diff --git "a/checkpoint-2823840/trainer_state.json" "b/checkpoint-2823840/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-2823840/trainer_state.json"
@@ -0,0 +1,33411 @@
+{
+  "best_metric": 3.972881317138672,
+  "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/existential-there-quantifier/lstm/3/checkpoints/checkpoint-2823840",
+  "epoch": 0.025000606015738065,
+  "eval_steps": 10,
+  "global_step": 2823840,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999998362119627e-05,
+      "loss": 10.8215,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999161405248948e-05,
+      "loss": 7.5553,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.998322810497896e-05,
+      "loss": 7.06,
+      "step": 1024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.997484215746844e-05,
+      "loss": 6.9993,
+      "step": 1536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996645620995792e-05,
+      "loss": 6.9452,
+      "step": 2048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99580702624474e-05,
+      "loss": 6.887,
+      "step": 2560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994968431493688e-05,
+      "loss": 6.7272,
+      "step": 3072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.994129836742636e-05,
+      "loss": 6.623,
+      "step": 3584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.993291241991584e-05,
+      "loss": 6.5272,
+      "step": 4096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992452647240532e-05,
+      "loss": 6.4534,
+      "step": 4608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99161405248948e-05,
+      "loss": 6.3851,
+      "step": 5120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.990775457738428e-05,
+      "loss": 6.3195,
+      "step": 5632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989936862987376e-05,
+      "loss": 6.2438,
+      "step": 6144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.989098268236324e-05,
+      "loss": 6.1818,
+      "step": 6656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.988259673485272e-05,
+      "loss": 6.1253,
+      "step": 7168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98742107873422e-05,
+      "loss": 6.0748,
+      "step": 7680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.986582483983168e-05,
+      "loss": 6.0306,
+      "step": 8192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.985743889232116e-05,
+      "loss": 5.9982,
+      "step": 8704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984905294481064e-05,
+      "loss": 5.9498,
+      "step": 9216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.984068337610385e-05,
+      "loss": 5.9156,
+      "step": 9728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983229742859333e-05,
+      "loss": 5.8805,
+      "step": 10240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.982391148108281e-05,
+      "loss": 5.8319,
+      "step": 10752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981552553357229e-05,
+      "loss": 5.8005,
+      "step": 11264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9807155964865506e-05,
+      "loss": 5.7676,
+      "step": 11776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9798770017354986e-05,
+      "loss": 5.7518,
+      "step": 12288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9790384069844466e-05,
+      "loss": 5.7142,
+      "step": 12800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9781998122333946e-05,
+      "loss": 5.6862,
+      "step": 13312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9773612174823426e-05,
+      "loss": 5.667,
+      "step": 13824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.97652262273129e-05,
+      "loss": 5.6411,
+      "step": 14336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975684027980238e-05,
+      "loss": 5.6139,
+      "step": 14848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974845433229186e-05,
+      "loss": 5.5904,
+      "step": 15360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.974008476358507e-05,
+      "loss": 5.5832,
+      "step": 15872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.973169881607455e-05,
+      "loss": 5.5618,
+      "step": 16384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972331286856403e-05,
+      "loss": 5.5511,
+      "step": 16896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9714926921053515e-05,
+      "loss": 5.5301,
+      "step": 17408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9706557352346724e-05,
+      "loss": 5.5108,
+      "step": 17920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9698171404836204e-05,
+      "loss": 5.5063,
+      "step": 18432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9689785457325684e-05,
+      "loss": 5.4668,
+      "step": 18944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9681399509815164e-05,
+      "loss": 5.4575,
+      "step": 19456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9673013562304644e-05,
+      "loss": 5.44,
+      "step": 19968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.966464399359785e-05,
+      "loss": 5.4334,
+      "step": 20480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.965625804608733e-05,
+      "loss": 5.4098,
+      "step": 20992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.964787209857681e-05,
+      "loss": 5.4172,
+      "step": 21504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.963948615106629e-05,
+      "loss": 5.3823,
+      "step": 22016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.96311165823595e-05,
+      "loss": 5.3822,
+      "step": 22528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.962273063484898e-05,
+      "loss": 5.3764,
+      "step": 23040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.961434468733847e-05,
+      "loss": 5.3677,
+      "step": 23552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.960595873982795e-05,
+      "loss": 5.3599,
+      "step": 24064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959757279231743e-05,
+      "loss": 5.3376,
+      "step": 24576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958920322361064e-05,
+      "loss": 5.3161,
+      "step": 25088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.958081727610012e-05,
+      "loss": 5.3305,
+      "step": 25600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.95724313285896e-05,
+      "loss": 5.3098,
+      "step": 26112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.956406175988281e-05,
+      "loss": 5.2965,
+      "step": 26624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.955567581237229e-05,
+      "loss": 5.2795,
+      "step": 27136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954728986486177e-05,
+      "loss": 5.2859,
+      "step": 27648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.953890391735125e-05,
+      "loss": 5.2659,
+      "step": 28160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9530517969840727e-05,
+      "loss": 5.28,
+      "step": 28672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9522132022330207e-05,
+      "loss": 5.2493,
+      "step": 29184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9513746074819686e-05,
+      "loss": 5.2411,
+      "step": 29696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9505360127309166e-05,
+      "loss": 5.2422,
+      "step": 30208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9496974179798646e-05,
+      "loss": 5.2266,
+      "step": 30720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948858823228813e-05,
+      "loss": 5.2044,
+      "step": 31232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948021866358134e-05,
+      "loss": 5.2103,
+      "step": 31744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.947183271607082e-05,
+      "loss": 5.1871,
+      "step": 32256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.94634467685603e-05,
+      "loss": 5.1888,
+      "step": 32768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.945506082104978e-05,
+      "loss": 5.1996,
+      "step": 33280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9446674873539255e-05,
+      "loss": 5.1912,
+      "step": 33792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9438288926028735e-05,
+      "loss": 5.172,
+      "step": 34304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9429902978518215e-05,
+      "loss": 5.1534,
+      "step": 34816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9421517031007695e-05,
+      "loss": 5.1481,
+      "step": 35328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9413131083497175e-05,
+      "loss": 5.1584,
+      "step": 35840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.940477789359412e-05,
+      "loss": 5.1469,
+      "step": 36352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.93963919460836e-05,
+      "loss": 5.1432,
+      "step": 36864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.938800599857309e-05,
+      "loss": 5.1413,
+      "step": 37376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937962005106256e-05,
+      "loss": 5.1408,
+      "step": 37888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.937123410355204e-05,
+      "loss": 5.1315,
+      "step": 38400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.936284815604152e-05,
+      "loss": 5.1038,
+      "step": 38912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9354462208531e-05,
+      "loss": 5.1086,
+      "step": 39424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.934607626102048e-05,
+      "loss": 5.1027,
+      "step": 39936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.933769031350996e-05,
+      "loss": 5.0883,
+      "step": 40448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932932074480317e-05,
+      "loss": 5.1068,
+      "step": 40960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.932093479729265e-05,
+      "loss": 5.0859,
+      "step": 41472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.931254884978213e-05,
+      "loss": 5.0922,
+      "step": 41984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.930416290227161e-05,
+      "loss": 5.0771,
+      "step": 42496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9295793333564825e-05,
+      "loss": 5.0525,
+      "step": 43008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9287407386054305e-05,
+      "loss": 5.0583,
+      "step": 43520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9279037817347514e-05,
+      "loss": 5.0611,
+      "step": 44032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9270651869836994e-05,
+      "loss": 5.0563,
+      "step": 44544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9262265922326474e-05,
+      "loss": 5.0505,
+      "step": 45056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9253879974815954e-05,
+      "loss": 5.0374,
+      "step": 45568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924551040610916e-05,
+      "loss": 5.0355,
+      "step": 46080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.923712445859864e-05,
+      "loss": 5.0247,
+      "step": 46592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.922873851108812e-05,
+      "loss": 5.0171,
+      "step": 47104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.92203525635776e-05,
+      "loss": 5.0155,
+      "step": 47616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.921196661606708e-05,
+      "loss": 5.008,
+      "step": 48128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.920358066855656e-05,
+      "loss": 5.0159,
+      "step": 48640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.919519472104604e-05,
+      "loss": 4.9988,
+      "step": 49152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.918680877353552e-05,
+      "loss": 4.9903,
+      "step": 49664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917842282602501e-05,
+      "loss": 4.9836,
+      "step": 50176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.917003687851449e-05,
+      "loss": 4.9863,
+      "step": 50688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.91616673098077e-05,
+      "loss": 4.9844,
+      "step": 51200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.915328136229718e-05,
+      "loss": 4.972,
+      "step": 51712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.914489541478666e-05,
+      "loss": 4.9655,
+      "step": 52224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.913650946727614e-05,
+      "loss": 4.9507,
+      "step": 52736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.912812351976562e-05,
+      "loss": 4.9588,
+      "step": 53248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911975395105883e-05,
+      "loss": 4.9455,
+      "step": 53760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.911136800354831e-05,
+      "loss": 4.9402,
+      "step": 54272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9102998434841516e-05,
+      "loss": 4.942,
+      "step": 54784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9094612487330996e-05,
+      "loss": 4.9412,
+      "step": 55296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9086226539820476e-05,
+      "loss": 4.934,
+      "step": 55808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.907784059230996e-05,
+      "loss": 4.9307,
+      "step": 56320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906945464479944e-05,
+      "loss": 4.9209,
+      "step": 56832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.906106869728892e-05,
+      "loss": 4.9214,
+      "step": 57344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9052682749778396e-05,
+      "loss": 4.9111,
+      "step": 57856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9044296802267876e-05,
+      "loss": 4.9058,
+      "step": 58368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9035910854757356e-05,
+      "loss": 4.9145,
+      "step": 58880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9027541286050565e-05,
+      "loss": 4.9045,
+      "step": 59392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901917171734378e-05,
+      "loss": 4.9033,
+      "step": 59904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.901078576983326e-05,
+      "loss": 4.8934,
+      "step": 60416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.900239982232274e-05,
+      "loss": 4.9003,
+      "step": 60928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8994013874812214e-05,
+      "loss": 4.8959,
+      "step": 61440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.89856279273017e-05,
+      "loss": 4.8768,
+      "step": 61952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.897725835859491e-05,
+      "loss": 4.8927,
+      "step": 62464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8968888789888125e-05,
+      "loss": 4.8795,
+      "step": 62976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8960502842377605e-05,
+      "loss": 4.8807,
+      "step": 63488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8952116894867085e-05,
+      "loss": 4.8594,
+      "step": 64000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8943730947356565e-05,
+      "loss": 4.8535,
+      "step": 64512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.893534499984604e-05,
+      "loss": 4.8489,
+      "step": 65024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892695905233552e-05,
+      "loss": 4.8533,
+      "step": 65536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8918573104825e-05,
+      "loss": 4.8427,
+      "step": 66048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.891018715731448e-05,
+      "loss": 4.8606,
+      "step": 66560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.890180120980396e-05,
+      "loss": 4.8555,
+      "step": 67072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.889341526229344e-05,
+      "loss": 4.8375,
+      "step": 67584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.888502931478292e-05,
+      "loss": 4.8421,
+      "step": 68096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88766433672724e-05,
+      "loss": 4.8377,
+      "step": 68608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.886825741976188e-05,
+      "loss": 4.8377,
+      "step": 69120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8859871472251365e-05,
+      "loss": 4.8376,
+      "step": 69632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8851501903544574e-05,
+      "loss": 4.8381,
+      "step": 70144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.884313233483778e-05,
+      "loss": 4.8197,
+      "step": 70656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.883474638732726e-05,
+      "loss": 4.8222,
+      "step": 71168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.882636043981674e-05,
+      "loss": 4.8165,
+      "step": 71680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881797449230622e-05,
+      "loss": 4.8068,
+      "step": 72192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.88095885447957e-05,
+      "loss": 4.8206,
+      "step": 72704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.880120259728518e-05,
+      "loss": 4.8058,
+      "step": 73216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.879281664977466e-05,
+      "loss": 4.7958,
+      "step": 73728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.878444708106787e-05,
+      "loss": 4.8108,
+      "step": 74240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.877606113355735e-05,
