939e-05,
+      "loss": 0.141,
+      "step": 13680
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.3627011179924011,
+      "learning_rate": 7.995688373476064e-05,
+      "loss": 0.1414,
+      "step": 13690
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.7024916410446167,
+      "learning_rate": 7.994201605709188e-05,
+      "loss": 0.1403,
+      "step": 13700
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 1.041019082069397,
+      "learning_rate": 7.992714837942314e-05,
+      "loss": 0.1461,
+      "step": 13710
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.9740252494812012,
+      "learning_rate": 7.991228070175439e-05,
+      "loss": 0.1423,
+      "step": 13720
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.3003685176372528,
+      "learning_rate": 7.989741302408565e-05,
+      "loss": 0.1615,
+      "step": 13730
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.32233926653862,
+      "learning_rate": 7.988254534641689e-05,
+      "loss": 0.1328,
+      "step": 13740
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.23946711421012878,
+      "learning_rate": 7.986767766874814e-05,
+      "loss": 0.1512,
+      "step": 13750
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.9506574273109436,
+      "learning_rate": 7.98528099910794e-05,
+      "loss": 0.1485,
+      "step": 13760
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 2.029719114303589,
+      "learning_rate": 7.983794231341065e-05,
+      "loss": 0.1527,
+      "step": 13770
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.22863923013210297,
+      "learning_rate": 7.982307463574191e-05,
+      "loss": 0.1448,
+      "step": 13780
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.300920307636261,
+      "learning_rate": 7.980820695807315e-05,
+      "loss": 0.1477,
+      "step": 13790
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.3894307315349579,
+      "learning_rate": 7.97933392804044e-05,
+      "loss": 0.1428,
+      "step": 13800
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.7516509294509888,
+      "learning_rate": 7.977847160273566e-05,
+      "loss": 0.1423,
+      "step": 13810
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.6397115588188171,
+      "learning_rate": 7.976360392506691e-05,
+      "loss": 0.1446,
+      "step": 13820
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.33174657821655273,
+      "learning_rate": 7.975022301516504e-05,
+      "loss": 0.1415,
+      "step": 13830
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.393704891204834,
+      "learning_rate": 7.973535533749629e-05,
+      "loss": 0.1471,
+      "step": 13840
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.0416827201843262,
+      "learning_rate": 7.972048765982753e-05,
+      "loss": 0.1369,
+      "step": 13850
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 1.268273949623108,
+      "learning_rate": 7.970561998215879e-05,
+      "loss": 0.1461,
+      "step": 13860
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.27585187554359436,
+      "learning_rate": 7.969075230449004e-05,
+      "loss": 0.1483,
+      "step": 13870
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.27429184317588806,
+      "learning_rate": 7.96758846268213e-05,
+      "loss": 0.1434,
+      "step": 13880
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.2318065166473389,
+      "learning_rate": 7.966101694915254e-05,
+      "loss": 0.1402,
+      "step": 13890
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.34538692235946655,
+      "learning_rate": 7.96461492714838e-05,
+      "loss": 0.1368,
+      "step": 13900
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.7623499631881714,
+      "learning_rate": 7.963128159381505e-05,
+      "loss": 0.1447,
+      "step": 13910
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 1.2780420780181885,
+      "learning_rate": 7.96164139161463e-05,
+      "loss": 0.1466,
+      "step": 13920
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.6700235605239868,
+      "learning_rate": 7.960154623847756e-05,
+      "loss": 0.1429,
+      "step": 13930
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.46738964319229126,
+      "learning_rate": 7.95866785608088e-05,
+      "loss": 0.1411,
+      "step": 13940
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.8376699090003967,
+      "learning_rate": 7.957181088314005e-05,
+      "loss": 0.1376,
+      "step": 13950
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.43358296155929565,
+      "learning_rate": 7.955694320547131e-05,
+      "loss": 0.1482,
+      "step": 13960
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.7221712470054626,
+      "learning_rate": 7.954207552780256e-05,
+      "loss": 0.139,
+      "step": 13970
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.6004229187965393,
+      "learning_rate": 7.95272078501338e-05,
+      "loss": 0.1446,
+      "step": 13980
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.7398319840431213,
+      "learning_rate": 7.951234017246506e-05,
+      "loss": 0.14,
+      "step": 13990
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.39415350556373596,
+      "learning_rate": 7.949747249479631e-05,
+      "loss": 0.1425,
+      "step": 14000
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.32933810353279114,
+      "learning_rate": 7.948260481712757e-05,
+      "loss": 0.141,
+      "step": 14010
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.5092155933380127,
+      "learning_rate": 7.946773713945882e-05,
+      "loss": 0.1498,
+      "step": 14020
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.36288583278656,
+      "learning_rate": 7.945286946179007e-05,
+      "loss": 0.1441,
+      "step": 14030
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.4052281081676483,
+      "learning_rate": 7.943800178412132e-05,
+      "loss": 0.1387,
+      "step": 14040
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 1.4487920999526978,
+      "learning_rate": 7.942313410645257e-05,
+      "loss": 0.1451,
+      "step": 14050
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.6891205906867981,
+      "learning_rate": 7.940826642878383e-05,
+      "loss": 0.1397,
+      "step": 14060
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.8474586606025696,
+      "learning_rate": 7.939339875111508e-05,
+      "loss": 0.1501,
+      "step": 14070
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.4147012531757355,
+      "learning_rate": 7.937853107344633e-05,
+      "loss": 0.131,
+      "step": 14080
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 1.2289682626724243,
+      "learning_rate": 7.93636633957776e-05,
+      "loss": 0.1447,
+      "step": 14090
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.4058608114719391,
+      "learning_rate": 7.934879571810884e-05,
+      "loss": 0.144,
+      "step": 14100
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.19358326494693756,
+      "learning_rate": 7.933392804044009e-05,
+      "loss": 0.1414,
+      "step": 14110
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.3566335439682007,
+      "learning_rate": 7.931906036277133e-05,
+      "loss": 0.137,
+      "step": 14120
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.7426595091819763,
+      "learning_rate": 7.93041926851026e-05,
+      "loss": 0.1302,
+      "step": 14130
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.8978757858276367,
+      "learning_rate": 7.928932500743384e-05,
+      "loss": 0.146,
+      "step": 14140
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.46663588285446167,
+      "learning_rate": 7.92744573297651e-05,
+      "loss": 0.144,
+      "step": 14150
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.2703957259654999,
+      "learning_rate": 7.925958965209635e-05,
+      "loss": 0.1376,
+      "step": 14160
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.3722328841686249,
+      "learning_rate": 7.924472197442759e-05,
+      "loss": 0.1486,
+      "step": 14170
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 1.1712533235549927,
+      "learning_rate": 7.922985429675886e-05,
+      "loss": 0.1369,
+      "step": 14180
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.2290862798690796,
+      "learning_rate": 7.92149866190901e-05,
+      "loss": 0.1448,
+      "step": 14190
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.2642858028411865,
+      "learning_rate": 7.920011894142136e-05,
+      "loss": 0.1406,
+      "step": 14200
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.7167410850524902,
+      "learning_rate": 7.91852512637526e-05,
+      "loss": 0.1376,
+      "step": 14210
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 1.1799192428588867,
+      "learning_rate": 7.917038358608387e-05,
+      "loss": 0.1393,
+      "step": 14220
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.2893933951854706,
+      "learning_rate": 7.91555159084151e-05,
+      "loss": 0.1457,
+      "step": 14230
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.45935529470443726,
+      "learning_rate": 7.914064823074636e-05,
+      "loss": 0.1436,
+      "step": 14240
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.2944597005844116,
+      "learning_rate": 7.912578055307762e-05,
+      "loss": 0.1412,
+      "step": 14250
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.5963975191116333,
+      "learning_rate": 7.911091287540886e-05,
+      "loss": 0.1474,
+      "step": 14260
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.6776736378669739,
+      "learning_rate": 7.909604519774013e-05,
+      "loss": 0.1374,
+      "step": 14270
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.3224552869796753,
+      "learning_rate": 7.908117752007137e-05,
+      "loss": 0.1385,
+      "step": 14280
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4411897361278534,
+      "learning_rate": 7.906630984240262e-05,
+      "loss": 0.1445,
+      "step": 14290
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.4700928330421448,
+      "learning_rate": 7.905144216473386e-05,
+      "loss": 0.1388,
+      "step": 14300
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.5549743175506592,
+      "learning_rate": 7.903657448706513e-05,
+      "loss": 0.1418,
+      "step": 14310
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.7805799841880798,
+      "learning_rate": 7.902170680939637e-05,
+      "loss": 0.1429,
+      "step": 14320
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.6569302678108215,
+      "learning_rate": 7.900683913172763e-05,
+      "loss": 0.1413,
+      "step": 14330
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.9454148411750793,
+      "learning_rate": 7.899197145405888e-05,
+      "loss": 0.1422,
+      "step": 14340
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.41172319650650024,
+      "learning_rate": 7.897710377639012e-05,
+      "loss": 0.1411,
+      "step": 14350
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.7529705166816711,
+      "learning_rate": 7.896223609872139e-05,
+      "loss": 0.1388,
+      "step": 14360
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.5935275554656982,
+      "learning_rate": 7.894736842105263e-05,
+      "loss": 0.1448,
+      "step": 14370
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 1.1465091705322266,
+      "learning_rate": 7.893250074338389e-05,
+      "loss": 0.132,
+      "step": 14380
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.5180565714836121,
+      "learning_rate": 7.891763306571513e-05,
+      "loss": 0.1394,
+      "step": 14390
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.3858262896537781,
+      "learning_rate": 7.89027653880464e-05,
+      "loss": 0.1426,
+      "step": 14400
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.3078696131706238,
+      "learning_rate": 7.888789771037765e-05,
+      "loss": 0.1347,
+      "step": 14410
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.2894585430622101,
+      "learning_rate": 7.887303003270889e-05,
+      "loss": 0.1412,
+      "step": 14420
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.9229311943054199,
+      "learning_rate": 7.885816235504015e-05,
+      "loss": 0.145,
+      "step": 14430
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.8589563369750977,
+      "learning_rate": 7.884329467737139e-05,
+      "loss": 0.1423,
+      "step": 14440
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.33812132477760315,
+      "learning_rate": 7.882842699970266e-05,
+      "loss": 0.1312,
+      "step": 14450
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.4707375466823578,
+      "learning_rate": 7.88135593220339e-05,
+      "loss": 0.1333,
+      "step": 14460
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.35323792695999146,
+      "learning_rate": 7.879869164436515e-05,
+      "loss": 0.1503,
+      "step": 14470
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.7402976751327515,
+      "learning_rate": 7.878382396669641e-05,
+      "loss": 0.1429,
+      "step": 14480
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.49308282136917114,
+      "learning_rate": 7.876895628902766e-05,
+      "loss": 0.138,
+      "step": 14490
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.824302613735199,
+      "learning_rate": 7.875408861135892e-05,
+      "loss": 0.1456,
+      "step": 14500