+      "loss": 4.794,
+      "step": 74752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.876767518604683e-05,
+      "loss": 4.7911,
+      "step": 75264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.875928923853632e-05,
+      "loss": 4.7942,
+      "step": 75776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.87509032910258e-05,
+      "loss": 4.787,
+      "step": 76288
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.749973297119141,
+      "eval_runtime": 308.6341,
+      "eval_samples_per_second": 1236.386,
+      "eval_steps_per_second": 38.638,
+      "step": 76320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.874251734351528e-05,
+      "loss": 4.773,
+      "step": 76800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.873413139600475e-05,
+      "loss": 4.7833,
+      "step": 77312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.872574544849423e-05,
+      "loss": 4.7971,
+      "step": 77824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.871735950098371e-05,
+      "loss": 4.7722,
+      "step": 78336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870897355347319e-05,
+      "loss": 4.7864,
+      "step": 78848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.870058760596267e-05,
+      "loss": 4.7584,
+      "step": 79360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.869221803725588e-05,
+      "loss": 4.7752,
+      "step": 79872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8683848468549096e-05,
+      "loss": 4.7521,
+      "step": 80384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8675462521038576e-05,
+      "loss": 4.7619,
+      "step": 80896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8667076573528056e-05,
+      "loss": 4.7614,
+      "step": 81408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8658690626017536e-05,
+      "loss": 4.7653,
+      "step": 81920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8650304678507016e-05,
+      "loss": 4.7597,
+      "step": 82432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8641935109800225e-05,
+      "loss": 4.7424,
+      "step": 82944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8633549162289705e-05,
+      "loss": 4.7391,
+      "step": 83456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.862517959358292e-05,
+      "loss": 4.7341,
+      "step": 83968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.86167936460724e-05,
+      "loss": 4.7398,
+      "step": 84480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8608407698561874e-05,
+      "loss": 4.7517,
+      "step": 84992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8600021751051354e-05,
+      "loss": 4.7337,
+      "step": 85504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8591635803540834e-05,
+      "loss": 4.7408,
+      "step": 86016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8583249856030314e-05,
+      "loss": 4.7521,
+      "step": 86528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8574863908519794e-05,
+      "loss": 4.727,
+      "step": 87040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8566477961009274e-05,
+      "loss": 4.7239,
+      "step": 87552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.8558092013498754e-05,
+      "loss": 4.7244,
+      "step": 88064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854970606598824e-05,
+      "loss": 4.7358,
+      "step": 88576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.854132011847772e-05,
+      "loss": 4.7148,
+      "step": 89088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.853295054977093e-05,
+      "loss": 4.7168,
+      "step": 89600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.852456460226041e-05,
+      "loss": 4.7152,
+      "step": 90112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.851617865474989e-05,
+      "loss": 4.7141,
+      "step": 90624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.850779270723937e-05,
+      "loss": 4.6959,
+      "step": 91136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849942313853258e-05,
+      "loss": 4.7005,
+      "step": 91648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.849103719102206e-05,
+      "loss": 4.6971,
+      "step": 92160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.848265124351154e-05,
+      "loss": 4.7056,
+      "step": 92672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.847426529600102e-05,
+      "loss": 4.7102,
+      "step": 93184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.84658793484905e-05,
+      "loss": 4.6944,
+      "step": 93696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.845749340097998e-05,
+      "loss": 4.6919,
+      "step": 94208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.844910745346946e-05,
+      "loss": 4.7031,
+      "step": 94720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.844072150595894e-05,
+      "loss": 4.6767,
+      "step": 95232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.843233555844842e-05,
+      "loss": 4.6836,
+      "step": 95744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8423965989741634e-05,
+      "loss": 4.6719,
+      "step": 96256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8415580042231114e-05,
+      "loss": 4.6799,
+      "step": 96768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.840719409472059e-05,
+      "loss": 4.6702,
+      "step": 97280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.83988245260138e-05,
+      "loss": 4.6856,
+      "step": 97792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.839043857850328e-05,
+      "loss": 4.6668,
+      "step": 98304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.838205263099276e-05,
+      "loss": 4.6745,
+      "step": 98816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8373666683482236e-05,
+      "loss": 4.6671,
+      "step": 99328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8365280735971716e-05,
+      "loss": 4.6726,
+      "step": 99840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8356894788461196e-05,
+      "loss": 4.6692,
+      "step": 100352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8348508840950676e-05,
+      "loss": 4.6585,
+      "step": 100864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.834013927224389e-05,
+      "loss": 4.6463,
+      "step": 101376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.833175332473337e-05,
+      "loss": 4.6697,
+      "step": 101888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.832336737722285e-05,
+      "loss": 4.6572,
+      "step": 102400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.831498142971233e-05,
+      "loss": 4.6495,
+      "step": 102912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.830659548220181e-05,
+      "loss": 4.6443,
+      "step": 103424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.829820953469129e-05,
+      "loss": 4.6514,
+      "step": 103936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828982358718077e-05,
+      "loss": 4.6403,
+      "step": 104448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.828143763967025e-05,
+      "loss": 4.6595,
+      "step": 104960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.827305169215973e-05,
+      "loss": 4.6405,
+      "step": 105472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.826468212345294e-05,
+      "loss": 4.6373,
+      "step": 105984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.825629617594242e-05,
+      "loss": 4.6466,
+      "step": 106496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.82479102284319e-05,
+      "loss": 4.6268,
+      "step": 107008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823952428092138e-05,
+      "loss": 4.6222,
+      "step": 107520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.823113833341086e-05,
+      "loss": 4.6336,
+      "step": 108032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8222768764704077e-05,
+      "loss": 4.6131,
+      "step": 108544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8214382817193557e-05,
+      "loss": 4.6217,
+      "step": 109056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8205996869683037e-05,
+      "loss": 4.6315,
+      "step": 109568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8197610922172516e-05,
+      "loss": 4.6349,
+      "step": 110080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8189224974661996e-05,
+      "loss": 4.6215,
+      "step": 110592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8180839027151476e-05,
+      "loss": 4.6132,
+      "step": 111104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.817245307964095e-05,
+      "loss": 4.6066,
+      "step": 111616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8164083510934165e-05,
+      "loss": 4.6256,
+      "step": 112128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8155697563423645e-05,
+      "loss": 4.6192,
+      "step": 112640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8147311615913125e-05,
+      "loss": 4.6198,
+      "step": 113152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8138942047206334e-05,
+      "loss": 4.6181,
+      "step": 113664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8130556099695814e-05,
+      "loss": 4.6265,
+      "step": 114176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8122170152185294e-05,
+      "loss": 4.6207,
+      "step": 114688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8113784204674774e-05,
+      "loss": 4.5981,
+      "step": 115200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8105398257164254e-05,
+      "loss": 4.6107,
+      "step": 115712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8097012309653734e-05,
+      "loss": 4.6092,
+      "step": 116224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8088626362143214e-05,
+      "loss": 4.5906,
+      "step": 116736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8080240414632694e-05,
+      "loss": 4.6198,
+      "step": 117248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80718708459259e-05,
+      "loss": 4.6045,
+      "step": 117760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.806348489841538e-05,
+      "loss": 4.6099,
+      "step": 118272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.805509895090486e-05,
+      "loss": 4.6065,
+      "step": 118784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.804671300339434e-05,
+      "loss": 4.5812,
+      "step": 119296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.803832705588382e-05,
+      "loss": 4.5973,
+      "step": 119808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.80299411083733e-05,
+      "loss": 4.5946,
+      "step": 120320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.802155516086278e-05,
+      "loss": 4.5978,
+      "step": 120832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.8013185592156e-05,
+      "loss": 4.5957,
+      "step": 121344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.800481602344921e-05,
+      "loss": 4.5913,
+      "step": 121856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.799643007593869e-05,
+      "loss": 4.5919,
+      "step": 122368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.798804412842817e-05,
+      "loss": 4.5878,
+      "step": 122880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797967455972138e-05,
+      "loss": 4.573,
+      "step": 123392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.797128861221086e-05,
+      "loss": 4.58,
+      "step": 123904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.796290266470034e-05,
+      "loss": 4.5838,
+      "step": 124416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.795451671718982e-05,
+      "loss": 4.5871,
+      "step": 124928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7946147148483026e-05,
+      "loss": 4.5745,
+      "step": 125440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7937761200972506e-05,
+      "loss": 4.5683,
+      "step": 125952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7929375253461986e-05,
+      "loss": 4.5696,
+      "step": 126464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.792098930595147e-05,
+      "loss": 4.5779,
+      "step": 126976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.791260335844095e-05,
+      "loss": 4.5759,
+      "step": 127488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.790421741093043e-05,
+      "loss": 4.5692,
+      "step": 128000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.789583146341991e-05,
+      "loss": 4.561,
+      "step": 128512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.788744551590939e-05,
+      "loss": 4.555,
+      "step": 129024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.78790759472026e-05,
+      "loss": 4.5643,
+      "step": 129536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.787068999969208e-05,
+      "loss": 4.5512,
+      "step": 130048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.786230405218156e-05,
+      "loss": 4.5558,
+      "step": 130560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.785391810467104e-05,
+      "loss": 4.5576,
+      "step": 131072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.784554853596425e-05,
+      "loss": 4.557,
+      "step": 131584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.783716258845373e-05,
+      "loss": 4.5544,
+      "step": 132096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782877664094321e-05,
+      "loss": 4.559,
+      "step": 132608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.782040707223642e-05,
+      "loss": 4.5464,
+      "step": 133120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7812021124725906e-05,
+      "loss": 4.549,
+      "step": 133632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7803635177215386e-05,
+      "loss": 4.5459,
+      "step": 134144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7795249229704866e-05,
+      "loss": 4.5408,
+      "step": 134656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7786863282194346e-05,
+      "loss": 4.556,
+      "step": 135168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7778477334683826e-05,
+      "loss": 4.5489,
+      "step": 135680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7770091387173306e-05,
+      "loss": 4.5484,
+      "step": 136192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7761705439662786e-05,
+      "loss": 4.5413,
+      "step": 136704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7753335870955995e-05,
+      "loss": 4.5491,
+      "step": 137216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7744949923445475e-05,
+      "loss": 4.5463,
+      "step": 137728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7736563975934955e-05,
+      "loss": 4.5331,
+      "step": 138240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7728178028424435e-05,
+      "loss": 4.5536,
+      "step": 138752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7719808459717644e-05,
+      "loss": 4.5384,
+      "step": 139264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7711422512207124e-05,
+      "loss": 4.5417,
+      "step": 139776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7703036564696604e-05,
+      "loss": 4.5251,
+      "step": 140288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7694650617186084e-05,
+      "loss": 4.5262,
+      "step": 140800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7686264669675564e-05,
+      "loss": 4.5203,
+      "step": 141312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.767789510096878e-05,
+      "loss": 4.5235,
+      "step": 141824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766950915345826e-05,
+      "loss": 4.5192,
+      "step": 142336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.766112320594773e-05,
+      "loss": 4.5362,
+      "step": 142848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.765273725843721e-05,
+      "loss": 4.5377,
+      "step": 143360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.764435131092669e-05,
+      "loss": 4.5251,
+      "step": 143872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.763596536341617e-05,
+      "loss": 4.5193,
+      "step": 144384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.762757941590565e-05,
+      "loss": 4.5234,
+      "step": 144896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761919346839513e-05,
+      "loss": 4.5262,
+      "step": 145408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.761080752088461e-05,
+      "loss": 4.532,
+      "step": 145920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.760243795217783e-05,
+      "loss": 4.5272,
+      "step": 146432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.759405200466731e-05,
+      "loss": 4.5198,
+      "step": 146944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.758566605715679e-05,
+      "loss": 4.5121,
+      "step": 147456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.757728010964627e-05,
+      "loss": 4.5202,
+      "step": 147968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756889416213575e-05,
+      "loss": 4.5129,
+      "step": 148480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.756052459342896e-05,
+      "loss": 4.5191,
+      "step": 148992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.755213864591844e-05,
+      "loss": 4.5135,
+      "step": 149504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7543769077211646e-05,
+      "loss": 4.5083,
+      "step": 150016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7535383129701126e-05,
+      "loss": 4.5212,
+      "step": 150528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7526997182190606e-05,
+      "loss": 4.5073,
+      "step": 151040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7518611234680086e-05,
+      "loss": 4.502,
+      "step": 151552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7510225287169566e-05,
+      "loss": 4.5065,
+      "step": 152064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.7501839339659046e-05,
+      "loss": 4.5023,
+      "step": 152576
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.472645282745361,
+      "eval_runtime": 305.2704,
+      "eval_samples_per_second": 1250.01,
+      "eval_steps_per_second": 39.064,
+      "step": 152640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7493453392148526e-05,
+      "loss": 4.4888,
+      "step": 153088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.748506744463801e-05,
+      "loss": 4.5014,
+      "step": 153600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.747668149712749e-05,
+      "loss": 4.5164,
+      "step": 154112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.746829554961697e-05,
+      "loss": 4.496,
+      "step": 154624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7459909602106446e-05,
+      "loss": 4.5078,
+      "step": 155136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7451523654595926e-05,
+      "loss": 4.4881,
+      "step": 155648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7443137707085406e-05,
+      "loss": 4.5028,
+      "step": 156160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7434751759574886e-05,
+      "loss": 4.4786,
+      "step": 156672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7426365812064366e-05,
+      "loss": 4.4957,
+      "step": 157184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7417979864553846e-05,
+      "loss": 4.4958,
+      "step": 157696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7409593917043326e-05,
+      "loss": 4.4962,
+      "step": 158208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7401207969532806e-05,
+      "loss": 4.5029,
+      "step": 158720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7392838400826015e-05,
+      "loss": 4.4784,
+      "step": 159232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7384452453315495e-05,
+      "loss": 4.4802,
+      "step": 159744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7376066505804975e-05,
+      "loss": 4.4765,
+      "step": 160256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.736768055829446e-05,
+      "loss": 4.4858,
+      "step": 160768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735929461078394e-05,
+      "loss": 4.4903,
+      "step": 161280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.735090866327342e-05,
+      "loss": 4.4831,
+      "step": 161792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.73425227157629e-05,
+      "loss": 4.4865,
+      "step": 162304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.733415314705611e-05,
+      "loss": 4.5014,
+      "step": 162816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.732576719954559e-05,
+      "loss": 4.4771,
+      "step": 163328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.731738125203507e-05,
+      "loss": 4.4741,
+      "step": 163840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730899530452455e-05,
+      "loss": 4.4784,
+      "step": 164352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.730060935701403e-05,
+      "loss": 4.4904,
+      "step": 164864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.729222340950351e-05,
+      "loss": 4.4725,
+      "step": 165376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728383746199298e-05,
+      "loss": 4.4717,
+      "step": 165888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.727545151448246e-05,
+      "loss": 4.4726,
+      "step": 166400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.726706556697194e-05,
+      "loss": 4.4727,
+      "step": 166912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.725867961946143e-05,
+      "loss": 4.4563,
+      "step": 167424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.725029367195091e-05,
+      "loss": 4.4648,
+      "step": 167936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.724192410324412e-05,
+      "loss": 4.4642,
+      "step": 168448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7233554534537335e-05,
+      "loss": 4.4673,
+      "step": 168960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.722516858702681e-05,
+      "loss": 4.4796,
+      "step": 169472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.721678263951629e-05,
+      "loss": 4.4591,
+      "step": 169984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720839669200577e-05,
+      "loss": 4.4628,
+      "step": 170496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.720001074449525e-05,
+      "loss": 4.471,
+      "step": 171008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.719162479698473e-05,
+      "loss": 4.4538,
+      "step": 171520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.718323884947421e-05,
+      "loss": 4.4527,
+      "step": 172032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.717485290196369e-05,
+      "loss": 4.4473,
+      "step": 172544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.71664833332569e-05,
+      "loss": 4.456,
+      "step": 173056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7158097385746384e-05,
+      "loss": 4.4456,
+      "step": 173568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7149711438235864e-05,
+      "loss": 4.4649,
+      "step": 174080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7141325490725343e-05,
+      "loss": 4.4425,
+      "step": 174592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.713295592201855e-05,
+      "loss": 4.4596,
+      "step": 175104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.712456997450803e-05,
+      "loss": 4.4509,
+      "step": 175616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.711618402699751e-05,
+      "loss": 4.4499,
+      "step": 176128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.710779807948699e-05,
+      "loss": 4.4553,
+      "step": 176640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709941213197647e-05,
+      "loss": 4.4391,
+      "step": 177152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.709104256326968e-05,
+      "loss": 4.4324,
+      "step": 177664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.708265661575916e-05,
+      "loss": 4.4544,
+      "step": 178176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.707427066824864e-05,
+      "loss": 4.4456,
+      "step": 178688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.706588472073812e-05,
+      "loss": 4.436,
+      "step": 179200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705751515203134e-05,
+      "loss": 4.4384,
+      "step": 179712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.704912920452082e-05,
+      "loss": 4.4441,
+      "step": 180224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.70407432570103e-05,
+      "loss": 4.4264,
+      "step": 180736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7032373688303506e-05,
+      "loss": 4.4492,
+      "step": 181248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7024004119596715e-05,
+      "loss": 4.434,
+      "step": 181760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7015618172086195e-05,
+      "loss": 4.4322,
+      "step": 182272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7007232224575675e-05,
+      "loss": 4.4442,
+      "step": 182784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6998846277065155e-05,
+      "loss": 4.4233,
+      "step": 183296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6990460329554635e-05,
+      "loss": 4.4198,
+      "step": 183808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6982074382044115e-05,
+      "loss": 4.4338,
+      "step": 184320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6973688434533595e-05,
+      "loss": 4.4162,
+      "step": 184832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6965318865826804e-05,
+      "loss": 4.4193,
+      "step": 185344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.695693291831629e-05,
+      "loss": 4.4328,
+      "step": 185856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694854697080577e-05,
+      "loss": 4.437,
+      "step": 186368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.694016102329525e-05,
+      "loss": 4.4213,
+      "step": 186880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693177507578473e-05,
+      "loss": 4.4199,
+      "step": 187392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.692338912827421e-05,
+      "loss": 4.4094,
+      "step": 187904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.691500318076369e-05,
+      "loss": 4.4329,
+      "step": 188416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.690661723325317e-05,
+      "loss": 4.4229,
+      "step": 188928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6898231285742644e-05,
+      "loss": 4.4303,
+      "step": 189440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6889845338232124e-05,
+      "loss": 4.4243,
+      "step": 189952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6881459390721604e-05,
+      "loss": 4.4337,
+      "step": 190464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6873073443211084e-05,
+      "loss": 4.4358,
+      "step": 190976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.686470387450429e-05,
+      "loss": 4.4102,
+      "step": 191488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.685631792699377e-05,
+      "loss": 4.4197,
+      "step": 192000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.684793197948325e-05,
+      "loss": 4.4245,
+      "step": 192512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683956241077647e-05,
+      "loss": 4.4052,
+      "step": 193024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.683117646326595e-05,
+      "loss": 4.4317,
+      "step": 193536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6822806894559164e-05,
+      "loss": 4.421,
+      "step": 194048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6814420947048644e-05,
+      "loss": 4.4261,
+      "step": 194560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.680603499953812e-05,
+      "loss": 4.426,
+      "step": 195072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.67976490520276e-05,
+      "loss": 4.4003,
+      "step": 195584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678926310451708e-05,
+      "loss": 4.4164,
+      "step": 196096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.678087715700656e-05,
+      "loss": 4.4129,
+      "step": 196608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.677249120949604e-05,
+      "loss": 4.4196,
+      "step": 197120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.676410526198552e-05,
+      "loss": 4.4126,
+      "step": 197632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.675575207208246e-05,
+      "loss": 4.4154,
+      "step": 198144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.674736612457194e-05,
+      "loss": 4.4127,
+      "step": 198656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.673898017706142e-05,
+      "loss": 4.4098,
+      "step": 199168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.67305942295509e-05,
+      "loss": 4.3974,
+      "step": 199680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.672220828204038e-05,
+      "loss": 4.4032,
+      "step": 200192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.671382233452986e-05,
+      "loss": 4.4094,
+      "step": 200704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.670543638701934e-05,
+      "loss": 4.4107,
+      "step": 201216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.669705043950882e-05,
+      "loss": 4.4053,
+      "step": 201728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.66886644919983e-05,
+      "loss": 4.3927,
+      "step": 202240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.668027854448778e-05,
+      "loss": 4.3971,
+      "step": 202752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.667189259697726e-05,
+      "loss": 4.4085,
+      "step": 203264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.666350664946674e-05,
+      "loss": 4.4038,
+      "step": 203776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665512070195622e-05,
+      "loss": 4.3996,
+      "step": 204288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.664675113324943e-05,
+      "loss": 4.393,
+      "step": 204800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.663836518573891e-05,
+      "loss": 4.3843,
+      "step": 205312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662997923822839e-05,
+      "loss": 4.4016,
+      "step": 205824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.662159329071788e-05,
+      "loss": 4.3779,
+      "step": 206336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.661320734320736e-05,
+      "loss": 4.3935,
+      "step": 206848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6604837774500567e-05,
+      "loss": 4.386,
+      "step": 207360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6596451826990047e-05,
+      "loss": 4.3995,
+      "step": 207872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6588082258283256e-05,
+      "loss": 4.3871,
+      "step": 208384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6579696310772736e-05,
+      "loss": 4.3977,
+      "step": 208896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6571310363262216e-05,
+      "loss": 4.3839,
+      "step": 209408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6562924415751695e-05,
+      "loss": 4.3826,
+      "step": 209920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6554538468241175e-05,
+      "loss": 4.3893,
+      "step": 210432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.654615252073065e-05,
+      "loss": 4.3769,
+      "step": 210944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.653776657322013e-05,
+      "loss": 4.3971,
+      "step": 211456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6529380625709615e-05,
+      "loss": 4.392,
+      "step": 211968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6520994678199095e-05,
+      "loss": 4.3912,
+      "step": 212480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6512608730688575e-05,
+      "loss": 4.3814,
+      "step": 212992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6504222783178055e-05,
+      "loss": 4.3879,
+      "step": 213504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6495836835667535e-05,
+      "loss": 4.3919,
+      "step": 214016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6487467266960744e-05,
+      "loss": 4.3723,
+      "step": 214528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6479081319450224e-05,
+      "loss": 4.3961,
+      "step": 215040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6470695371939704e-05,
+      "loss": 4.3849,
+      "step": 215552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.646232580323291e-05,
+      "loss": 4.3839,
+      "step": 216064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.645393985572239e-05,
+      "loss": 4.3747,
+      "step": 216576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.64455702870156e-05,
+      "loss": 4.3766,
+      "step": 217088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.643718433950508e-05,
+      "loss": 4.3608,
+      "step": 217600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642879839199457e-05,
+      "loss": 4.3726,
+      "step": 218112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.642041244448405e-05,
+      "loss": 4.367,
+      "step": 218624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.641202649697353e-05,
+      "loss": 4.3829,
+      "step": 219136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.640364054946301e-05,
+      "loss": 4.3872,
+      "step": 219648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.639525460195249e-05,
+      "loss": 4.3748,
+      "step": 220160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.638686865444197e-05,
+      "loss": 4.3702,
+      "step": 220672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637848270693145e-05,
+      "loss": 4.3727,
+      "step": 221184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.637009675942093e-05,
+      "loss": 4.3785,
+      "step": 221696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.636172719071414e-05,
+      "loss": 4.3823,
+      "step": 222208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.635334124320362e-05,
+      "loss": 4.3796,
+      "step": 222720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63449552956931e-05,
+      "loss": 4.3683,
+      "step": 223232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.633656934818258e-05,
+      "loss": 4.3729,
+      "step": 223744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.632818340067206e-05,
+      "loss": 4.3772,
+      "step": 224256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631979745316154e-05,
+      "loss": 4.3633,