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 1.5500679016113281,
+      "learning_rate": 7.873922093369016e-05,
+      "loss": 0.1466,
+      "step": 14510
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.34378743171691895,
+      "learning_rate": 7.872435325602141e-05,
+      "loss": 0.1382,
+      "step": 14520
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 0.48903822898864746,
+      "learning_rate": 7.870948557835265e-05,
+      "loss": 0.1398,
+      "step": 14530
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.954559862613678,
+      "learning_rate": 7.869461790068392e-05,
+      "loss": 0.1406,
+      "step": 14540
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.3512939214706421,
+      "learning_rate": 7.867975022301516e-05,
+      "loss": 0.1404,
+      "step": 14550
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.6540653705596924,
+      "learning_rate": 7.866488254534642e-05,
+      "loss": 0.1399,
+      "step": 14560
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.4381154775619507,
+      "learning_rate": 7.865001486767767e-05,
+      "loss": 0.1456,
+      "step": 14570
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.2611878514289856,
+      "learning_rate": 7.863514719000893e-05,
+      "loss": 0.1436,
+      "step": 14580
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.1198517084121704,
+      "learning_rate": 7.862027951234018e-05,
+      "loss": 0.1421,
+      "step": 14590
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 1.4881504774093628,
+      "learning_rate": 7.860541183467142e-05,
+      "loss": 0.1431,
+      "step": 14600
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.5614832043647766,
+      "learning_rate": 7.859054415700268e-05,
+      "loss": 0.1444,
+      "step": 14610
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 1.2753747701644897,
+      "learning_rate": 7.857567647933393e-05,
+      "loss": 0.1355,
+      "step": 14620
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.7569044232368469,
+      "learning_rate": 7.856080880166519e-05,
+      "loss": 0.1403,
+      "step": 14630
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.8014097213745117,
+      "learning_rate": 7.854594112399643e-05,
+      "loss": 0.1444,
+      "step": 14640
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.4055616855621338,
+      "learning_rate": 7.853107344632768e-05,
+      "loss": 0.1384,
+      "step": 14650
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.8355204463005066,
+      "learning_rate": 7.851620576865894e-05,
+      "loss": 0.1374,
+      "step": 14660
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.5635973215103149,
+      "learning_rate": 7.85013380909902e-05,
+      "loss": 0.1453,
+      "step": 14670
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.3572874069213867,
+      "learning_rate": 7.848647041332145e-05,
+      "loss": 0.1474,
+      "step": 14680
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.31111082434654236,
+      "learning_rate": 7.847160273565269e-05,
+      "loss": 0.1436,
+      "step": 14690
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.38512876629829407,
+      "learning_rate": 7.845673505798394e-05,
+      "loss": 0.138,
+      "step": 14700
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.26960206031799316,
+      "learning_rate": 7.84418673803152e-05,
+      "loss": 0.147,
+      "step": 14710
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.29528844356536865,
+      "learning_rate": 7.842699970264645e-05,
+      "loss": 0.1416,
+      "step": 14720
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.3126089572906494,
+      "learning_rate": 7.841213202497771e-05,
+      "loss": 0.1417,
+      "step": 14730
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 1.1311891078948975,
+      "learning_rate": 7.839726434730895e-05,
+      "loss": 0.1365,
+      "step": 14740
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.27010664343833923,
+      "learning_rate": 7.83823966696402e-05,
+      "loss": 0.1367,
+      "step": 14750
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 1.2416595220565796,
+      "learning_rate": 7.836752899197146e-05,
+      "loss": 0.137,
+      "step": 14760
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.2980870306491852,
+      "learning_rate": 7.835266131430271e-05,
+      "loss": 0.1486,
+      "step": 14770
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.44527989625930786,
+      "learning_rate": 7.833779363663396e-05,
+      "loss": 0.1392,
+      "step": 14780
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.36573994159698486,
+      "learning_rate": 7.832292595896521e-05,
+      "loss": 0.1469,
+      "step": 14790
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.3939895033836365,
+      "learning_rate": 7.830805828129647e-05,
+      "loss": 0.1449,
+      "step": 14800
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.5952978134155273,
+      "learning_rate": 7.829319060362772e-05,
+      "loss": 0.1399,
+      "step": 14810
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.4172046184539795,
+      "learning_rate": 7.827832292595897e-05,
+      "loss": 0.1417,
+      "step": 14820
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.6859674453735352,
+      "learning_rate": 7.826345524829022e-05,
+      "loss": 0.1503,
+      "step": 14830
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.4280993640422821,
+      "learning_rate": 7.824858757062147e-05,
+      "loss": 0.1466,
+      "step": 14840
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.28708526492118835,
+      "learning_rate": 7.823371989295273e-05,
+      "loss": 0.1401,
+      "step": 14850
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 1.4044026136398315,
+      "learning_rate": 7.821885221528398e-05,
+      "loss": 0.142,
+      "step": 14860
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.4096578061580658,
+      "learning_rate": 7.820398453761522e-05,
+      "loss": 0.1433,
+      "step": 14870
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.49273645877838135,
+      "learning_rate": 7.818911685994648e-05,
+      "loss": 0.1411,
+      "step": 14880
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.1558241844177246,
+      "learning_rate": 7.817424918227773e-05,
+      "loss": 0.1325,
+      "step": 14890
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.2521809935569763,
+      "learning_rate": 7.815938150460899e-05,
+      "loss": 0.1436,
+      "step": 14900
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.3384026288986206,
+      "learning_rate": 7.814451382694024e-05,
+      "loss": 0.1353,
+      "step": 14910
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.9027255177497864,
+      "learning_rate": 7.812964614927148e-05,
+      "loss": 0.1396,
+      "step": 14920
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.5526256561279297,
+      "learning_rate": 7.811477847160274e-05,
+      "loss": 0.1385,
+      "step": 14930
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 1.069063425064087,
+      "learning_rate": 7.809991079393399e-05,
+      "loss": 0.1359,
+      "step": 14940
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.2974967956542969,
+      "learning_rate": 7.808504311626525e-05,
+      "loss": 0.1342,
+      "step": 14950
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.25086086988449097,
+      "learning_rate": 7.807017543859649e-05,
+      "loss": 0.1332,
+      "step": 14960
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.6883974671363831,
+      "learning_rate": 7.805530776092774e-05,
+      "loss": 0.1399,
+      "step": 14970
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.8510827422142029,
+      "learning_rate": 7.8040440083259e-05,
+      "loss": 0.1441,
+      "step": 14980
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 1.4618726968765259,
+      "learning_rate": 7.802557240559025e-05,
+      "loss": 0.147,
+      "step": 14990
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.8736234307289124,
+      "learning_rate": 7.80107047279215e-05,
+      "loss": 0.1422,
+      "step": 15000
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.5815086364746094,
+      "learning_rate": 7.799583705025275e-05,
+      "loss": 0.1413,
+      "step": 15010
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.29546529054641724,
+      "learning_rate": 7.7980969372584e-05,
+      "loss": 0.1404,
+      "step": 15020
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.3832775056362152,
+      "learning_rate": 7.796610169491526e-05,
+      "loss": 0.1327,
+      "step": 15030
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.2785964608192444,
+      "learning_rate": 7.795123401724651e-05,
+      "loss": 0.1329,
+      "step": 15040
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.5065478086471558,
+      "learning_rate": 7.793636633957775e-05,
+      "loss": 0.1527,
+      "step": 15050
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.5339558124542236,
+      "learning_rate": 7.792149866190901e-05,
+      "loss": 0.1431,
+      "step": 15060
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 1.5240484476089478,
+      "learning_rate": 7.790663098424028e-05,
+      "loss": 0.1465,
+      "step": 15070
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3933528661727905,
+      "learning_rate": 7.789176330657152e-05,
+      "loss": 0.1408,
+      "step": 15080
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.600188672542572,
+      "learning_rate": 7.787689562890277e-05,
+      "loss": 0.1413,
+      "step": 15090
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.3889021575450897,
+      "learning_rate": 7.786202795123401e-05,
+      "loss": 0.1377,
+      "step": 15100
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.4499037265777588,
+      "learning_rate": 7.784716027356528e-05,
+      "loss": 0.1401,
+      "step": 15110
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.772075355052948,
+      "learning_rate": 7.783229259589652e-05,
+      "loss": 0.1359,
+      "step": 15120
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.5900905132293701,
+      "learning_rate": 7.781742491822778e-05,
+      "loss": 0.1374,
+      "step": 15130
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 1.9632419347763062,
+      "learning_rate": 7.780255724055903e-05,
+      "loss": 0.1485,
+      "step": 15140
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 1.079269289970398,
+      "learning_rate": 7.778768956289027e-05,
+      "loss": 0.1428,
+      "step": 15150
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.5753262042999268,
+      "learning_rate": 7.777282188522154e-05,
+      "loss": 0.1429,
+      "step": 15160
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.28684714436531067,
+      "learning_rate": 7.775795420755278e-05,
+      "loss": 0.1363,
+      "step": 15170
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.2784688174724579,
+      "learning_rate": 7.774308652988404e-05,
+      "loss": 0.1447,
+      "step": 15180
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.4321437478065491,
+      "learning_rate": 7.772821885221528e-05,
+      "loss": 0.1437,
+      "step": 15190
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.4593127369880676,
+      "learning_rate": 7.771335117454655e-05,
+      "loss": 0.1394,
+      "step": 15200
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.6981044411659241,
+      "learning_rate": 7.769848349687779e-05,
+      "loss": 0.1469,
+      "step": 15210
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.9199114441871643,
+      "learning_rate": 7.768361581920904e-05,
+      "loss": 0.1396,
+      "step": 15220
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.5728020668029785,
+      "learning_rate": 7.76687481415403e-05,
+      "loss": 0.1433,
+      "step": 15230
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.3551998734474182,
+      "learning_rate": 7.765388046387154e-05,
+      "loss": 0.1418,
+      "step": 15240
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.25206756591796875,
+      "learning_rate": 7.763901278620281e-05,
+      "loss": 0.1405,
+      "step": 15250
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.40802913904190063,
+      "learning_rate": 7.762414510853405e-05,
+      "loss": 0.1406,
+      "step": 15260
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.4538644552230835,
+      "learning_rate": 7.76092774308653e-05,
+      "loss": 0.1414,
+      "step": 15270
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.5877019166946411,
+      "learning_rate": 7.759440975319655e-05,
+      "loss": 0.1398,
+      "step": 15280
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.7143536806106567,
+      "learning_rate": 7.757954207552781e-05,
+      "loss": 0.1444,
+      "step": 15290
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.8385623693466187,
+      "learning_rate": 7.756467439785905e-05,
+      "loss": 0.1453,
+      "step": 15300
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.5438138246536255,
+      "learning_rate": 7.754980672019031e-05,
+      "loss": 0.1476,
+      "step": 15310
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 1.067368507385254,
+      "learning_rate": 7.753493904252156e-05,
+      "loss": 0.1493,
+      "step": 15320
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.43237921595573425,
+      "learning_rate": 7.75200713648528e-05,
+      "loss": 0.1418,
+      "step": 15330
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 0.7449706792831421,
+      "learning_rate": 7.750520368718407e-05,
+      "loss": 0.1372,
+      "step": 15340
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 1.315883994102478,
+      "learning_rate": 7.749033600951531e-05,
+      "loss": 0.1482,
+      "step": 15350
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.20625711977481842,
+      "learning_rate": 7.747546833184657e-05,
+      "loss": 0.1374,
+      "step": 15360
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.29170262813568115,
+      "learning_rate": 7.746060065417781e-05,
+      "loss": 0.1434,
+      "step": 15370
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.6575577855110168,
+      "learning_rate": 7.744573297650908e-05,
+      "loss": 0.1411,
+      "step": 15380
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.36040765047073364,
+      "learning_rate": 7.743086529884032e-05,
+      "loss": 0.1387,
+      "step": 15390
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.6309950947761536,
+      "learning_rate": 7.741599762117158e-05,
+      "loss": 0.1478,
+      "step": 15400
+    },
+    {
+      "epoch": 2.28,
+      "grad_norm": 0.2758764624595642,
+      "learning_rate": 7.740112994350283e-05,
+      "loss": 0.1437,
+      "step": 15410
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.648673951625824,
+      "learning_rate": 7.738626226583407e-05,
+      "loss": 0.1365,
+      "step": 15420
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 1.0597361326217651,
+      "learning_rate": 7.737139458816534e-05,
+      "loss": 0.1408,
+      "step": 15430
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.3122524321079254,
+      "learning_rate": 7.735652691049658e-05,
+      "loss": 0.1481,
+      "step": 15440
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.9538226127624512,
+      "learning_rate": 7.734165923282784e-05,
+      "loss": 0.149,
+      "step": 15450
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.28215184807777405,
+      "learning_rate": 7.732679155515909e-05,
+      "loss": 0.1316,
+      "step": 15460
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.4941287338733673,
+      "learning_rate": 7.731192387749034e-05,
+      "loss": 0.1409,
+      "step": 15470
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.3369395136833191,
+      "learning_rate": 7.72970561998216e-05,
+      "loss": 0.145,
+      "step": 15480
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.8342441320419312,
+      "learning_rate": 7.728218852215284e-05,
+      "loss": 0.1377,
+      "step": 15490
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 2.002681255340576,
+      "learning_rate": 7.72673208444841e-05,
+      "loss": 0.1436,
+      "step": 15500
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 1.1296112537384033,
+      "learning_rate": 7.725245316681534e-05,
+      "loss": 0.1372,
+      "step": 15510
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.4747876822948456,
+      "learning_rate": 7.72375854891466e-05,
+      "loss": 0.1391,
+      "step": 15520
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.37686342000961304,
+      "learning_rate": 7.722271781147785e-05,
+      "loss": 0.1417,
+      "step": 15530
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.6419855952262878,
+      "learning_rate": 7.72078501338091e-05,
+      "loss": 0.1507,
+      "step": 15540
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.5197919011116028,
+      "learning_rate": 7.719298245614036e-05,
+      "loss": 0.1406,
+      "step": 15550
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.6364248991012573,
+      "learning_rate": 7.717811477847161e-05,
+      "loss": 0.139,
+      "step": 15560
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.2479657530784607,
+      "learning_rate": 7.716324710080287e-05,
+      "loss": 0.1347,
+      "step": 15570
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.46477001905441284,
+      "learning_rate": 7.71483794231341e-05,
+      "loss": 0.1431,
+      "step": 15580
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.32222843170166016,
+      "learning_rate": 7.713351174546536e-05,
+      "loss": 0.1487,
+      "step": 15590
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.847084641456604,
+      "learning_rate": 7.711864406779662e-05,
+      "loss": 0.1503,
+      "step": 15600
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.4260410964488983,
+      "learning_rate": 7.710377639012787e-05,
+      "loss": 0.1367,
+      "step": 15610
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.7543812394142151,
+      "learning_rate": 7.708890871245911e-05,
+      "loss": 0.1353,
+      "step": 15620
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.0072932243347168,
+      "learning_rate": 7.707404103479037e-05,
+      "loss": 0.1471,
+      "step": 15630
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.30913281440734863,
+      "learning_rate": 7.705917335712162e-05,
+      "loss": 0.145,
+      "step": 15640
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.6405145525932312,
+      "learning_rate": 7.704430567945288e-05,
+      "loss": 0.1342,
+      "step": 15650
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 1.67426598072052,
+      "learning_rate": 7.702943800178413e-05,
+      "loss": 0.1342,
+      "step": 15660
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.5174002647399902,
+      "learning_rate": 7.701457032411537e-05,
+      "loss": 0.1419,
+      "step": 15670
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.7070042490959167,
+      "learning_rate": 7.699970264644663e-05,
+      "loss": 0.1388,
+      "step": 15680
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.992650032043457,
+      "learning_rate": 7.698483496877788e-05,
+      "loss": 0.1361,
+      "step": 15690
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.519932210445404,
+      "learning_rate": 7.696996729110914e-05,
+      "loss": 0.1409,
+      "step": 15700
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.5189968943595886,
+      "learning_rate": 7.695509961344038e-05,
+      "loss": 0.14,
+      "step": 15710
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 1.0756185054779053,
+      "learning_rate": 7.694023193577163e-05,
+      "loss": 0.1407,
+      "step": 15720
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.6444623470306396,
+      "learning_rate": 7.692536425810289e-05,
+      "loss": 0.1466,
+      "step": 15730
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.3298320174217224,
+      "learning_rate": 7.691049658043414e-05,
+      "loss": 0.136,
+      "step": 15740
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.20186583697795868,
+      "learning_rate": 7.68956289027654e-05,
+      "loss": 0.1393,
+      "step": 15750
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.335825115442276,
+      "learning_rate": 7.688076122509664e-05,
+      "loss": 0.1377,
+      "step": 15760
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.36127769947052,
+      "learning_rate": 7.686589354742789e-05,
+      "loss": 0.1457,
+      "step": 15770
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.2447770833969116,
+      "learning_rate": 7.685102586975915e-05,
+      "loss": 0.144,
+      "step": 15780
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.7446007132530212,
+      "learning_rate": 7.68361581920904e-05,
+      "loss": 0.1407,
+      "step": 15790
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 1.47913658618927,
+      "learning_rate": 7.682129051442166e-05,
+      "loss": 0.1473,
+      "step": 15800
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.7737151980400085,
+      "learning_rate": 7.68064228367529e-05,
+      "loss": 0.138,
+      "step": 15810
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.3995475172996521,
+      "learning_rate": 7.679155515908415e-05,
+      "loss": 0.1473,
+      "step": 15820
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.6491892337799072,
+      "learning_rate": 7.677668748141541e-05,
+      "loss": 0.1408,
+      "step": 15830
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.3302996754646301,
+      "learning_rate": 7.676181980374666e-05,
+      "loss": 0.1456,
+      "step": 15840
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 1.0726025104522705,
+      "learning_rate": 7.67469521260779e-05,
+      "loss": 0.1448,
+      "step": 15850
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.29980069398880005,
+      "learning_rate": 7.673208444840916e-05,
+      "loss": 0.1394,
+      "step": 15860
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.3691909611225128,
+      "learning_rate": 7.671721677074041e-05,
+      "loss": 0.139,
+      "step": 15870
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.26005685329437256,
+      "learning_rate": 7.670234909307167e-05,
+      "loss": 0.1369,
+      "step": 15880
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.5521954894065857,
+      "learning_rate": 7.668748141540292e-05,
+      "loss": 0.1394,
+      "step": 15890
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.44873157143592834,
+      "learning_rate": 7.667261373773416e-05,
+      "loss": 0.1408,
+      "step": 15900
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.8657715916633606,
+      "learning_rate": 7.665774606006542e-05,
+      "loss": 0.1368,
+      "step": 15910
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.876879870891571,
+      "learning_rate": 7.664287838239667e-05,
+      "loss": 0.1395,
+      "step": 15920
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.5348753929138184,
+      "learning_rate": 7.662801070472793e-05,
+      "loss": 0.147,
+      "step": 15930
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 1.3844401836395264,
+      "learning_rate": 7.661314302705917e-05,
+      "loss": 0.1448,
+      "step": 15940
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.532243013381958,
+      "learning_rate": 7.659827534939042e-05,
+      "loss": 0.1453,
+      "step": 15950
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.49074360728263855,
+      "learning_rate": 7.658340767172168e-05,
+      "loss": 0.1393,
+      "step": 15960
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.7487906217575073,
+      "learning_rate": 7.656853999405293e-05,
+      "loss": 0.1369,
+      "step": 15970
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.24927030503749847,
+      "learning_rate": 7.655367231638419e-05,
+      "loss": 0.146,
+      "step": 15980
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.3706786036491394,
+      "learning_rate": 7.653880463871543e-05,
+      "loss": 0.1432,
+      "step": 15990
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.342334508895874,
+      "learning_rate": 7.652393696104668e-05,
+      "loss": 0.1466,
+      "step": 16000
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.24996308982372284,
+      "learning_rate": 7.650906928337794e-05,
+      "loss": 0.1383,
+      "step": 16010
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.39586570858955383,
+      "learning_rate": 7.64942016057092e-05,
+      "loss": 0.1392,
+      "step": 16020
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.6906977891921997,
+      "learning_rate": 7.647933392804044e-05,
+      "loss": 0.1363,
+      "step": 16030
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 1.0736937522888184,
+      "learning_rate": 7.646446625037169e-05,
+      "loss": 0.1384,
+      "step": 16040
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.8316856622695923,
+      "learning_rate": 7.644959857270295e-05,
+      "loss": 0.1358,
+      "step": 16050
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.44582003355026245,
+      "learning_rate": 7.64347308950342e-05,
+      "loss": 0.1439,
+      "step": 16060
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.3568483591079712,
+      "learning_rate": 7.641986321736545e-05,
+      "loss": 0.1418,
+      "step": 16070
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.5303836464881897,
+      "learning_rate": 7.64049955396967e-05,
+      "loss": 0.1381,
+      "step": 16080
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.48559141159057617,
+      "learning_rate": 7.639012786202796e-05,
+      "loss": 0.1312,
+      "step": 16090
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.39514291286468506,
+      "learning_rate": 7.63752601843592e-05,
+      "loss": 0.1452,
+      "step": 16100
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.7182896137237549,
+      "learning_rate": 7.636039250669046e-05,
+      "loss": 0.133,
+      "step": 16110
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 1.0193687677383423,
+      "learning_rate": 7.634552482902171e-05,
+      "loss": 0.1429,
+      "step": 16120
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.29207757115364075,
+      "learning_rate": 7.633065715135296e-05,
+      "loss": 0.1457,
+      "step": 16130
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.8084988594055176,
+      "learning_rate": 7.631578947368422e-05,
+      "loss": 0.1446,
+      "step": 16140
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 0.3876637816429138,