+      "step": 224768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.631141150565102e-05,
+      "loss": 4.3731,
+      "step": 225280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.63030255581405e-05,
+      "loss": 4.3663,
+      "step": 225792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.629465598943371e-05,
+      "loss": 4.3635,
+      "step": 226304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.628627004192319e-05,
+      "loss": 4.3768,
+      "step": 226816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.62779004732164e-05,
+      "loss": 4.3648,
+      "step": 227328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626951452570588e-05,
+      "loss": 4.358,
+      "step": 227840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.626112857819536e-05,
+      "loss": 4.3626,
+      "step": 228352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.6252742630684835e-05,
+      "loss": 4.3574,
+      "step": 228864
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.3384690284729,
+      "eval_runtime": 294.8635,
+      "eval_samples_per_second": 1294.128,
+      "eval_steps_per_second": 40.442,
+      "step": 228960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6244356683174315e-05,
+      "loss": 4.3548,
+      "step": 229376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6235970735663795e-05,
+      "loss": 4.3566,
+      "step": 229888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6227584788153275e-05,
+      "loss": 4.377,
+      "step": 230400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6219215219446484e-05,
+      "loss": 4.3561,
+      "step": 230912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.621082927193597e-05,
+      "loss": 4.373,
+      "step": 231424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.620244332442545e-05,
+      "loss": 4.3465,
+      "step": 231936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.619405737691493e-05,
+      "loss": 4.365,
+      "step": 232448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.618567142940441e-05,
+      "loss": 4.3426,
+      "step": 232960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.617730186069762e-05,
+      "loss": 4.3597,
+      "step": 233472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.61689159131871e-05,
+      "loss": 4.3622,
+      "step": 233984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.616052996567658e-05,
+      "loss": 4.359,
+      "step": 234496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.615214401816606e-05,
+      "loss": 4.3641,
+      "step": 235008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.614377444945927e-05,
+      "loss": 4.3448,
+      "step": 235520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.613538850194875e-05,
+      "loss": 4.3453,
+      "step": 236032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.612700255443823e-05,
+      "loss": 4.3442,
+      "step": 236544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611861660692771e-05,
+      "loss": 4.3475,
+      "step": 237056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.611023065941719e-05,
+      "loss": 4.3572,
+      "step": 237568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.610184471190667e-05,
+      "loss": 4.3523,
+      "step": 238080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6093458764396156e-05,
+      "loss": 4.3519,
+      "step": 238592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6085072816885636e-05,
+      "loss": 4.3677,
+      "step": 239104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6076686869375116e-05,
+      "loss": 4.3497,
+      "step": 239616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6068300921864595e-05,
+      "loss": 4.3476,
+      "step": 240128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6059914974354075e-05,
+      "loss": 4.3504,
+      "step": 240640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6051529026843555e-05,
+      "loss": 4.3553,
+      "step": 241152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6043159458136764e-05,
+      "loss": 4.3404,
+      "step": 241664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6034789889429974e-05,
+      "loss": 4.3454,
+      "step": 242176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6026403941919454e-05,
+      "loss": 4.3447,
+      "step": 242688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6018017994408933e-05,
+      "loss": 4.338,
+      "step": 243200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6009632046898413e-05,
+      "loss": 4.332,
+      "step": 243712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.6001246099387893e-05,
+      "loss": 4.3346,
+      "step": 244224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.599286015187737e-05,
+      "loss": 4.3422,
+      "step": 244736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.598449058317059e-05,
+      "loss": 4.3409,
+      "step": 245248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.597610463566007e-05,
+      "loss": 4.3513,
+      "step": 245760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.596771868814955e-05,
+      "loss": 4.3421,
+      "step": 246272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595934911944276e-05,
+      "loss": 4.3298,
+      "step": 246784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.595096317193224e-05,
+      "loss": 4.3424,
+      "step": 247296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.594257722442172e-05,
+      "loss": 4.3354,
+      "step": 247808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.59341912769112e-05,
+      "loss": 4.3259,
+      "step": 248320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.592580532940067e-05,
+      "loss": 4.3285,
+      "step": 248832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.591741938189015e-05,
+      "loss": 4.3286,
+      "step": 249344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.590903343437963e-05,
+      "loss": 4.3262,
+      "step": 249856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.590064748686911e-05,
+      "loss": 4.3421,
+      "step": 250368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.589226153935859e-05,
+      "loss": 4.3234,
+      "step": 250880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.588387559184808e-05,
+      "loss": 4.3368,
+      "step": 251392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.587548964433756e-05,
+      "loss": 4.3336,
+      "step": 251904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.586710369682704e-05,
+      "loss": 4.3261,
+      "step": 252416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585873412812025e-05,
+      "loss": 4.3361,
+      "step": 252928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.585034818060973e-05,
+      "loss": 4.3186,
+      "step": 253440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.584196223309921e-05,
+      "loss": 4.3073,
+      "step": 253952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5833592664392416e-05,
+      "loss": 4.3397,
+      "step": 254464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5825206716881896e-05,
+      "loss": 4.3269,
+      "step": 254976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5816820769371376e-05,
+      "loss": 4.3223,
+      "step": 255488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5808434821860856e-05,
+      "loss": 4.3179,
+      "step": 256000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5800048874350336e-05,
+      "loss": 4.3233,
+      "step": 256512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5791662926839816e-05,
+      "loss": 4.3123,
+      "step": 257024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5783293358133025e-05,
+      "loss": 4.3285,
+      "step": 257536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.577490741062251e-05,
+      "loss": 4.3223,
+      "step": 258048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.576653784191572e-05,
+      "loss": 4.3164,
+      "step": 258560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57581518944052e-05,
+      "loss": 4.3273,
+      "step": 259072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574976594689468e-05,
+      "loss": 4.3074,
+      "step": 259584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.574137999938416e-05,
+      "loss": 4.306,
+      "step": 260096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.573299405187364e-05,
+      "loss": 4.3237,
+      "step": 260608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.572460810436312e-05,
+      "loss": 4.2981,
+      "step": 261120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.57162221568526e-05,
+      "loss": 4.3026,
+      "step": 261632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.570783620934208e-05,
+      "loss": 4.3214,
+      "step": 262144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569946664063529e-05,
+      "loss": 4.3242,
+      "step": 262656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.569108069312477e-05,
+      "loss": 4.3048,
+      "step": 263168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.568269474561425e-05,
+      "loss": 4.3148,
+      "step": 263680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.567430879810373e-05,
+      "loss": 4.2965,
+      "step": 264192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.566592285059321e-05,
+      "loss": 4.3194,
+      "step": 264704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5657553281886425e-05,
+      "loss": 4.3116,
+      "step": 265216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5649167334375905e-05,
+      "loss": 4.3209,
+      "step": 265728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5640781386865385e-05,
+      "loss": 4.3116,
+      "step": 266240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.563239543935486e-05,
+      "loss": 4.3227,
+      "step": 266752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5624025870648074e-05,
+      "loss": 4.3267,
+      "step": 267264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5615639923137554e-05,
+      "loss": 4.2964,
+      "step": 267776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.5607253975627034e-05,
+      "loss": 4.3151,
+      "step": 268288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559886802811651e-05,
+      "loss": 4.3173,
+      "step": 268800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.559048208060599e-05,
+      "loss": 4.2891,
+      "step": 269312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.55821125118992e-05,
+      "loss": 4.3259,
+      "step": 269824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.557372656438868e-05,
+      "loss": 4.313,
+      "step": 270336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.556534061687816e-05,
+      "loss": 4.3144,
+      "step": 270848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.555695466936764e-05,
+      "loss": 4.3197,
+      "step": 271360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554858510066086e-05,
+      "loss": 4.296,
+      "step": 271872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.554019915315033e-05,
+      "loss": 4.3025,
+      "step": 272384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.553181320563981e-05,
+      "loss": 4.3081,
+      "step": 272896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.552342725812929e-05,
+      "loss": 4.3118,
+      "step": 273408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.551504131061877e-05,
+      "loss": 4.3056,
+      "step": 273920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.550667174191198e-05,
+      "loss": 4.3105,
+      "step": 274432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.549828579440146e-05,
+      "loss": 4.3004,
+      "step": 274944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548989984689094e-05,
+      "loss": 4.3101,
+      "step": 275456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.548151389938042e-05,
+      "loss": 4.291,
+      "step": 275968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.54731279518699e-05,
+      "loss": 4.2996,
+      "step": 276480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.546474200435939e-05,
+      "loss": 4.3067,
+      "step": 276992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.545635605684887e-05,
+      "loss": 4.308,
+      "step": 277504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.544797010933835e-05,
+      "loss": 4.3018,
+      "step": 278016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.543958416182783e-05,
+      "loss": 4.2878,
+      "step": 278528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5431214593121036e-05,
+      "loss": 4.2928,
+      "step": 279040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5422828645610516e-05,
+      "loss": 4.3021,
+      "step": 279552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5414442698099996e-05,
+      "loss": 4.3034,
+      "step": 280064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5406056750589476e-05,
+      "loss": 4.2988,
+      "step": 280576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5397687181882685e-05,
+      "loss": 4.2899,
+      "step": 281088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5389301234372165e-05,
+      "loss": 4.2845,
+      "step": 281600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5380915286861645e-05,
+      "loss": 4.2984,
+      "step": 282112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5372529339351125e-05,
+      "loss": 4.2797,
+      "step": 282624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.536415977064434e-05,
+      "loss": 4.2903,
+      "step": 283136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.535579020193755e-05,
+      "loss": 4.2824,
+      "step": 283648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.534742063323076e-05,
+      "loss": 4.2965,
+      "step": 284160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.533903468572024e-05,
+      "loss": 4.2876,
+      "step": 284672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.533064873820972e-05,
+      "loss": 4.3011,
+      "step": 285184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.53222627906992e-05,
+      "loss": 4.2802,
+      "step": 285696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.531387684318868e-05,
+      "loss": 4.2854,
+      "step": 286208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.530549089567816e-05,
+      "loss": 4.2881,
+      "step": 286720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.529710494816764e-05,
+      "loss": 4.2833,
+      "step": 287232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.528871900065712e-05,
+      "loss": 4.293,
+      "step": 287744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.52803330531466e-05,
+      "loss": 4.2978,
+      "step": 288256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.527196348443981e-05,
+      "loss": 4.2911,
+      "step": 288768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5263577536929295e-05,
+      "loss": 4.2828,
+      "step": 289280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5255191589418775e-05,
+      "loss": 4.2891,
+      "step": 289792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5246805641908255e-05,
+      "loss": 4.2932,
+      "step": 290304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5238419694397735e-05,
+      "loss": 4.277,
+      "step": 290816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5230033746887215e-05,
+      "loss": 4.2983,
+      "step": 291328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5221647799376694e-05,
+      "loss": 4.2838,
+      "step": 291840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5213278230669904e-05,
+      "loss": 4.2902,
+      "step": 292352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5204892283159384e-05,
+      "loss": 4.2808,
+      "step": 292864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5196506335648863e-05,
+      "loss": 4.2805,
+      "step": 293376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5188120388138343e-05,
+      "loss": 4.2648,
+      "step": 293888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.517976719823528e-05,
+      "loss": 4.2822,
+      "step": 294400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.517138125072476e-05,
+      "loss": 4.271,
+      "step": 294912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.516299530321425e-05,
+      "loss": 4.2839,
+      "step": 295424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.515460935570373e-05,
+      "loss": 4.2885,
+      "step": 295936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.514622340819321e-05,
+      "loss": 4.2865,
+      "step": 296448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.513783746068269e-05,
+      "loss": 4.2719,
+      "step": 296960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.512945151317217e-05,
+      "loss": 4.2827,
+      "step": 297472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.512106556566164e-05,
+      "loss": 4.2871,
+      "step": 297984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.511267961815112e-05,
+      "loss": 4.2871,
+      "step": 298496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.51042936706406e-05,
+      "loss": 4.2858,
+      "step": 299008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.509590772313008e-05,
+      "loss": 4.2776,
+      "step": 299520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.508752177561956e-05,