+      "learning_rate": 7.630092179601547e-05,
+      "loss": 0.1339,
+      "step": 16150
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.8244117498397827,
+      "learning_rate": 7.628605411834672e-05,
+      "loss": 0.1456,
+      "step": 16160
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.2684149444103241,
+      "learning_rate": 7.627118644067796e-05,
+      "loss": 0.1347,
+      "step": 16170
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.710655927658081,
+      "learning_rate": 7.625631876300923e-05,
+      "loss": 0.1438,
+      "step": 16180
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 1.0059056282043457,
+      "learning_rate": 7.624145108534047e-05,
+      "loss": 0.1427,
+      "step": 16190
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.8452532887458801,
+      "learning_rate": 7.622658340767173e-05,
+      "loss": 0.1484,
+      "step": 16200
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.28337419033050537,
+      "learning_rate": 7.621171573000298e-05,
+      "loss": 0.1421,
+      "step": 16210
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.3758450448513031,
+      "learning_rate": 7.619684805233422e-05,
+      "loss": 0.1385,
+      "step": 16220
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.6121477484703064,
+      "learning_rate": 7.618198037466549e-05,
+      "loss": 0.1454,
+      "step": 16230
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.5063532590866089,
+      "learning_rate": 7.616711269699673e-05,
+      "loss": 0.1422,
+      "step": 16240
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 1.5977672338485718,
+      "learning_rate": 7.615224501932799e-05,
+      "loss": 0.1472,
+      "step": 16250
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.38274484872817993,
+      "learning_rate": 7.613737734165923e-05,
+      "loss": 0.1479,
+      "step": 16260
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.6593810319900513,
+      "learning_rate": 7.61225096639905e-05,
+      "loss": 0.1436,
+      "step": 16270
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.3354584276676178,
+      "learning_rate": 7.610764198632174e-05,
+      "loss": 0.1341,
+      "step": 16280
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.642591655254364,
+      "learning_rate": 7.609277430865299e-05,
+      "loss": 0.1391,
+      "step": 16290
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.25391778349876404,
+      "learning_rate": 7.607790663098425e-05,
+      "loss": 0.1409,
+      "step": 16300
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.478242427110672,
+      "learning_rate": 7.606303895331549e-05,
+      "loss": 0.1324,
+      "step": 16310
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 1.1422983407974243,
+      "learning_rate": 7.604817127564676e-05,
+      "loss": 0.1378,
+      "step": 16320
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.5172997713088989,
+      "learning_rate": 7.6033303597978e-05,
+      "loss": 0.142,
+      "step": 16330
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.5715184807777405,
+      "learning_rate": 7.601843592030925e-05,
+      "loss": 0.1425,
+      "step": 16340
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.6750907897949219,
+      "learning_rate": 7.60035682426405e-05,
+      "loss": 0.1496,
+      "step": 16350
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.7555719017982483,
+      "learning_rate": 7.598870056497176e-05,
+      "loss": 0.1384,
+      "step": 16360
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.9847421050071716,
+      "learning_rate": 7.5973832887303e-05,
+      "loss": 0.1445,
+      "step": 16370
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 2.2487268447875977,
+      "learning_rate": 7.595896520963426e-05,
+      "loss": 0.1437,
+      "step": 16380
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.2849732041358948,
+      "learning_rate": 7.594409753196551e-05,
+      "loss": 0.1373,
+      "step": 16390
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.9714717268943787,
+      "learning_rate": 7.592922985429675e-05,
+      "loss": 0.1395,
+      "step": 16400
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.6926866173744202,
+      "learning_rate": 7.591436217662802e-05,
+      "loss": 0.1417,
+      "step": 16410
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.687318742275238,
+      "learning_rate": 7.589949449895926e-05,
+      "loss": 0.1339,
+      "step": 16420
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.5382028222084045,
+      "learning_rate": 7.588462682129052e-05,
+      "loss": 0.1466,
+      "step": 16430
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.7722700238227844,
+      "learning_rate": 7.586975914362177e-05,
+      "loss": 0.1394,
+      "step": 16440
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.8510586619377136,
+      "learning_rate": 7.585489146595303e-05,
+      "loss": 0.1389,
+      "step": 16450
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.35136356949806213,
+      "learning_rate": 7.584002378828428e-05,
+      "loss": 0.1414,
+      "step": 16460
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.2880131006240845,
+      "learning_rate": 7.582515611061552e-05,
+      "loss": 0.1411,
+      "step": 16470
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.2339642494916916,
+      "learning_rate": 7.581028843294678e-05,
+      "loss": 0.14,
+      "step": 16480
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.9338073134422302,
+      "learning_rate": 7.579542075527802e-05,
+      "loss": 0.1414,
+      "step": 16490
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.46103188395500183,
+      "learning_rate": 7.578055307760929e-05,
+      "loss": 0.1396,
+      "step": 16500
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.4752241373062134,
+      "learning_rate": 7.576568539994053e-05,
+      "loss": 0.1455,
+      "step": 16510
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.4363197684288025,
+      "learning_rate": 7.575081772227178e-05,
+      "loss": 0.1391,
+      "step": 16520
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.7979761362075806,
+      "learning_rate": 7.573595004460304e-05,
+      "loss": 0.1379,
+      "step": 16530
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.3751250207424164,
+      "learning_rate": 7.572108236693429e-05,
+      "loss": 0.1444,
+      "step": 16540
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.4037092328071594,
+      "learning_rate": 7.570621468926555e-05,
+      "loss": 0.1438,
+      "step": 16550
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 0.4799390137195587,
+      "learning_rate": 7.569134701159679e-05,
+      "loss": 0.1403,
+      "step": 16560
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.8113133311271667,
+      "learning_rate": 7.567647933392804e-05,
+      "loss": 0.1372,
+      "step": 16570
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 1.4397985935211182,
+      "learning_rate": 7.56616116562593e-05,
+      "loss": 0.1379,
+      "step": 16580
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3244461417198181,
+      "learning_rate": 7.564674397859055e-05,
+      "loss": 0.14,
+      "step": 16590
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3702487647533417,
+      "learning_rate": 7.56318763009218e-05,
+      "loss": 0.1388,
+      "step": 16600
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.5789926648139954,
+      "learning_rate": 7.561700862325305e-05,
+      "loss": 0.1455,
+      "step": 16610
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.3471226096153259,
+      "learning_rate": 7.56021409455843e-05,
+      "loss": 0.1339,
+      "step": 16620
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.26994770765304565,
+      "learning_rate": 7.558727326791556e-05,
+      "loss": 0.1378,
+      "step": 16630
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.5144693851470947,
+      "learning_rate": 7.557240559024681e-05,
+      "loss": 0.1468,
+      "step": 16640
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.29384931921958923,
+      "learning_rate": 7.555753791257805e-05,
+      "loss": 0.1399,
+      "step": 16650
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.8755843639373779,
+      "learning_rate": 7.554267023490931e-05,
+      "loss": 0.1365,
+      "step": 16660
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.43195444345474243,
+      "learning_rate": 7.552780255724056e-05,
+      "loss": 0.1496,
+      "step": 16670
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.4010591506958008,
+      "learning_rate": 7.551293487957182e-05,
+      "loss": 0.1387,
+      "step": 16680
+    },
+    {
+      "epoch": 2.47,
+      "grad_norm": 0.33434391021728516,
+      "learning_rate": 7.549806720190306e-05,
+      "loss": 0.1422,
+      "step": 16690
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.1877391338348389,
+      "learning_rate": 7.548319952423432e-05,
+      "loss": 0.1342,
+      "step": 16700
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.7518399357795715,
+      "learning_rate": 7.546833184656557e-05,
+      "loss": 0.1479,
+      "step": 16710
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.5379367470741272,
+      "learning_rate": 7.545346416889682e-05,
+      "loss": 0.1384,
+      "step": 16720
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.8337181210517883,
+      "learning_rate": 7.543859649122808e-05,
+      "loss": 0.1445,
+      "step": 16730
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.9827070832252502,
+      "learning_rate": 7.542372881355932e-05,
+      "loss": 0.1377,
+      "step": 16740
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.5948978066444397,
+      "learning_rate": 7.540886113589058e-05,
+      "loss": 0.1458,
+      "step": 16750
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 1.3337939977645874,
+      "learning_rate": 7.539399345822183e-05,
+      "loss": 0.1422,
+      "step": 16760
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.48446279764175415,
+      "learning_rate": 7.537912578055309e-05,
+      "loss": 0.1393,
+      "step": 16770
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.37337374687194824,
+      "learning_rate": 7.536425810288434e-05,
+      "loss": 0.1383,
+      "step": 16780
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.37720489501953125,
+      "learning_rate": 7.534939042521558e-05,
+      "loss": 0.1323,
+      "step": 16790
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.3711811304092407,
+      "learning_rate": 7.533452274754684e-05,
+      "loss": 0.1362,
+      "step": 16800
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.36137768626213074,
+      "learning_rate": 7.531965506987809e-05,
+      "loss": 0.1311,
+      "step": 16810
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.5882296562194824,
+      "learning_rate": 7.530478739220935e-05,
+      "loss": 0.1407,
+      "step": 16820
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.5880404114723206,
+      "learning_rate": 7.528991971454059e-05,
+      "loss": 0.1402,
+      "step": 16830
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.5515339374542236,
+      "learning_rate": 7.527505203687184e-05,
+      "loss": 0.1407,
+      "step": 16840
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.28839829564094543,
+      "learning_rate": 7.52601843592031e-05,
+      "loss": 0.1333,
+      "step": 16850
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.4314468502998352,
+      "learning_rate": 7.524531668153435e-05,
+      "loss": 0.1424,
+      "step": 16860
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.4594912827014923,
+      "learning_rate": 7.52304490038656e-05,
+      "loss": 0.1435,
+      "step": 16870
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.3527372479438782,
+      "learning_rate": 7.521558132619685e-05,
+      "loss": 0.139,
+      "step": 16880
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 0.6145148277282715,
+      "learning_rate": 7.52007136485281e-05,
+      "loss": 0.1386,
+      "step": 16890
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.47387057542800903,
+      "learning_rate": 7.518584597085936e-05,
+      "loss": 0.1448,
+      "step": 16900
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.3137947916984558,
+      "learning_rate": 7.517097829319061e-05,
+      "loss": 0.1415,
+      "step": 16910
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.25534674525260925,
+      "learning_rate": 7.515611061552185e-05,
+      "loss": 0.1413,
+      "step": 16920
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.8711992502212524,
+      "learning_rate": 7.514124293785311e-05,
+      "loss": 0.141,
+      "step": 16930
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.5922618508338928,
+      "learning_rate": 7.512637526018436e-05,
+      "loss": 0.1372,
+      "step": 16940
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.2968009114265442,
+      "learning_rate": 7.511150758251562e-05,
+      "loss": 0.1443,
+      "step": 16950
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.7671728730201721,
+      "learning_rate": 7.509663990484687e-05,
+      "loss": 0.1426,
+      "step": 16960
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.7778077721595764,
+      "learning_rate": 7.508177222717811e-05,
+      "loss": 0.1391,
+      "step": 16970
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.5975720882415771,
+      "learning_rate": 7.506690454950937e-05,
+      "loss": 0.1399,
+      "step": 16980
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.3030705153942108,
+      "learning_rate": 7.505203687184062e-05,
+      "loss": 0.1411,
+      "step": 16990
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.6025968790054321,
+      "learning_rate": 7.503716919417188e-05,
+      "loss": 0.1429,
+      "step": 17000
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.262866735458374,
+      "learning_rate": 7.502230151650312e-05,
+      "loss": 0.1359,
+      "step": 17010
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 1.0509122610092163,
+      "learning_rate": 7.500743383883437e-05,
+      "loss": 0.1488,
+      "step": 17020
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.8183063268661499,
+      "learning_rate": 7.499256616116563e-05,
+      "loss": 0.1436,
+      "step": 17030
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.247868612408638,
+      "learning_rate": 7.497769848349688e-05,
+      "loss": 0.1381,
+      "step": 17040
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.292298287153244,
+      "learning_rate": 7.496283080582814e-05,
+      "loss": 0.145,
+      "step": 17050
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.6261994242668152,
+      "learning_rate": 7.494796312815938e-05,
+      "loss": 0.1494,
+      "step": 17060
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.6475508213043213,
+      "learning_rate": 7.493309545049065e-05,
+      "loss": 0.1359,
+      "step": 17070
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.24762730300426483,
+      "learning_rate": 7.491822777282189e-05,
+      "loss": 0.1477,
+      "step": 17080
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 1.116014003753662,
+      "learning_rate": 7.490336009515314e-05,
+      "loss": 0.1436,
+      "step": 17090
+    },
+    {
+      "epoch": 2.53,
+      "grad_norm": 0.24743910133838654,
+      "learning_rate": 7.488849241748438e-05,
+      "loss": 0.1452,
+      "step": 17100
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.4202406406402588,
+      "learning_rate": 7.487362473981564e-05,
+      "loss": 0.14,
+      "step": 17110
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.25671133399009705,
+      "learning_rate": 7.485875706214691e-05,
+      "loss": 0.1421,
+      "step": 17120
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 1.6126033067703247,
+      "learning_rate": 7.484388938447815e-05,
+      "loss": 0.1455,
+      "step": 17130
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.7320016026496887,
+      "learning_rate": 7.48290217068094e-05,
+      "loss": 0.1455,
+      "step": 17140
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.4042165279388428,
+      "learning_rate": 7.481415402914064e-05,
+      "loss": 0.138,
+      "step": 17150
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.34952160716056824,
+      "learning_rate": 7.479928635147191e-05,
+      "loss": 0.1372,
+      "step": 17160
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.43814823031425476,
+      "learning_rate": 7.478441867380315e-05,
+      "loss": 0.1426,
+      "step": 17170
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.30086013674736023,
+      "learning_rate": 7.476955099613441e-05,
+      "loss": 0.1333,
+      "step": 17180
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.6000199317932129,
+      "learning_rate": 7.475468331846566e-05,
+      "loss": 0.1436,
+      "step": 17190
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.31554168462753296,
+      "learning_rate": 7.47398156407969e-05,
+      "loss": 0.1476,
+      "step": 17200
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.5113075971603394,
+      "learning_rate": 7.472494796312817e-05,
+      "loss": 0.1442,
+      "step": 17210
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.6701615452766418,
+      "learning_rate": 7.471008028545941e-05,
+      "loss": 0.1412,
+      "step": 17220
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 0.9265919327735901,
+      "learning_rate": 7.469521260779067e-05,
+      "loss": 0.1443,
+      "step": 17230
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.5859856605529785,
+      "learning_rate": 7.468034493012191e-05,
+      "loss": 0.1479,
+      "step": 17240
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.39752840995788574,
+      "learning_rate": 7.466547725245318e-05,
+      "loss": 0.1487,
+      "step": 17250
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.5720125436782837,
+      "learning_rate": 7.465060957478442e-05,
+      "loss": 0.1419,
+      "step": 17260
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.754892110824585,
+      "learning_rate": 7.463574189711567e-05,
+      "loss": 0.1425,
+      "step": 17270
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.7733323574066162,
+      "learning_rate": 7.462087421944693e-05,
+      "loss": 0.1453,
+      "step": 17280
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 1.4931153059005737,
+      "learning_rate": 7.460600654177817e-05,
+      "loss": 0.135,
+      "step": 17290
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.3707599639892578,
+      "learning_rate": 7.459113886410944e-05,
+      "loss": 0.1415,
+      "step": 17300
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3874984681606293,
+      "learning_rate": 7.457627118644068e-05,
+      "loss": 0.1476,
+      "step": 17310
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3033313751220703,
+      "learning_rate": 7.456140350877193e-05,
+      "loss": 0.1414,
+      "step": 17320
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.6481125354766846,
+      "learning_rate": 7.454653583110318e-05,
+      "loss": 0.1432,
+      "step": 17330
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 1.2512054443359375,
+      "learning_rate": 7.453166815343444e-05,
+      "loss": 0.1396,
+      "step": 17340
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.6148348450660706,
+      "learning_rate": 7.451680047576569e-05,
+      "loss": 0.139,
+      "step": 17350
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.3186105191707611,
+      "learning_rate": 7.450193279809694e-05,
+      "loss": 0.1282,
+      "step": 17360
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 0.31010115146636963,
+      "learning_rate": 7.44870651204282e-05,
+      "loss": 0.1444,
+      "step": 17370
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.8012278079986572,
+      "learning_rate": 7.447219744275944e-05,
+      "loss": 0.1356,
+      "step": 17380
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.26003143191337585,
+      "learning_rate": 7.44573297650907e-05,
+      "loss": 0.1387,
+      "step": 17390
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.284709095954895,
+      "learning_rate": 7.444246208742195e-05,
+      "loss": 0.1456,
+      "step": 17400
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 1.3213711977005005,
+      "learning_rate": 7.44275944097532e-05,
+      "loss": 0.1347,
+      "step": 17410
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.8473525047302246,
+      "learning_rate": 7.441272673208444e-05,
+      "loss": 0.1453,
+      "step": 17420
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.36998623609542847,
+      "learning_rate": 7.439785905441571e-05,
+      "loss": 0.1308,
+      "step": 17430
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 2.219050168991089,
+      "learning_rate": 7.438299137674695e-05,
+      "loss": 0.1497,
+      "step": 17440
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.2892897129058838,
+      "learning_rate": 7.43681236990782e-05,
+      "loss": 0.1392,
+      "step": 17450
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.3273633122444153,
+      "learning_rate": 7.435325602140946e-05,
+      "loss": 0.1459,
+      "step": 17460
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 1.172643780708313,
+      "learning_rate": 7.43383883437407e-05,
+      "loss": 0.1396,
+      "step": 17470
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.7880512475967407,
+      "learning_rate": 7.432352066607197e-05,
+      "loss": 0.1404,
+      "step": 17480
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.3355169892311096,
+      "learning_rate": 7.430865298840321e-05,
+      "loss": 0.1513,
+      "step": 17490
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 0.5652661919593811,
+      "learning_rate": 7.429378531073447e-05,
+      "loss": 0.1349,
+      "step": 17500
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.34006038308143616,
+      "learning_rate": 7.427891763306572e-05,
+      "loss": 0.1406,
+      "step": 17510
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.258184552192688,
+      "learning_rate": 7.426404995539698e-05,
+      "loss": 0.1422,
+      "step": 17520
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.49831709265708923,
+      "learning_rate": 7.424918227772823e-05,
+      "loss": 0.1373,
+      "step": 17530
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 1.0806423425674438,
+      "learning_rate": 7.423431460005947e-05,
+      "loss": 0.1468,
+      "step": 17540
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.307546466588974,
+      "learning_rate": 7.421944692239073e-05,
+      "loss": 0.1376,
+      "step": 17550
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.5520646572113037,
+      "learning_rate": 7.420457924472198e-05,
+      "loss": 0.1426,
+      "step": 17560
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.2815137505531311,
+      "learning_rate": 7.418971156705324e-05,
+      "loss": 0.1359,
+      "step": 17570
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.4063834846019745,
+      "learning_rate": 7.417484388938448e-05,
+      "loss": 0.1448,
+      "step": 17580
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.4033089876174927,
+      "learning_rate": 7.415997621171573e-05,
+      "loss": 0.1406,
+      "step": 17590
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.300946980714798,
+      "learning_rate": 7.414510853404699e-05,
+      "loss": 0.1339,
+      "step": 17600
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.9063736200332642,
+      "learning_rate": 7.413024085637824e-05,
+      "loss": 0.146,
+      "step": 17610
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 1.2720870971679688,
+      "learning_rate": 7.41153731787095e-05,
+      "loss": 0.1394,
+      "step": 17620
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.36671578884124756,
+      "learning_rate": 7.410050550104074e-05,
+      "loss": 0.1451,
+      "step": 17630
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.28279754519462585,
+      "learning_rate": 7.408563782337199e-05,
+      "loss": 0.1421,
+      "step": 17640
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 1.5014578104019165,
+      "learning_rate": 7.407077014570325e-05,
+      "loss": 0.1362,
+      "step": 17650
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.6834672689437866,
+      "learning_rate": 7.40559024680345e-05,
+      "loss": 0.1504,
+      "step": 17660
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.5922780632972717,
+      "learning_rate": 7.404103479036574e-05,
+      "loss": 0.1375,
+      "step": 17670
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.2400318682193756,
+      "learning_rate": 7.4026167112697e-05,
+      "loss": 0.1401,
+      "step": 17680
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.8875470161437988,
+      "learning_rate": 7.401129943502825e-05,
+      "loss": 0.1465,
+      "step": 17690
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.2918970584869385,
+      "learning_rate": 7.399643175735951e-05,
+      "loss": 0.141,
+      "step": 17700
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.6647611260414124,
+      "learning_rate": 7.398156407969076e-05,
+      "loss": 0.1399,
+      "step": 17710
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.23712264001369476,
+      "learning_rate": 7.3966696402022e-05,
+      "loss": 0.1331,
+      "step": 17720
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.7306144833564758,
+      "learning_rate": 7.395182872435326e-05,
+      "loss": 0.1417,
+      "step": 17730
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.40289196372032166,
+      "learning_rate": 7.393696104668451e-05,
+      "loss": 0.1437,
+      "step": 17740
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.719495415687561,
+      "learning_rate": 7.392209336901577e-05,
+      "loss": 0.1472,
+      "step": 17750
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.24560745060443878,
+      "learning_rate": 7.390722569134701e-05,
+      "loss": 0.1434,
+      "step": 17760
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 1.0618599653244019,
+      "learning_rate": 7.389235801367826e-05,
+      "loss": 0.1481,
+      "step": 17770
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.5430271029472351,
+      "learning_rate": 7.387749033600952e-05,
+      "loss": 0.139,
+      "step": 17780
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.0089772939682007,
+      "learning_rate": 7.386262265834077e-05,