+      "loss": 4.2765,
+      "step": 300032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.507915220691277e-05,
+      "loss": 4.286,
+      "step": 300544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5070782638205986e-05,
+      "loss": 4.2706,
+      "step": 301056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5062396690695466e-05,
+      "loss": 4.2813,
+      "step": 301568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5054010743184946e-05,
+      "loss": 4.2718,
+      "step": 302080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5045624795674426e-05,
+      "loss": 4.2755,
+      "step": 302592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5037238848163906e-05,
+      "loss": 4.2866,
+      "step": 303104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5028852900653386e-05,
+      "loss": 4.2734,
+      "step": 303616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5020466953142866e-05,
+      "loss": 4.2636,
+      "step": 304128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5012081005632346e-05,
+      "loss": 4.2698,
+      "step": 304640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.5003711436925555e-05,
+      "loss": 4.2706,
+      "step": 305152
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.256267547607422,
+      "eval_runtime": 295.3709,
+      "eval_samples_per_second": 1291.905,
+      "eval_steps_per_second": 40.373,
+      "step": 305280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4995325489415035e-05,
+      "loss": 4.2581,
+      "step": 305664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4986939541904515e-05,
+      "loss": 4.2676,
+      "step": 306176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4978553594393995e-05,
+      "loss": 4.2837,
+      "step": 306688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4970167646883475e-05,
+      "loss": 4.2673,
+      "step": 307200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4961781699372955e-05,
+      "loss": 4.2813,
+      "step": 307712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4953395751862435e-05,
+      "loss": 4.2644,
+      "step": 308224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4945009804351915e-05,
+      "loss": 4.2707,
+      "step": 308736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4936623856841395e-05,
+      "loss": 4.2596,
+      "step": 309248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.492823790933088e-05,
+      "loss": 4.2662,
+      "step": 309760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4919851961820354e-05,
+      "loss": 4.2726,
+      "step": 310272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4911466014309834e-05,
+      "loss": 4.2713,
+      "step": 310784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4903080066799314e-05,
+      "loss": 4.2766,
+      "step": 311296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.489471049809253e-05,
+      "loss": 4.256,
+      "step": 311808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4886324550582003e-05,
+      "loss": 4.2596,
+      "step": 312320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4877938603071483e-05,
+      "loss": 4.2573,
+      "step": 312832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.486955265556096e-05,
+      "loss": 4.2586,
+      "step": 313344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.486116670805044e-05,
+      "loss": 4.2709,
+      "step": 313856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.485278076053992e-05,
+      "loss": 4.2635,
+      "step": 314368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.48443948130294e-05,
+      "loss": 4.2673,
+      "step": 314880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.483600886551888e-05,
+      "loss": 4.2822,
+      "step": 315392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.482762291800836e-05,
+      "loss": 4.2681,
+      "step": 315904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481923697049785e-05,
+      "loss": 4.263,
+      "step": 316416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.481085102298733e-05,
+      "loss": 4.2623,
+      "step": 316928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.480248145428054e-05,
+      "loss": 4.2674,
+      "step": 317440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.479409550677002e-05,
+      "loss": 4.2559,
+      "step": 317952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.47857095592595e-05,
+      "loss": 4.2618,
+      "step": 318464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.477732361174898e-05,
+      "loss": 4.2574,
+      "step": 318976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476893766423846e-05,
+      "loss": 4.2559,
+      "step": 319488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.476055171672794e-05,
+      "loss": 4.2499,
+      "step": 320000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475216576921742e-05,
+      "loss": 4.2528,
+      "step": 320512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.474377982170689e-05,
+      "loss": 4.2546,
+      "step": 321024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.473541025300011e-05,
+      "loss": 4.2596,
+      "step": 321536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472702430548959e-05,
+      "loss": 4.2674,
+      "step": 322048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4718654736782804e-05,
+      "loss": 4.259,
+      "step": 322560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4710268789272283e-05,
+      "loss": 4.248,
+      "step": 323072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4701882841761763e-05,
+      "loss": 4.2617,
+      "step": 323584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.469351327305497e-05,
+      "loss": 4.2589,
+      "step": 324096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.468512732554445e-05,
+      "loss": 4.2381,
+      "step": 324608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.467674137803393e-05,
+      "loss": 4.2431,
+      "step": 325120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.466835543052341e-05,
+      "loss": 4.2488,
+      "step": 325632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.465996948301289e-05,
+      "loss": 4.2423,
+      "step": 326144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4651583535502366e-05,
+      "loss": 4.262,
+      "step": 326656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4643197587991846e-05,
+      "loss": 4.2427,
+      "step": 327168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4634811640481325e-05,
+      "loss": 4.2524,
+      "step": 327680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4626425692970805e-05,
+      "loss": 4.2519,
+      "step": 328192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.461805612426402e-05,
+      "loss": 4.25,
+      "step": 328704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46096701767535e-05,
+      "loss": 4.2536,
+      "step": 329216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.460128422924298e-05,
+      "loss": 4.2388,
+      "step": 329728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.459291466053619e-05,
+      "loss": 4.2238,
+      "step": 330240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.458452871302567e-05,
+      "loss": 4.263,
+      "step": 330752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.457614276551515e-05,
+      "loss": 4.2439,
+      "step": 331264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.456775681800463e-05,
+      "loss": 4.2447,
+      "step": 331776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455937087049411e-05,
+      "loss": 4.2415,
+      "step": 332288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455098492298359e-05,
+      "loss": 4.2385,
+      "step": 332800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.454259897547307e-05,
+      "loss": 4.2356,
+      "step": 333312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.453421302796255e-05,
+      "loss": 4.2451,
+      "step": 333824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.452582708045203e-05,
+      "loss": 4.2429,
+      "step": 334336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.451745751174524e-05,
+      "loss": 4.2377,
+      "step": 334848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.450907156423472e-05,
+      "loss": 4.2539,
+      "step": 335360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4500685616724206e-05,
+      "loss": 4.2268,
+      "step": 335872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4492299669213686e-05,
+      "loss": 4.2249,
+      "step": 336384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4483913721703166e-05,
+      "loss": 4.2511,
+      "step": 336896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4475527774192646e-05,
+      "loss": 4.217,
+      "step": 337408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4467141826682126e-05,
+      "loss": 4.2242,
+      "step": 337920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4458755879171606e-05,
+      "loss": 4.2462,
+      "step": 338432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4450386310464815e-05,
+      "loss": 4.2469,
+      "step": 338944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4442000362954295e-05,
+      "loss": 4.2293,
+      "step": 339456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4433630794247504e-05,
+      "loss": 4.2388,
+      "step": 339968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4425244846736984e-05,
+      "loss": 4.22,
+      "step": 340480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4416858899226464e-05,
+      "loss": 4.2412,
+      "step": 340992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4408472951715944e-05,
+      "loss": 4.2329,
+      "step": 341504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4400087004205423e-05,
+      "loss": 4.2418,
+      "step": 342016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4391701056694903e-05,
+      "loss": 4.2355,
+      "step": 342528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.438333148798812e-05,
+      "loss": 4.2461,
+      "step": 343040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43749455404776e-05,
+      "loss": 4.25,
+      "step": 343552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.436655959296708e-05,
+      "loss": 4.2234,
+      "step": 344064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.435817364545655e-05,
+      "loss": 4.2345,
+      "step": 344576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.434978769794603e-05,
+      "loss": 4.2424,
+      "step": 345088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.434140175043551e-05,
+      "loss": 4.2178,
+      "step": 345600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.433303218172873e-05,
+      "loss": 4.2466,
+      "step": 346112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.43246462342182e-05,
+      "loss": 4.243,
+      "step": 346624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.431626028670768e-05,
+      "loss": 4.2419,
+      "step": 347136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.430787433919716e-05,
+      "loss": 4.2455,
+      "step": 347648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429948839168664e-05,
+      "loss": 4.2227,
+      "step": 348160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.429111882297986e-05,
+      "loss": 4.2271,
+      "step": 348672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.428273287546934e-05,
+      "loss": 4.2341,
+      "step": 349184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.427434692795882e-05,
+      "loss": 4.2411,
+      "step": 349696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.42659609804483e-05,
+      "loss": 4.2344,
+      "step": 350208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4257591411741506e-05,
+      "loss": 4.2351,
+      "step": 350720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4249205464230986e-05,
+      "loss": 4.224,
+      "step": 351232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4240819516720466e-05,
+      "loss": 4.2384,
+      "step": 351744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4232433569209946e-05,
+      "loss": 4.2247,
+      "step": 352256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4224047621699426e-05,
+      "loss": 4.2216,
+      "step": 352768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4215678052992635e-05,
+      "loss": 4.232,
+      "step": 353280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4207292105482115e-05,
+      "loss": 4.2364,
+      "step": 353792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4198906157971595e-05,
+      "loss": 4.2287,
+      "step": 354304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.419052021046108e-05,
+      "loss": 4.2188,
+      "step": 354816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.418213426295056e-05,
+      "loss": 4.2201,
+      "step": 355328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.417374831544004e-05,
+      "loss": 4.2284,
+      "step": 355840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.416536236792952e-05,
+      "loss": 4.2347,
+      "step": 356352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4156976420419e-05,
+      "loss": 4.2249,
+      "step": 356864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.414860685171221e-05,
+      "loss": 4.2233,
+      "step": 357376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.414023728300542e-05,
+      "loss": 4.2144,
+      "step": 357888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.41318513354949e-05,
+      "loss": 4.2224,
+      "step": 358400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.412346538798438e-05,
+      "loss": 4.2112,
+      "step": 358912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.411507944047386e-05,
+      "loss": 4.2228,
+      "step": 359424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.410670987176707e-05,
+      "loss": 4.205,
+      "step": 359936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.409832392425655e-05,
+      "loss": 4.2343,
+      "step": 360448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.408993797674603e-05,
+      "loss": 4.2191,
+      "step": 360960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4081552029235515e-05,
+      "loss": 4.2247,
+      "step": 361472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4073166081724995e-05,
+      "loss": 4.213,
+      "step": 361984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4064796513018204e-05,
+      "loss": 4.216,
+      "step": 362496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4056410565507684e-05,
+      "loss": 4.2194,
+      "step": 363008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4048024617997164e-05,
+      "loss": 4.2158,
+      "step": 363520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4039638670486644e-05,
+      "loss": 4.2234,
+      "step": 364032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4031252722976124e-05,
+      "loss": 4.2266,
+      "step": 364544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4022866775465604e-05,
+      "loss": 4.2216,
+      "step": 365056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4014480827955084e-05,
+      "loss": 4.2156,
+      "step": 365568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.400609488044456e-05,
+      "loss": 4.2192,
+      "step": 366080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.399770893293404e-05,
+      "loss": 4.2278,
+      "step": 366592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398933936422725e-05,
+      "loss": 4.203,
+      "step": 367104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.398095341671673e-05,
+      "loss": 4.2292,
+      "step": 367616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.397256746920621e-05,
+      "loss": 4.2138,
+      "step": 368128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.396419790049943e-05,
+      "loss": 4.2234,
+      "step": 368640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.395581195298891e-05,
+      "loss": 4.2116,
+      "step": 369152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.394742600547839e-05,
+      "loss": 4.215,
+      "step": 369664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.393904005796786e-05,
+      "loss": 4.2007,
+      "step": 370176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.393067048926108e-05,
+      "loss": 4.2112,
+      "step": 370688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.392228454175056e-05,
+      "loss": 4.2075,
+      "step": 371200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.391389859424003e-05,
+      "loss": 4.2145,
+      "step": 371712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.390551264672951e-05,
+      "loss": 4.2175,
+      "step": 372224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.389712669921899e-05,
+      "loss": 4.2218,
+      "step": 372736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388875713051221e-05,
+      "loss": 4.2039,
+      "step": 373248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388037118300169e-05,
+      "loss": 4.2151,
+      "step": 373760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.387198523549117e-05,
+      "loss": 4.2217,
+      "step": 374272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3863599287980647e-05,
+      "loss": 4.2194,