+      "loss": 0.1429,
+      "step": 17790
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.2407023161649704,
+      "learning_rate": 7.384775498067203e-05,
+      "loss": 0.1383,
+      "step": 17800
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.26621460914611816,
+      "learning_rate": 7.383288730300327e-05,
+      "loss": 0.139,
+      "step": 17810
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.26956406235694885,
+      "learning_rate": 7.381801962533452e-05,
+      "loss": 0.1417,
+      "step": 17820
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.2954663038253784,
+      "learning_rate": 7.380315194766578e-05,
+      "loss": 0.1337,
+      "step": 17830
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 1.1632752418518066,
+      "learning_rate": 7.378828426999703e-05,
+      "loss": 0.1466,
+      "step": 17840
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 1.1635485887527466,
+      "learning_rate": 7.377341659232829e-05,
+      "loss": 0.1374,
+      "step": 17850
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.622186541557312,
+      "learning_rate": 7.375854891465953e-05,
+      "loss": 0.1385,
+      "step": 17860
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.2939455211162567,
+      "learning_rate": 7.374368123699078e-05,
+      "loss": 0.137,
+      "step": 17870
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.38761767745018005,
+      "learning_rate": 7.372881355932204e-05,
+      "loss": 0.1361,
+      "step": 17880
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.9036255478858948,
+      "learning_rate": 7.37139458816533e-05,
+      "loss": 0.1374,
+      "step": 17890
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.34090399742126465,
+      "learning_rate": 7.369907820398453e-05,
+      "loss": 0.1325,
+      "step": 17900
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 0.29713836312294006,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.1408,
+      "step": 17910
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.826836884021759,
+      "learning_rate": 7.366934284864704e-05,
+      "loss": 0.1498,
+      "step": 17920
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.8794457316398621,
+      "learning_rate": 7.36544751709783e-05,
+      "loss": 0.1441,
+      "step": 17930
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.3596440851688385,
+      "learning_rate": 7.363960749330955e-05,
+      "loss": 0.1452,
+      "step": 17940
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.4660389721393585,
+      "learning_rate": 7.36247398156408e-05,
+      "loss": 0.1397,
+      "step": 17950
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 1.148728847503662,
+      "learning_rate": 7.360987213797205e-05,
+      "loss": 0.1429,
+      "step": 17960
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.5068449378013611,
+      "learning_rate": 7.35950044603033e-05,
+      "loss": 0.1367,
+      "step": 17970
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.3436690866947174,
+      "learning_rate": 7.358013678263456e-05,
+      "loss": 0.1434,
+      "step": 17980
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.28847482800483704,
+      "learning_rate": 7.35652691049658e-05,
+      "loss": 0.1452,
+      "step": 17990
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.3127191960811615,
+      "learning_rate": 7.355040142729706e-05,
+      "loss": 0.1468,
+      "step": 18000
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.3851916491985321,
+      "learning_rate": 7.353553374962831e-05,
+      "loss": 0.1416,
+      "step": 18010
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 1.0425105094909668,
+      "learning_rate": 7.352066607195956e-05,
+      "loss": 0.1336,
+      "step": 18020
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.5313891172409058,
+      "learning_rate": 7.350579839429082e-05,
+      "loss": 0.1449,
+      "step": 18030
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.3654390871524811,
+      "learning_rate": 7.349093071662206e-05,
+      "loss": 0.1304,
+      "step": 18040
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.8208723664283752,
+      "learning_rate": 7.347606303895333e-05,
+      "loss": 0.1395,
+      "step": 18050
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.265028178691864,
+      "learning_rate": 7.346119536128457e-05,
+      "loss": 0.1423,
+      "step": 18060
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.5879223346710205,
+      "learning_rate": 7.344632768361583e-05,
+      "loss": 0.1346,
+      "step": 18070
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 1.9254904985427856,
+      "learning_rate": 7.343146000594707e-05,
+      "loss": 0.1455,
+      "step": 18080
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.7956934571266174,
+      "learning_rate": 7.341659232827832e-05,
+      "loss": 0.1395,
+      "step": 18090
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.3552607595920563,
+      "learning_rate": 7.340172465060958e-05,
+      "loss": 0.1381,
+      "step": 18100
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.8107473254203796,
+      "learning_rate": 7.338685697294083e-05,
+      "loss": 0.1417,
+      "step": 18110
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.28250497579574585,
+      "learning_rate": 7.337198929527209e-05,
+      "loss": 0.144,
+      "step": 18120
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.26782554388046265,
+      "learning_rate": 7.335712161760333e-05,
+      "loss": 0.1386,
+      "step": 18130
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.5947808623313904,
+      "learning_rate": 7.33422539399346e-05,
+      "loss": 0.1424,
+      "step": 18140
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.2512226998806,
+      "learning_rate": 7.332738626226584e-05,
+      "loss": 0.1352,
+      "step": 18150
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.35504305362701416,
+      "learning_rate": 7.331251858459709e-05,
+      "loss": 0.1416,
+      "step": 18160
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 1.3061926364898682,
+      "learning_rate": 7.329765090692835e-05,
+      "loss": 0.1444,
+      "step": 18170
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.6368234157562256,
+      "learning_rate": 7.328278322925959e-05,
+      "loss": 0.1403,
+      "step": 18180
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.42611798644065857,
+      "learning_rate": 7.326791555159086e-05,
+      "loss": 0.1386,
+      "step": 18190
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.6984022855758667,
+      "learning_rate": 7.32530478739221e-05,
+      "loss": 0.1451,
+      "step": 18200
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.4926019310951233,
+      "learning_rate": 7.323818019625335e-05,
+      "loss": 0.1399,
+      "step": 18210
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.8579056262969971,
+      "learning_rate": 7.322331251858459e-05,
+      "loss": 0.143,
+      "step": 18220
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.9211030602455139,
+      "learning_rate": 7.320844484091586e-05,
+      "loss": 0.1469,
+      "step": 18230
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.8623634576797485,
+      "learning_rate": 7.31935771632471e-05,
+      "loss": 0.1424,
+      "step": 18240
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.2992829382419586,
+      "learning_rate": 7.317870948557836e-05,
+      "loss": 0.1417,
+      "step": 18250
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.6315504908561707,
+      "learning_rate": 7.316384180790961e-05,
+      "loss": 0.1417,
+      "step": 18260
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.9249059557914734,
+      "learning_rate": 7.314897413024085e-05,
+      "loss": 0.1468,
+      "step": 18270
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.35125473141670227,
+      "learning_rate": 7.313410645257212e-05,
+      "loss": 0.145,
+      "step": 18280
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.4258287250995636,
+      "learning_rate": 7.311923877490336e-05,
+      "loss": 0.1401,
+      "step": 18290
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.26481908559799194,
+      "learning_rate": 7.310437109723462e-05,
+      "loss": 0.1431,
+      "step": 18300
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 1.5498566627502441,
+      "learning_rate": 7.308950341956586e-05,
+      "loss": 0.1455,
+      "step": 18310
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.0560815334320068,
+      "learning_rate": 7.307463574189713e-05,
+      "loss": 0.1475,
+      "step": 18320
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.7546606659889221,
+      "learning_rate": 7.305976806422837e-05,
+      "loss": 0.1465,
+      "step": 18330
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.151910424232483,
+      "learning_rate": 7.304490038655962e-05,
+      "loss": 0.1435,
+      "step": 18340
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.38248512148857117,
+      "learning_rate": 7.303003270889088e-05,
+      "loss": 0.1462,
+      "step": 18350
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.24999620020389557,
+      "learning_rate": 7.301516503122212e-05,
+      "loss": 0.1423,
+      "step": 18360
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 1.2096368074417114,
+      "learning_rate": 7.300029735355339e-05,
+      "loss": 0.1402,
+      "step": 18370
+    },
+    {
+      "epoch": 2.72,
+      "grad_norm": 0.6669314503669739,
+      "learning_rate": 7.298542967588463e-05,
+      "loss": 0.1419,
+      "step": 18380
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.7453274726867676,
+      "learning_rate": 7.297056199821588e-05,
+      "loss": 0.1507,
+      "step": 18390
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.3134549558162689,
+      "learning_rate": 7.295569432054712e-05,
+      "loss": 0.1401,
+      "step": 18400
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.509635329246521,
+      "learning_rate": 7.294082664287839e-05,
+      "loss": 0.1427,
+      "step": 18410
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 1.5291324853897095,
+      "learning_rate": 7.292595896520963e-05,
+      "loss": 0.1382,
+      "step": 18420
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.8732107877731323,
+      "learning_rate": 7.291109128754089e-05,
+      "loss": 0.1425,
+      "step": 18430
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.951556384563446,
+      "learning_rate": 7.289622360987214e-05,
+      "loss": 0.1417,
+      "step": 18440
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 0.22064682841300964,
+      "learning_rate": 7.288135593220338e-05,
+      "loss": 0.1389,
+      "step": 18450
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.36942338943481445,
+      "learning_rate": 7.286648825453465e-05,
+      "loss": 0.1356,
+      "step": 18460
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.9014129638671875,
+      "learning_rate": 7.28516205768659e-05,
+      "loss": 0.1439,
+      "step": 18470
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.31892070174217224,
+      "learning_rate": 7.283675289919715e-05,
+      "loss": 0.1439,
+      "step": 18480
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.3054444491863251,
+      "learning_rate": 7.28218852215284e-05,
+      "loss": 0.1385,
+      "step": 18490
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.23387959599494934,
+      "learning_rate": 7.280701754385966e-05,
+      "loss": 0.1406,
+      "step": 18500
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.26444515585899353,
+      "learning_rate": 7.279214986619091e-05,
+      "loss": 0.1412,
+      "step": 18510
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.8463549613952637,
+      "learning_rate": 7.277728218852215e-05,
+      "loss": 0.1437,
+      "step": 18520
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.9371518492698669,
+      "learning_rate": 7.276241451085341e-05,
+      "loss": 0.1401,
+      "step": 18530
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.3569345772266388,
+      "learning_rate": 7.274754683318466e-05,
+      "loss": 0.132,
+      "step": 18540
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.25233766436576843,
+      "learning_rate": 7.273267915551592e-05,
+      "loss": 0.1409,
+      "step": 18550
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.46643221378326416,
+      "learning_rate": 7.271781147784716e-05,
+      "loss": 0.1378,
+      "step": 18560
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 1.249118447303772,
+      "learning_rate": 7.270294380017841e-05,
+      "loss": 0.1477,
+      "step": 18570
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.29594314098358154,
+      "learning_rate": 7.268807612250967e-05,
+      "loss": 0.15,
+      "step": 18580
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.2833545207977295,
+      "learning_rate": 7.267320844484092e-05,
+      "loss": 0.1405,
+      "step": 18590
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.5924676656723022,
+      "learning_rate": 7.265834076717218e-05,
+      "loss": 0.1372,
+      "step": 18600
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.3902270495891571,
+      "learning_rate": 7.264347308950342e-05,
+      "loss": 0.1413,
+      "step": 18610
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.447142630815506,
+      "learning_rate": 7.262860541183467e-05,
+      "loss": 0.1426,
+      "step": 18620
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.4772433340549469,
+      "learning_rate": 7.261373773416593e-05,
+      "loss": 0.1394,
+      "step": 18630
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.45227327942848206,
+      "learning_rate": 7.259887005649718e-05,
+      "loss": 0.1315,
+      "step": 18640
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 1.1519386768341064,
+      "learning_rate": 7.258400237882843e-05,
+      "loss": 0.1408,
+      "step": 18650
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.8935173749923706,
+      "learning_rate": 7.256913470115968e-05,
+      "loss": 0.1429,
+      "step": 18660
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.24923725426197052,
+      "learning_rate": 7.255426702349093e-05,
+      "loss": 0.132,
+      "step": 18670
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.4108467102050781,
+      "learning_rate": 7.253939934582219e-05,
+      "loss": 0.1406,
+      "step": 18680
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.7500537037849426,
+      "learning_rate": 7.252453166815344e-05,
+      "loss": 0.1435,
+      "step": 18690
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.3367678225040436,
+      "learning_rate": 7.250966399048469e-05,
+      "loss": 0.1423,
+      "step": 18700
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.30210843682289124,
+      "learning_rate": 7.249479631281594e-05,
+      "loss": 0.1513,
+      "step": 18710
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.7015200257301331,
+      "learning_rate": 7.24799286351472e-05,
+      "loss": 0.14,
+      "step": 18720
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.45465758442878723,
+      "learning_rate": 7.246506095747845e-05,
+      "loss": 0.1443,
+      "step": 18730
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.9578850269317627,
+      "learning_rate": 7.245019327980969e-05,
+      "loss": 0.1461,
+      "step": 18740
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.20722438395023346,
+      "learning_rate": 7.243532560214095e-05,
+      "loss": 0.1311,
+      "step": 18750
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.9666074514389038,
+      "learning_rate": 7.24204579244722e-05,
+      "loss": 0.1319,
+      "step": 18760
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.286914199590683,
+      "learning_rate": 7.240559024680346e-05,
+      "loss": 0.1404,
+      "step": 18770
+    },
+    {
+      "epoch": 2.78,
+      "grad_norm": 0.635274350643158,
+      "learning_rate": 7.239072256913471e-05,
+      "loss": 0.1369,
+      "step": 18780
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.46256542205810547,
+      "learning_rate": 7.237585489146595e-05,
+      "loss": 0.1416,
+      "step": 18790
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.5526597499847412,
+      "learning_rate": 7.23609872137972e-05,
+      "loss": 0.14,
+      "step": 18800
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 1.3282074928283691,
+      "learning_rate": 7.234611953612846e-05,
+      "loss": 0.1523,
+      "step": 18810
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.2414388209581375,
+      "learning_rate": 7.233125185845972e-05,
+      "loss": 0.133,
+      "step": 18820
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.8080794215202332,
+      "learning_rate": 7.231638418079097e-05,
+      "loss": 0.143,
+      "step": 18830
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.4373549818992615,
+      "learning_rate": 7.230151650312221e-05,
+      "loss": 0.1352,
+      "step": 18840
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.6084916591644287,
+      "learning_rate": 7.228664882545347e-05,
+      "loss": 0.1433,
+      "step": 18850
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.41089242696762085,
+      "learning_rate": 7.227178114778472e-05,
+      "loss": 0.145,
+      "step": 18860
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.2655513882637024,
+      "learning_rate": 7.225691347011598e-05,
+      "loss": 0.1423,
+      "step": 18870
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4778943359851837,
+      "learning_rate": 7.224204579244722e-05,
+      "loss": 0.1348,
+      "step": 18880
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.4908297061920166,
+      "learning_rate": 7.222717811477847e-05,
+      "loss": 0.1434,
+      "step": 18890
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.3996427059173584,
+      "learning_rate": 7.221231043710973e-05,
+      "loss": 0.1404,
+      "step": 18900
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.6099945306777954,
+      "learning_rate": 7.219744275944098e-05,
+      "loss": 0.1454,
+      "step": 18910
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.45081639289855957,
+      "learning_rate": 7.218257508177224e-05,
+      "loss": 0.1354,
+      "step": 18920
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.0426857471466064,
+      "learning_rate": 7.216770740410348e-05,
+      "loss": 0.1491,
+      "step": 18930
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.772966206073761,
+      "learning_rate": 7.215283972643473e-05,
+      "loss": 0.1365,
+      "step": 18940
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.3265818655490875,
+      "learning_rate": 7.213797204876599e-05,
+      "loss": 0.1498,
+      "step": 18950
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.2635791003704071,
+      "learning_rate": 7.212310437109724e-05,
+      "loss": 0.1363,
+      "step": 18960
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.9569715857505798,
+      "learning_rate": 7.210823669342848e-05,
+      "loss": 0.1327,
+      "step": 18970
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.20809006690979004,
+      "learning_rate": 7.209336901575974e-05,
+      "loss": 0.1377,
+      "step": 18980
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 1.0042877197265625,
+      "learning_rate": 7.207850133809099e-05,
+      "loss": 0.1371,
+      "step": 18990
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.5755969882011414,
+      "learning_rate": 7.206363366042225e-05,
+      "loss": 0.1355,
+      "step": 19000
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.8315523862838745,
+      "learning_rate": 7.20487659827535e-05,
+      "loss": 0.1435,
+      "step": 19010
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.5126630663871765,
+      "learning_rate": 7.203389830508474e-05,
+      "loss": 0.1358,
+      "step": 19020
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.5610932111740112,
+      "learning_rate": 7.201903062741601e-05,
+      "loss": 0.1303,
+      "step": 19030
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.49236372113227844,
+      "learning_rate": 7.200416294974725e-05,
+      "loss": 0.1428,
+      "step": 19040
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.26628008484840393,
+      "learning_rate": 7.198929527207851e-05,
+      "loss": 0.1407,
+      "step": 19050
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 1.6078991889953613,
+      "learning_rate": 7.197442759440975e-05,
+      "loss": 0.135,
+      "step": 19060
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.3959672749042511,
+      "learning_rate": 7.1959559916741e-05,
+      "loss": 0.1418,
+      "step": 19070
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.35960277915000916,
+      "learning_rate": 7.194469223907226e-05,
+      "loss": 0.1386,
+      "step": 19080
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.46168479323387146,
+      "learning_rate": 7.192982456140351e-05,
+      "loss": 0.1465,
+      "step": 19090
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.7458998560905457,
+      "learning_rate": 7.191495688373477e-05,
+      "loss": 0.1347,
+      "step": 19100
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.629227876663208,
+      "learning_rate": 7.190008920606601e-05,
+      "loss": 0.1382,
+      "step": 19110
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.48990973830223083,
+      "learning_rate": 7.188522152839728e-05,
+      "loss": 0.1346,
+      "step": 19120
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 1.1364151239395142,
+      "learning_rate": 7.187035385072852e-05,
+      "loss": 0.1346,
+      "step": 19130
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.39966854453086853,
+      "learning_rate": 7.185548617305977e-05,
+      "loss": 0.1365,
+      "step": 19140
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.7130772471427917,
+      "learning_rate": 7.184061849539101e-05,
+      "loss": 0.1474,
+      "step": 19150
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.47073498368263245,
+      "learning_rate": 7.182575081772227e-05,
+      "loss": 0.136,
+      "step": 19160
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.7420998811721802,
+      "learning_rate": 7.181088314005354e-05,
+      "loss": 0.1423,
+      "step": 19170
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.5848856568336487,
+      "learning_rate": 7.179601546238478e-05,
+      "loss": 0.1435,
+      "step": 19180
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.41992321610450745,
+      "learning_rate": 7.178114778471603e-05,
+      "loss": 0.1385,
+      "step": 19190
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.2730390131473541,
+      "learning_rate": 7.176628010704727e-05,
+      "loss": 0.135,
+      "step": 19200
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.503334105014801,
+      "learning_rate": 7.175141242937854e-05,
+      "loss": 0.141,
+      "step": 19210
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.7644150257110596,
+      "learning_rate": 7.173654475170978e-05,
+      "loss": 0.1387,
+      "step": 19220
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.33800143003463745,
+      "learning_rate": 7.172167707404104e-05,
+      "loss": 0.1381,
+      "step": 19230
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.5975232124328613,
+      "learning_rate": 7.17068093963723e-05,
+      "loss": 0.1403,
+      "step": 19240
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.3202337920665741,
+      "learning_rate": 7.169194171870353e-05,
+      "loss": 0.1415,
+      "step": 19250
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.2667614221572876,
+      "learning_rate": 7.16770740410348e-05,
+      "loss": 0.1404,
+      "step": 19260
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.28970199823379517,
+      "learning_rate": 7.166220636336604e-05,
+      "loss": 0.144,
+      "step": 19270
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.9261873364448547,
+      "learning_rate": 7.16473386856973e-05,
+      "loss": 0.1355,
+      "step": 19280
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.30007144808769226,
+      "learning_rate": 7.163247100802854e-05,
+      "loss": 0.1357,
+      "step": 19290
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 1.1617704629898071,
+      "learning_rate": 7.161760333035981e-05,
+      "loss": 0.1433,
+      "step": 19300
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.5835827589035034,
+      "learning_rate": 7.160273565269105e-05,
+      "loss": 0.1412,
+      "step": 19310
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.3388794958591461,
+      "learning_rate": 7.15878679750223e-05,
+      "loss": 0.1467,
+      "step": 19320
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.5224015712738037,
+      "learning_rate": 7.157300029735356e-05,
+      "loss": 0.1409,
+      "step": 19330
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.2810278534889221,
+      "learning_rate": 7.15581326196848e-05,
+      "loss": 0.14,
+      "step": 19340
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.6086370348930359,
+      "learning_rate": 7.154326494201607e-05,
+      "loss": 0.1415,
+      "step": 19350
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.28138285875320435,
+      "learning_rate": 7.152839726434731e-05,
+      "loss": 0.1351,
+      "step": 19360
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.2224389910697937,
+      "learning_rate": 7.151352958667857e-05,
+      "loss": 0.1389,
+      "step": 19370
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.6104931235313416,
+      "learning_rate": 7.14986619090098e-05,
+      "loss": 0.144,
+      "step": 19380
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.29406848549842834,
+      "learning_rate": 7.148379423134107e-05,
+      "loss": 0.1367,
+      "step": 19390
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.9205821752548218,
+      "learning_rate": 7.146892655367232e-05,
+      "loss": 0.1391,
+      "step": 19400
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.89859539270401,
+      "learning_rate": 7.145405887600357e-05,
+      "loss": 0.1401,
+      "step": 19410
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.37001898884773254,
+      "learning_rate": 7.143919119833483e-05,
+      "loss": 0.1421,
+      "step": 19420
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.483979731798172,
+      "learning_rate": 7.142432352066607e-05,
+      "loss": 0.1388,
+      "step": 19430
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.3105422556400299,