+      "step": 374784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.385522971927386e-05,
+      "loss": 4.2157,
+      "step": 375296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3846843771763336e-05,
+      "loss": 4.2186,
+      "step": 375808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3838457824252816e-05,
+      "loss": 4.2075,
+      "step": 376320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3830071876742296e-05,
+      "loss": 4.2204,
+      "step": 376832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3821685929231775e-05,
+      "loss": 4.2028,
+      "step": 377344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3813299981721255e-05,
+      "loss": 4.2169,
+      "step": 377856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3804914034210735e-05,
+      "loss": 4.2069,
+      "step": 378368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3796528086700215e-05,
+      "loss": 4.2088,
+      "step": 378880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3788158517993424e-05,
+      "loss": 4.2204,
+      "step": 379392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.3779772570482904e-05,
+      "loss": 4.2112,
+      "step": 379904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.377138662297239e-05,
+      "loss": 4.2026,
+      "step": 380416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.376300067546187e-05,
+      "loss": 4.1998,
+      "step": 380928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.375463110675508e-05,
+      "loss": 4.2067,
+      "step": 381440
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.20039701461792,
+      "eval_runtime": 287.3669,
+      "eval_samples_per_second": 1327.888,
+      "eval_steps_per_second": 41.497,
+      "step": 381600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.374624515924456e-05,
+      "loss": 4.2019,
+      "step": 381952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.373785921173404e-05,
+      "loss": 4.201,
+      "step": 382464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372947326422352e-05,
+      "loss": 4.2209,
+      "step": 382976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.372110369551673e-05,
+      "loss": 4.2066,
+      "step": 383488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.371271774800621e-05,
+      "loss": 4.2166,
+      "step": 384000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.370433180049569e-05,
+      "loss": 4.2034,
+      "step": 384512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.369594585298517e-05,
+      "loss": 4.2021,
+      "step": 385024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.368755990547465e-05,
+      "loss": 4.1975,
+      "step": 385536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3679206715571594e-05,
+      "loss": 4.2002,
+      "step": 386048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3670820768061074e-05,
+      "loss": 4.2113,
+      "step": 386560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.3662434820550554e-05,
+      "loss": 4.2104,
+      "step": 387072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.365406525184376e-05,
+      "loss": 4.2085,
+      "step": 387584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.364569568313698e-05,
+      "loss": 4.1986,
+      "step": 388096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.363730973562645e-05,
+      "loss": 4.1972,
+      "step": 388608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.362892378811593e-05,
+      "loss": 4.1959,
+      "step": 389120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.362053784060541e-05,
+      "loss": 4.194,
+      "step": 389632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.361215189309489e-05,
+      "loss": 4.2081,
+      "step": 390144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.360376594558437e-05,
+      "loss": 4.2025,
+      "step": 390656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.359537999807385e-05,
+      "loss": 4.2061,
+      "step": 391168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.358699405056333e-05,
+      "loss": 4.2175,
+      "step": 391680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.357860810305281e-05,
+      "loss": 4.21,
+      "step": 392192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.35702221555423e-05,
+      "loss": 4.1957,
+      "step": 392704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.356183620803178e-05,
+      "loss": 4.2022,
+      "step": 393216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.355345026052126e-05,
+      "loss": 4.2056,
+      "step": 393728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.354506431301074e-05,
+      "loss": 4.1962,
+      "step": 394240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.353667836550022e-05,
+      "loss": 4.2002,
+      "step": 394752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.352829241798969e-05,
+      "loss": 4.1965,
+      "step": 395264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.351990647047917e-05,
+      "loss": 4.1964,
+      "step": 395776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.351152052296865e-05,
+      "loss": 4.193,
+      "step": 396288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.350313457545813e-05,
+      "loss": 4.1885,
+      "step": 396800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.349474862794761e-05,
+      "loss": 4.1923,
+      "step": 397312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3486395438044556e-05,
+      "loss": 4.1992,
+      "step": 397824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3478009490534036e-05,
+      "loss": 4.2046,
+      "step": 398336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3469623543023516e-05,
+      "loss": 4.1992,
+      "step": 398848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3461237595512996e-05,
+      "loss": 4.1874,
+      "step": 399360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3452851648002476e-05,
+      "loss": 4.2039,
+      "step": 399872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3444465700491956e-05,
+      "loss": 4.1991,
+      "step": 400384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3436079752981436e-05,
+      "loss": 4.1751,
+      "step": 400896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3427710184274645e-05,
+      "loss": 4.191,
+      "step": 401408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3419324236764125e-05,
+      "loss": 4.1857,
+      "step": 401920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3410938289253605e-05,
+      "loss": 4.1852,
+      "step": 402432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3402552341743085e-05,
+      "loss": 4.2019,
+      "step": 402944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3394166394232565e-05,
+      "loss": 4.1846,
+      "step": 403456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3385780446722045e-05,
+      "loss": 4.189,
+      "step": 403968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3377394499211525e-05,
+      "loss": 4.1917,
+      "step": 404480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3369008551701005e-05,
+      "loss": 4.1904,
+      "step": 404992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3360622604190485e-05,
+      "loss": 4.1989,
+      "step": 405504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.335226941428743e-05,
+      "loss": 4.181,
+      "step": 406016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.334388346677691e-05,
+      "loss": 4.1602,
+      "step": 406528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.333549751926639e-05,
+      "loss": 4.2059,
+      "step": 407040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.332711157175587e-05,
+      "loss": 4.1842,
+      "step": 407552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331872562424535e-05,
+      "loss": 4.1923,
+      "step": 408064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.331033967673483e-05,
+      "loss": 4.1792,
+      "step": 408576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.330195372922431e-05,
+      "loss": 4.1799,
+      "step": 409088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.329356778171379e-05,
+      "loss": 4.1811,
+      "step": 409600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3285198213007e-05,
+      "loss": 4.1828,
+      "step": 410112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.327681226549648e-05,
+      "loss": 4.1889,
+      "step": 410624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326842631798596e-05,
+      "loss": 4.178,
+      "step": 411136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.326004037047544e-05,
+      "loss": 4.1968,
+      "step": 411648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.325165442296492e-05,
+      "loss": 4.1745,
+      "step": 412160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.32432684754544e-05,
+      "loss": 4.1648,
+      "step": 412672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3234882527943885e-05,
+      "loss": 4.1926,
+      "step": 413184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.322649658043336e-05,
+      "loss": 4.1646,
+      "step": 413696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3218127011726574e-05,
+      "loss": 4.1663,
+      "step": 414208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3209741064216054e-05,
+      "loss": 4.1875,
+      "step": 414720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.320135511670553e-05,
+      "loss": 4.1878,
+      "step": 415232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.319298554799874e-05,
+      "loss": 4.176,
+      "step": 415744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.318459960048822e-05,
+      "loss": 4.184,
+      "step": 416256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.31762136529777e-05,
+      "loss": 4.1624,
+      "step": 416768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3167827705467176e-05,
+      "loss": 4.1857,
+      "step": 417280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3159441757956656e-05,
+      "loss": 4.1771,
+      "step": 417792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3151055810446136e-05,
+      "loss": 4.1884,
+      "step": 418304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.314266986293562e-05,
+      "loss": 4.1801,
+      "step": 418816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.31342839154251e-05,
+      "loss": 4.1871,
+      "step": 419328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.312591434671831e-05,
+      "loss": 4.1941,
+      "step": 419840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.311752839920779e-05,
+      "loss": 4.1681,
+      "step": 420352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.310914245169727e-05,
+      "loss": 4.181,
+      "step": 420864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.310077288299048e-05,
+      "loss": 4.1848,
+      "step": 421376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.309238693547996e-05,
+      "loss": 4.1652,
+      "step": 421888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.308400098796944e-05,
+      "loss": 4.1883,
+      "step": 422400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.307561504045892e-05,
+      "loss": 4.1892,
+      "step": 422912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.30672290929484e-05,
+      "loss": 4.1844,
+      "step": 423424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.305884314543788e-05,
+      "loss": 4.1978,
+      "step": 423936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.305045719792736e-05,
+      "loss": 4.1671,
+      "step": 424448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.304207125041684e-05,
+      "loss": 4.1711,
+      "step": 424960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3033701681710057e-05,
+      "loss": 4.1799,
+      "step": 425472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3025315734199536e-05,
+      "loss": 4.1825,
+      "step": 425984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3016929786689016e-05,
+      "loss": 4.184,
+      "step": 426496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.3008543839178496e-05,
+      "loss": 4.1787,
+      "step": 427008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.3000157891667976e-05,
+      "loss": 4.1673,
+      "step": 427520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2991771944157456e-05,
+      "loss": 4.1836,
+      "step": 428032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2983385996646936e-05,
+      "loss": 4.1713,
+      "step": 428544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2975000049136416e-05,
+      "loss": 4.1687,
+      "step": 429056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2966630480429625e-05,
+      "loss": 4.176,
+      "step": 429568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2958244532919105e-05,
+      "loss": 4.1859,
+      "step": 430080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2949858585408585e-05,
+      "loss": 4.1745,
+      "step": 430592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2941472637898065e-05,
+      "loss": 4.1632,
+      "step": 431104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2933103069191274e-05,
+      "loss": 4.1672,
+      "step": 431616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2924717121680754e-05,
+      "loss": 4.176,
+      "step": 432128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.291633117417024e-05,
+      "loss": 4.1785,
+      "step": 432640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.290796160546345e-05,
+      "loss": 4.173,
+      "step": 433152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.289957565795293e-05,
+      "loss": 4.1709,
+      "step": 433664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.289118971044241e-05,
+      "loss": 4.1597,
+      "step": 434176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.288280376293189e-05,
+      "loss": 4.1667,
+      "step": 434688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.287441781542136e-05,
+      "loss": 4.163,
+      "step": 435200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.286603186791084e-05,
+      "loss": 4.1677,
+      "step": 435712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.285766229920406e-05,
+      "loss": 4.1537,
+      "step": 436224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.284927635169354e-05,
+      "loss": 4.183,
+      "step": 436736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.284089040418301e-05,
+      "loss": 4.1664,
+      "step": 437248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.283250445667249e-05,
+      "loss": 4.1711,
+      "step": 437760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.282411850916198e-05,
+      "loss": 4.1576,
+      "step": 438272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.281573256165146e-05,
+      "loss": 4.1717,
+      "step": 438784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.280734661414094e-05,
+      "loss": 4.1642,
+      "step": 439296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.279896066663042e-05,
+      "loss": 4.1621,
+      "step": 439808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.279059109792363e-05,
+      "loss": 4.1749,
+      "step": 440320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.278220515041311e-05,
+      "loss": 4.1786,
+      "step": 440832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.277383558170632e-05,
+      "loss": 4.1625,
+      "step": 441344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.27654496341958e-05,
+      "loss": 4.1747,
+      "step": 441856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.275706368668528e-05,
+      "loss": 4.161,
+      "step": 442368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.274867773917476e-05,
+      "loss": 4.174,
+      "step": 442880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2740291791664237e-05,
+      "loss": 4.1572,
+      "step": 443392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2731905844153717e-05,
+      "loss": 4.1773,
+      "step": 443904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2723519896643196e-05,
+      "loss": 4.16,
+      "step": 444416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2715133949132676e-05,
+      "loss": 4.1731,
+      "step": 444928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.270674800162216e-05,
+      "loss": 4.1641,
+      "step": 445440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.269837843291537e-05,
+      "loss": 4.1624,
+      "step": 445952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.268999248540485e-05,
+      "loss": 4.1465,
+      "step": 446464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.268162291669806e-05,
+      "loss": 4.1641,
+      "step": 446976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.267323696918754e-05,
+      "loss": 4.157,
+      "step": 447488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.266485102167702e-05,
+      "loss": 4.1653,
+      "step": 448000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.26564650741665e-05,
+      "loss": 4.1617,
+      "step": 448512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.264809550545971e-05,
+      "loss": 4.1723,
+      "step": 449024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.263970955794919e-05,
+      "loss": 4.1555,
+      "step": 449536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.263132361043867e-05,