+      "learning_rate": 7.140945584299733e-05,
+      "loss": 0.1451,
+      "step": 19440
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.598884105682373,
+      "learning_rate": 7.139458816532858e-05,
+      "loss": 0.1378,
+      "step": 19450
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.6454994082450867,
+      "learning_rate": 7.137972048765983e-05,
+      "loss": 0.1382,
+      "step": 19460
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.18754199147224426,
+      "learning_rate": 7.136485280999107e-05,
+      "loss": 0.1276,
+      "step": 19470
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.31574535369873047,
+      "learning_rate": 7.134998513232234e-05,
+      "loss": 0.1422,
+      "step": 19480
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.33168497681617737,
+      "learning_rate": 7.133511745465358e-05,
+      "loss": 0.1333,
+      "step": 19490
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.3626178503036499,
+      "learning_rate": 7.132024977698484e-05,
+      "loss": 0.1411,
+      "step": 19500
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.19792704284191132,
+      "learning_rate": 7.130538209931609e-05,
+      "loss": 0.1357,
+      "step": 19510
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.45540741086006165,
+      "learning_rate": 7.129051442164735e-05,
+      "loss": 0.1373,
+      "step": 19520
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 1.1599476337432861,
+      "learning_rate": 7.12756467439786e-05,
+      "loss": 0.1394,
+      "step": 19530
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.3340741991996765,
+      "learning_rate": 7.126077906630984e-05,
+      "loss": 0.1394,
+      "step": 19540
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.3207394480705261,
+      "learning_rate": 7.12459113886411e-05,
+      "loss": 0.1456,
+      "step": 19550
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.8679114580154419,
+      "learning_rate": 7.123104371097235e-05,
+      "loss": 0.1392,
+      "step": 19560
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 1.6166648864746094,
+      "learning_rate": 7.12161760333036e-05,
+      "loss": 0.1494,
+      "step": 19570
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.4138358533382416,
+      "learning_rate": 7.120130835563486e-05,
+      "loss": 0.1461,
+      "step": 19580
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.23959451913833618,
+      "learning_rate": 7.11864406779661e-05,
+      "loss": 0.1339,
+      "step": 19590
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.32449159026145935,
+      "learning_rate": 7.117157300029736e-05,
+      "loss": 0.1282,
+      "step": 19600
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 2.3437106609344482,
+      "learning_rate": 7.115670532262861e-05,
+      "loss": 0.1494,
+      "step": 19610
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.4861600995063782,
+      "learning_rate": 7.114183764495987e-05,
+      "loss": 0.1428,
+      "step": 19620
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.6348639726638794,
+      "learning_rate": 7.112696996729111e-05,
+      "loss": 0.1397,
+      "step": 19630
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.27723273634910583,
+      "learning_rate": 7.111210228962236e-05,
+      "loss": 0.1427,
+      "step": 19640
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.36210235953330994,
+      "learning_rate": 7.109723461195362e-05,
+      "loss": 0.1429,
+      "step": 19650
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.48523059487342834,
+      "learning_rate": 7.108236693428487e-05,
+      "loss": 0.1331,
+      "step": 19660
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.6052648425102234,
+      "learning_rate": 7.106749925661613e-05,
+      "loss": 0.1447,
+      "step": 19670
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.2304733544588089,
+      "learning_rate": 7.105263157894737e-05,
+      "loss": 0.1311,
+      "step": 19680
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.2759769558906555,
+      "learning_rate": 7.103776390127862e-05,
+      "loss": 0.1388,
+      "step": 19690
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.9866233468055725,
+      "learning_rate": 7.102289622360988e-05,
+      "loss": 0.1448,
+      "step": 19700
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.8615758419036865,
+      "learning_rate": 7.100802854594113e-05,
+      "loss": 0.1501,
+      "step": 19710
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.49602025747299194,
+      "learning_rate": 7.099316086827237e-05,
+      "loss": 0.1455,
+      "step": 19720
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.47930097579956055,
+      "learning_rate": 7.097829319060363e-05,
+      "loss": 0.1398,
+      "step": 19730
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.4859764575958252,
+      "learning_rate": 7.096342551293488e-05,
+      "loss": 0.138,
+      "step": 19740
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.7272999882698059,
+      "learning_rate": 7.094855783526614e-05,
+      "loss": 0.1323,
+      "step": 19750
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.3457830250263214,
+      "learning_rate": 7.093369015759739e-05,
+      "loss": 0.1425,
+      "step": 19760
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.23593512177467346,
+      "learning_rate": 7.091882247992863e-05,
+      "loss": 0.1293,
+      "step": 19770
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.8734978437423706,
+      "learning_rate": 7.090395480225989e-05,
+      "loss": 0.136,
+      "step": 19780
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.3335154950618744,
+      "learning_rate": 7.088908712459114e-05,
+      "loss": 0.1376,
+      "step": 19790
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.34812235832214355,
+      "learning_rate": 7.08742194469224e-05,
+      "loss": 0.1357,
+      "step": 19800
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.2229343056678772,
+      "learning_rate": 7.085935176925364e-05,
+      "loss": 0.1397,
+      "step": 19810
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.29468947649002075,
+      "learning_rate": 7.08444840915849e-05,
+      "loss": 0.1386,
+      "step": 19820
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.8740628361701965,
+      "learning_rate": 7.082961641391615e-05,
+      "loss": 0.1403,
+      "step": 19830
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.5060070753097534,
+      "learning_rate": 7.08147487362474e-05,
+      "loss": 0.1439,
+      "step": 19840
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.2761290967464447,
+      "learning_rate": 7.079988105857866e-05,
+      "loss": 0.1454,
+      "step": 19850
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.2517363429069519,
+      "learning_rate": 7.07850133809099e-05,
+      "loss": 0.1381,
+      "step": 19860
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.6545493006706238,
+      "learning_rate": 7.077014570324115e-05,
+      "loss": 0.1349,
+      "step": 19870
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.6688421964645386,
+      "learning_rate": 7.075527802557241e-05,
+      "loss": 0.1384,
+      "step": 19880
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.26555657386779785,
+      "learning_rate": 7.074041034790366e-05,
+      "loss": 0.1385,
+      "step": 19890
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.32324767112731934,
+      "learning_rate": 7.072554267023492e-05,
+      "loss": 0.1362,
+      "step": 19900
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.7072975039482117,
+      "learning_rate": 7.071067499256616e-05,
+      "loss": 0.1416,
+      "step": 19910
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.9408333897590637,
+      "learning_rate": 7.069580731489741e-05,
+      "loss": 0.137,
+      "step": 19920
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.7711098194122314,
+      "learning_rate": 7.068093963722867e-05,
+      "loss": 0.1379,
+      "step": 19930
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.1616792678833008,
+      "learning_rate": 7.066607195955992e-05,
+      "loss": 0.1445,
+      "step": 19940
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3702917993068695,
+      "learning_rate": 7.065120428189117e-05,
+      "loss": 0.1516,
+      "step": 19950
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 1.209560513496399,
+      "learning_rate": 7.063633660422242e-05,
+      "loss": 0.1455,
+      "step": 19960
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.42448684573173523,
+      "learning_rate": 7.062146892655367e-05,
+      "loss": 0.1464,
+      "step": 19970
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.7959136366844177,
+      "learning_rate": 7.060660124888493e-05,
+      "loss": 0.1467,
+      "step": 19980
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.3942021131515503,
+      "learning_rate": 7.059173357121618e-05,
+      "loss": 0.14,
+      "step": 19990
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.2557319104671478,
+      "learning_rate": 7.057686589354743e-05,
+      "loss": 0.1373,
+      "step": 20000
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.9788160920143127,
+      "learning_rate": 7.05619982158787e-05,
+      "loss": 0.1346,
+      "step": 20010
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.39195263385772705,
+      "learning_rate": 7.054713053820994e-05,
+      "loss": 0.1364,
+      "step": 20020
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.37623974680900574,
+      "learning_rate": 7.053226286054119e-05,
+      "loss": 0.142,
+      "step": 20030
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.6487099528312683,
+      "learning_rate": 7.051739518287243e-05,
+      "loss": 0.1424,
+      "step": 20040
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.6612546443939209,
+      "learning_rate": 7.050252750520369e-05,
+      "loss": 0.1381,
+      "step": 20050
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.30832764506340027,
+      "learning_rate": 7.048765982753494e-05,
+      "loss": 0.139,
+      "step": 20060
+    },
+    {
+      "epoch": 2.97,
+      "grad_norm": 0.5305371880531311,
+      "learning_rate": 7.04727921498662e-05,
+      "loss": 0.1415,
+      "step": 20070
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.4207991361618042,
+      "learning_rate": 7.045792447219745e-05,
+      "loss": 0.1385,
+      "step": 20080
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.2595292627811432,
+      "learning_rate": 7.044305679452869e-05,
+      "loss": 0.1286,
+      "step": 20090
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.864605724811554,
+      "learning_rate": 7.042818911685996e-05,
+      "loss": 0.1325,
+      "step": 20100
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.4537006616592407,
+      "learning_rate": 7.04133214391912e-05,
+      "loss": 0.135,
+      "step": 20110
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.3340538740158081,
+      "learning_rate": 7.039845376152246e-05,
+      "loss": 0.1375,
+      "step": 20120
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 1.0329761505126953,
+      "learning_rate": 7.03835860838537e-05,
+      "loss": 0.137,
+      "step": 20130
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.6245138645172119,
+      "learning_rate": 7.036871840618495e-05,
+      "loss": 0.1372,
+      "step": 20140
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.35035985708236694,
+      "learning_rate": 7.03538507285162e-05,
+      "loss": 0.1395,
+      "step": 20150
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.351352721452713,
+      "learning_rate": 7.033898305084746e-05,
+      "loss": 0.1433,
+      "step": 20160
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.4944305717945099,
+      "learning_rate": 7.032411537317872e-05,
+      "loss": 0.1388,
+      "step": 20170
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.8752815127372742,
+      "learning_rate": 7.030924769550996e-05,
+      "loss": 0.1341,
+      "step": 20180
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.30437782406806946,
+      "learning_rate": 7.029438001784123e-05,
+      "loss": 0.1306,
+      "step": 20190
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.797670841217041,
+      "learning_rate": 7.027951234017247e-05,
+      "loss": 0.1469,
+      "step": 20200
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.6191693544387817,
+      "learning_rate": 7.026464466250372e-05,
+      "loss": 0.1346,
+      "step": 20210
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.9991375803947449,
+      "learning_rate": 7.024977698483498e-05,
+      "loss": 0.1402,
+      "step": 20220
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.0731664896011353,
+      "learning_rate": 7.023490930716622e-05,
+      "loss": 0.1451,
+      "step": 20230
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.15701617300510406,
+      "eval_runtime": 2481.1974,
+      "eval_samples_per_second": 235.24,
+      "eval_steps_per_second": 3.676,
+      "step": 20238
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 67460,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 8.63158939557036e+18,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}