+      "loss": 4.1657,
+      "step": 450048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.262293766292815e-05,
+      "loss": 4.1696,
+      "step": 450560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.261455171541763e-05,
+      "loss": 4.1716,
+      "step": 451072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.260616576790712e-05,
+      "loss": 4.1661,
+      "step": 451584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.25977798203966e-05,
+      "loss": 4.1658,
+      "step": 452096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258939387288608e-05,
+      "loss": 4.1558,
+      "step": 452608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.258100792537555e-05,
+      "loss": 4.1739,
+      "step": 453120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2572638356668766e-05,
+      "loss": 4.1532,
+      "step": 453632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2564252409158246e-05,
+      "loss": 4.1681,
+      "step": 454144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2555866461647726e-05,
+      "loss": 4.1565,
+      "step": 454656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.25474805141372e-05,
+      "loss": 4.1607,
+      "step": 455168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2539110945430415e-05,
+      "loss": 4.1688,
+      "step": 455680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2530724997919895e-05,
+      "loss": 4.1577,
+      "step": 456192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2522339050409375e-05,
+      "loss": 4.16,
+      "step": 456704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.2513969481702584e-05,
+      "loss": 4.1485,
+      "step": 457216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.250558353419207e-05,
+      "loss": 4.1598,
+      "step": 457728
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.159285068511963,
+      "eval_runtime": 289.1843,
+      "eval_samples_per_second": 1319.543,
+      "eval_steps_per_second": 41.237,
+      "step": 457920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.249719758668155e-05,
+      "loss": 4.1539,
+      "step": 458240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2488811639171024e-05,
+      "loss": 4.1484,
+      "step": 458752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2480425691660504e-05,
+      "loss": 4.1687,
+      "step": 459264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2472039744149984e-05,
+      "loss": 4.16,
+      "step": 459776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2463653796639464e-05,
+      "loss": 4.1655,
+      "step": 460288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2455267849128943e-05,
+      "loss": 4.1538,
+      "step": 460800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2446881901618423e-05,
+      "loss": 4.1522,
+      "step": 461312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.24384959541079e-05,
+      "loss": 4.1495,
+      "step": 461824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.243011000659738e-05,
+      "loss": 4.1533,
+      "step": 462336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.242172405908686e-05,
+      "loss": 4.1632,
+      "step": 462848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.241333811157634e-05,
+      "loss": 4.158,
+      "step": 463360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.240495216406582e-05,
+      "loss": 4.1601,
+      "step": 463872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.239658259535903e-05,
+      "loss": 4.1511,
+      "step": 464384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.238819664784852e-05,
+      "loss": 4.1516,
+      "step": 464896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2379810700338e-05,
+      "loss": 4.1481,
+      "step": 465408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.237142475282748e-05,
+      "loss": 4.1432,
+      "step": 465920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.236303880531696e-05,
+      "loss": 4.1597,
+      "step": 466432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.235465285780644e-05,
+      "loss": 4.156,
+      "step": 466944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.234626691029591e-05,
+      "loss": 4.1557,
+      "step": 467456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.233788096278539e-05,
+      "loss": 4.1683,
+      "step": 467968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.232951139407861e-05,
+      "loss": 4.168,
+      "step": 468480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.232112544656809e-05,
+      "loss": 4.1523,
+      "step": 468992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.231273949905756e-05,
+      "loss": 4.1516,
+      "step": 469504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.230435355154704e-05,
+      "loss": 4.1576,
+      "step": 470016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.229596760403652e-05,
+      "loss": 4.1535,
+      "step": 470528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2287581656526e-05,
+      "loss": 4.1519,
+      "step": 471040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.227919570901549e-05,
+      "loss": 4.1515,
+      "step": 471552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.227080976150497e-05,
+      "loss": 4.1494,
+      "step": 472064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.226244019279818e-05,
+      "loss": 4.1465,
+      "step": 472576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2254054245287657e-05,
+      "loss": 4.1444,
+      "step": 473088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2245668297777137e-05,
+      "loss": 4.1412,
+      "step": 473600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2237282350266617e-05,
+      "loss": 4.1514,
+      "step": 474112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2228912781559826e-05,
+      "loss": 4.1589,
+      "step": 474624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2220526834049306e-05,
+      "loss": 4.1569,
+      "step": 475136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2212140886538786e-05,
+      "loss": 4.1424,
+      "step": 475648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2203771317831995e-05,
+      "loss": 4.1531,
+      "step": 476160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.219540174912521e-05,
+      "loss": 4.1531,
+      "step": 476672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.218701580161469e-05,
+      "loss": 4.1315,
+      "step": 477184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217862985410417e-05,
+      "loss": 4.1445,
+      "step": 477696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.217024390659365e-05,
+      "loss": 4.1407,
+      "step": 478208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.216185795908313e-05,
+      "loss": 4.1363,
+      "step": 478720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.215347201157261e-05,
+      "loss": 4.1532,
+      "step": 479232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.214508606406209e-05,
+      "loss": 4.1402,
+      "step": 479744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.213670011655157e-05,
+      "loss": 4.1476,
+      "step": 480256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.212831416904105e-05,
+      "loss": 4.1442,
+      "step": 480768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.211992822153053e-05,
+      "loss": 4.1439,
+      "step": 481280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.211154227402001e-05,
+      "loss": 4.1541,
+      "step": 481792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210315632650949e-05,
+      "loss": 4.1382,
+      "step": 482304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.209477037899897e-05,
+      "loss": 4.1126,
+      "step": 482816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.208640081029218e-05,
+      "loss": 4.1588,
+      "step": 483328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.207801486278166e-05,
+      "loss": 4.1405,
+      "step": 483840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.206962891527114e-05,
+      "loss": 4.1498,
+      "step": 484352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2061242967760626e-05,
+      "loss": 4.136,
+      "step": 484864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2052873399053835e-05,
+      "loss": 4.1371,
+      "step": 485376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2044487451543315e-05,
+      "loss": 4.1336,
+      "step": 485888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2036117882836524e-05,
+      "loss": 4.1398,
+      "step": 486400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2027731935326004e-05,
+      "loss": 4.1469,
+      "step": 486912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2019345987815484e-05,
+      "loss": 4.13,
+      "step": 487424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2010960040304964e-05,
+      "loss": 4.1511,
+      "step": 487936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2002574092794444e-05,
+      "loss": 4.1338,
+      "step": 488448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1994188145283924e-05,
+      "loss": 4.1213,
+      "step": 488960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19858021977734e-05,
+      "loss": 4.143,
+      "step": 489472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.197741625026288e-05,
+      "loss": 4.1266,
+      "step": 489984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196904668155609e-05,
+      "loss": 4.1192,
+      "step": 490496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.196066073404557e-05,
+      "loss": 4.1451,
+      "step": 491008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.195229116533879e-05,
+      "loss": 4.1456,
+      "step": 491520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.194390521782827e-05,
+      "loss": 4.1323,
+      "step": 492032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.193551927031775e-05,
+      "loss": 4.1372,
+      "step": 492544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.192713332280722e-05,
+      "loss": 4.1225,
+      "step": 493056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.19187473752967e-05,
+      "loss": 4.1384,
+      "step": 493568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.191036142778618e-05,
+      "loss": 4.1381,
+      "step": 494080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.190197548027566e-05,
+      "loss": 4.1402,
+      "step": 494592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189360591156887e-05,
+      "loss": 4.1391,
+      "step": 495104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.188521996405835e-05,
+      "loss": 4.1419,
+      "step": 495616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.187683401654783e-05,
+      "loss": 4.153,
+      "step": 496128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.186844806903731e-05,
+      "loss": 4.1256,
+      "step": 496640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.18600621215268e-05,
+      "loss": 4.1419,
+      "step": 497152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1851692552820006e-05,
+      "loss": 4.1364,
+      "step": 497664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1843306605309486e-05,
+      "loss": 4.1259,
+      "step": 498176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1834920657798966e-05,
+      "loss": 4.1444,
+      "step": 498688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1826534710288446e-05,
+      "loss": 4.149,
+      "step": 499200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1818148762777926e-05,
+      "loss": 4.1405,
+      "step": 499712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1809762815267406e-05,
+      "loss": 4.1553,
+      "step": 500224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1801393246560615e-05,
+      "loss": 4.1234,
+      "step": 500736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1793007299050095e-05,
+      "loss": 4.1284,
+      "step": 501248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1784621351539575e-05,
+      "loss": 4.137,
+      "step": 501760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1776235404029055e-05,
+      "loss": 4.143,
+      "step": 502272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1767849456518535e-05,
+      "loss": 4.1342,
+      "step": 502784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.175947988781175e-05,
+      "loss": 4.1391,
+      "step": 503296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.175109394030123e-05,
+      "loss": 4.1301,
+      "step": 503808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.174270799279071e-05,
+      "loss": 4.1401,
+      "step": 504320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.173432204528019e-05,
+      "loss": 4.1285,
+      "step": 504832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.172593609776967e-05,
+      "loss": 4.126,
+      "step": 505344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.171755015025915e-05,
+      "loss": 4.1358,
+      "step": 505856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.170916420274863e-05,
+      "loss": 4.1451,
+      "step": 506368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.170077825523811e-05,
+      "loss": 4.1308,
+      "step": 506880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1692392307727584e-05,
+      "loss": 4.1186,
+      "step": 507392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.16840227390208e-05,
+      "loss": 4.1304,
+      "step": 507904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.167563679151028e-05,
+      "loss": 4.1308,
+      "step": 508416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.166725084399975e-05,
+      "loss": 4.1393,
+      "step": 508928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.165886489648923e-05,
+      "loss": 4.1294,
+      "step": 509440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.165049532778245e-05,
+      "loss": 4.1287,
+      "step": 509952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1642109380271935e-05,
+      "loss": 4.1229,
+      "step": 510464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.163372343276141e-05,
+      "loss": 4.1204,
+      "step": 510976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.162533748525089e-05,
+      "loss": 4.13,
+      "step": 511488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.161695153774037e-05,
+      "loss": 4.124,
+      "step": 512000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1608581969033584e-05,
+      "loss": 4.1123,
+      "step": 512512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.160019602152306e-05,
+      "loss": 4.1425,
+      "step": 513024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.159181007401254e-05,
+      "loss": 4.1241,
+      "step": 513536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.158342412650202e-05,
+      "loss": 4.1266,
+      "step": 514048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.15750381789915e-05,
+      "loss": 4.1188,
+      "step": 514560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.156665223148098e-05,
+      "loss": 4.1297,
+      "step": 515072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.155826628397046e-05,
+      "loss": 4.1271,
+      "step": 515584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.154988033645994e-05,
+      "loss": 4.1203,
+      "step": 516096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.154151076775315e-05,
+      "loss": 4.1343,
+      "step": 516608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.153312482024263e-05,
+      "loss": 4.1337,
+      "step": 517120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.152473887273211e-05,
+      "loss": 4.1263,
+      "step": 517632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.151635292522159e-05,
+      "loss": 4.1324,
+      "step": 518144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.15079833565148e-05,
+      "loss": 4.1189,
+      "step": 518656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.149959740900428e-05,
+      "loss": 4.1328,
+      "step": 519168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.149121146149376e-05,
+      "loss": 4.1223,
+      "step": 519680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.148282551398324e-05,
+      "loss": 4.1339,
+      "step": 520192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.147443956647272e-05,
+      "loss": 4.1208,
+      "step": 520704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.146606999776593e-05,
+      "loss": 4.1309,
+      "step": 521216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.145768405025541e-05,
+      "loss": 4.1282,
+      "step": 521728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144929810274489e-05,
+      "loss": 4.1222,
+      "step": 522240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.144091215523437e-05,
+      "loss": 4.1088,
+      "step": 522752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1432558965331316e-05,
+      "loss": 4.1216,
+      "step": 523264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1424173017820796e-05,
+      "loss": 4.1155,
+      "step": 523776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1415787070310276e-05,
+      "loss": 4.1314,
+      "step": 524288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1407401122799756e-05,
+      "loss": 4.1205,
+      "step": 524800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1399015175289236e-05,
+      "loss": 4.1331,
+      "step": 525312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1390629227778716e-05,
+      "loss": 4.1123,
+      "step": 525824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1382243280268195e-05,
+      "loss": 4.1292,
+      "step": 526336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1373873711561405e-05,
+      "loss": 4.129,
+      "step": 526848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1365487764050885e-05,
+      "loss": 4.131,
+      "step": 527360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1357101816540364e-05,
+      "loss": 4.128,
+      "step": 527872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1348715869029844e-05,
+      "loss": 4.1266,
+      "step": 528384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1340329921519324e-05,
+      "loss": 4.1174,
+      "step": 528896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1331943974008804e-05,
+      "loss": 4.1352,
+      "step": 529408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.132355802649829e-05,
+      "loss": 4.1116,
+      "step": 529920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.131517207898777e-05,
+      "loss": 4.129,
+      "step": 530432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.130680251028098e-05,
+      "loss": 4.119,
+      "step": 530944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.129841656277046e-05,
+      "loss": 4.1265,
+      "step": 531456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.129004699406367e-05,
+      "loss": 4.1225,
+      "step": 531968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.128166104655315e-05,
+      "loss": 4.1189,
+      "step": 532480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.127327509904263e-05,
+      "loss": 4.1213,
+      "step": 532992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.126490553033584e-05,
+      "loss": 4.1127,
+      "step": 533504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.125651958282532e-05,
+      "loss": 4.1222,
+      "step": 534016
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.128636360168457,
+      "eval_runtime": 290.8736,
+      "eval_samples_per_second": 1311.879,
+      "eval_steps_per_second": 40.997,
+      "step": 534240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.12481336353148e-05,
+      "loss": 4.1144,
+      "step": 534528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.123974768780428e-05,
+      "loss": 4.1081,
+      "step": 535040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.123136174029376e-05,
+      "loss": 4.1287,
+      "step": 535552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1222975792783245e-05,
+      "loss": 4.1206,
+      "step": 536064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.121458984527272e-05,
+      "loss": 4.1267,
+      "step": 536576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.12062038977622e-05,
+      "loss": 4.1188,
+      "step": 537088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.119781795025168e-05,
+      "loss": 4.1158,
+      "step": 537600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118944838154489e-05,
+      "loss": 4.1128,
+      "step": 538112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.118106243403437e-05,
+      "loss": 4.1144,
+      "step": 538624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.117267648652385e-05,
+      "loss": 4.1256,
+      "step": 539136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.116429053901333e-05,
+      "loss": 4.1172,
+      "step": 539648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1155920970306536e-05,
+      "loss": 4.1214,
+      "step": 540160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.114755140159975e-05,
+      "loss": 4.117,
+      "step": 540672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.113916545408923e-05,
+      "loss": 4.1111,
+      "step": 541184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.113077950657871e-05,
+      "loss": 4.1052,
+      "step": 541696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.112239355906819e-05,
+      "loss": 4.1061,
+      "step": 542208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.111400761155767e-05,
+      "loss": 4.1241,
+      "step": 542720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.110562166404715e-05,
+      "loss": 4.1189,
+      "step": 543232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.109723571653663e-05,
+      "loss": 4.1121,
+      "step": 543744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.108884976902611e-05,
+      "loss": 4.1316,
+      "step": 544256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.108046382151559e-05,
+      "loss": 4.1281,
+      "step": 544768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.107207787400507e-05,
+      "loss": 4.1161,
+      "step": 545280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.106369192649455e-05,
+      "loss": 4.1174,
+      "step": 545792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.105530597898403e-05,
+      "loss": 4.1114,
+      "step": 546304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.104693641027724e-05,
+      "loss": 4.1179,
+      "step": 546816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.103855046276672e-05,
+      "loss": 4.1147,
+      "step": 547328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.10301645152562e-05,
+      "loss": 4.1142,
+      "step": 547840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.102177856774568e-05,
+      "loss": 4.116,
+      "step": 548352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1013408999038896e-05,
+      "loss": 4.107,
+      "step": 548864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1005023051528376e-05,
+      "loss": 4.1076,
+      "step": 549376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0996637104017856e-05,
+      "loss": 4.0984,
+      "step": 549888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0988251156507336e-05,
+      "loss": 4.1189,
+      "step": 550400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0979881587800545e-05,
+      "loss": 4.1207,
+      "step": 550912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0971495640290025e-05,
+      "loss": 4.1193,
+      "step": 551424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0963109692779505e-05,
+      "loss": 4.11,
+      "step": 551936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0954740124072714e-05,
+      "loss": 4.1067,
+      "step": 552448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0946354176562194e-05,
+      "loss": 4.1177,
+      "step": 552960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0937968229051674e-05,
+      "loss": 4.1012,
+      "step": 553472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0929582281541154e-05,
+      "loss": 4.101,
+      "step": 553984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0921196334030634e-05,
+      "loss": 4.1058,
+      "step": 554496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.091282676532385e-05,
+      "loss": 4.1018,
+      "step": 555008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.090444081781333e-05,
+      "loss": 4.1124,
+      "step": 555520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.089605487030281e-05,
+      "loss": 4.1034,
+      "step": 556032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.088766892279229e-05,
+      "loss": 4.1125,
+      "step": 556544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.087928297528177e-05,
+      "loss": 4.1062,
+      "step": 557056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.087089702777125e-05,
+      "loss": 4.1092,
+      "step": 557568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.086251108026072e-05,
+      "loss": 4.116,
+      "step": 558080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.08541251327502e-05,
+      "loss": 4.1038,
+      "step": 558592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.084575556404342e-05,
+      "loss": 4.0754,
+      "step": 559104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.083738599533663e-05,
+      "loss": 4.1204,
+      "step": 559616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.082900004782611e-05,
+      "loss": 4.1036,
+      "step": 560128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.082061410031559e-05,
+      "loss": 4.1139,
+      "step": 560640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0812228152805074e-05,
+      "loss": 4.0994,
+      "step": 561152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0803842205294554e-05,
+      "loss": 4.1024,
+      "step": 561664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.079545625778403e-05,
+      "loss": 4.0926,
+      "step": 562176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.078708668907724e-05,
+      "loss": 4.1031,
+      "step": 562688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.077870074156672e-05,
+      "loss": 4.1122,
+      "step": 563200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0770314794056196e-05,
+      "loss": 4.0904,
+      "step": 563712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0761928846545676e-05,
+      "loss": 4.1204,
+      "step": 564224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0753542899035156e-05,
+      "loss": 4.0989,
+      "step": 564736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0745156951524636e-05,
+      "loss": 4.0869,
+      "step": 565248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0736771004014116e-05,
+      "loss": 4.1037,
+      "step": 565760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0728385056503596e-05,
+      "loss": 4.0903,
+      "step": 566272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.072001548779681e-05,
+      "loss": 4.0857,
+      "step": 566784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.071162954028629e-05,
+      "loss": 4.11,
+      "step": 567296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.070324359277577e-05,
+      "loss": 4.1084,
+      "step": 567808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.069487402406898e-05,
+      "loss": 4.0975,
+      "step": 568320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.068648807655846e-05,
+      "loss": 4.1038,
+      "step": 568832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.067810212904794e-05,
+      "loss": 4.0835,
+      "step": 569344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.066971618153742e-05,
+      "loss": 4.1048,
+      "step": 569856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.06613302340269e-05,
+      "loss": 4.0975,
+      "step": 570368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.065294428651638e-05,
+      "loss": 4.1037,
+      "step": 570880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.064455833900586e-05,
+      "loss": 4.1073,
+      "step": 571392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.063617239149534e-05,
+      "loss": 4.1029,
+      "step": 571904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.062778644398482e-05,
+      "loss": 4.1187,
+      "step": 572416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0619433254081766e-05,
+      "loss": 4.0892,
+      "step": 572928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0611047306571246e-05,
+      "loss": 4.1044,
+      "step": 573440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0602661359060726e-05,
+      "loss": 4.1011,
+      "step": 573952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0594275411550206e-05,
+      "loss": 4.0924,
+      "step": 574464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0585889464039686e-05,
+      "loss": 4.1063,
+      "step": 574976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0577503516529166e-05,
+      "loss": 4.1116,
+      "step": 575488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0569133947822375e-05,
+      "loss": 4.1047,
+      "step": 576000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0560748000311855e-05,
+      "loss": 4.1211,
+      "step": 576512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0552362052801335e-05,
+      "loss": 4.0907,
+      "step": 577024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0543976105290814e-05,
+      "loss": 4.0934,
+      "step": 577536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0535590157780294e-05,
+      "loss": 4.1002,
+      "step": 578048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0527204210269774e-05,
+      "loss": 4.1083,
+      "step": 578560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0518818262759254e-05,
+      "loss": 4.0972,
+      "step": 579072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.0510448694052463e-05,
+      "loss": 4.1061,
+      "step": 579584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0502062746541943e-05,
+      "loss": 4.0959,
+      "step": 580096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.049367679903143e-05,
+      "loss": 4.1024,
+      "step": 580608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.048529085152091e-05,
+      "loss": 4.097,
+      "step": 581120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.047690490401038e-05,
+      "loss": 4.0893,
+      "step": 581632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.04685353353036e-05,
+      "loss": 4.101,
+      "step": 582144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.046014938779308e-05,
+      "loss": 4.1102,
+      "step": 582656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.045176344028256e-05,
+      "loss": 4.0978,
+      "step": 583168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.044337749277203e-05,
+      "loss": 4.083,
+      "step": 583680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.043499154526151e-05,
+      "loss": 4.0945,
+      "step": 584192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.042660559775099e-05,
+      "loss": 4.0939,
+      "step": 584704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.041821965024047e-05,
+      "loss": 4.1065,
+      "step": 585216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.040983370272995e-05,
+      "loss": 4.1003,
+      "step": 585728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.040144775521943e-05,
+      "loss": 4.0902,
+      "step": 586240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.039307818651265e-05,
+      "loss": 4.0925,
+      "step": 586752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.038469223900213e-05,
+      "loss": 4.0845,
+      "step": 587264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.037630629149161e-05,
+      "loss": 4.0941,
+      "step": 587776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.036792034398109e-05,
+      "loss": 4.0894,
+      "step": 588288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.035956715407803e-05,
+      "loss": 4.0807,
+      "step": 588800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0351181206567506e-05,
+      "loss": 4.104,
+      "step": 589312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0342795259056986e-05,
+      "loss": 4.0956,
+      "step": 589824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0334409311546466e-05,
+      "loss": 4.0888,
+      "step": 590336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0326023364035946e-05,
+      "loss": 4.088,
+      "step": 590848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0317637416525426e-05,
+      "loss": 4.0918,
+      "step": 591360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0309267847818635e-05,
+      "loss": 4.0948,
+      "step": 591872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.030088190030812e-05,
+      "loss": 4.0872,
+      "step": 592384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.02924959527976e-05,
+      "loss": 4.1026,
+      "step": 592896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.028411000528708e-05,
+      "loss": 4.1,
+      "step": 593408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.027574043658029e-05,
+      "loss": 4.0904,
+      "step": 593920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.026735448906977e-05,
+      "loss": 4.0993,
+      "step": 594432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.025898492036298e-05,
+      "loss": 4.0875,
+      "step": 594944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.025059897285246e-05,
+      "loss": 4.1023,
+      "step": 595456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.024221302534194e-05,
+      "loss": 4.0896,
+      "step": 595968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.023382707783142e-05,
+      "loss": 4.0947,
+      "step": 596480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.02254411303209e-05,
+      "loss": 4.0898,
+      "step": 596992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.021705518281038e-05,
+      "loss": 4.0999,
+      "step": 597504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.020866923529986e-05,
+      "loss": 4.0972,
+      "step": 598016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.020028328778934e-05,
+      "loss": 4.088,
+      "step": 598528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.019189734027882e-05,
+      "loss": 4.071,
+      "step": 599040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0183544150375764e-05,
+      "loss": 4.094,
+      "step": 599552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0175158202865244e-05,
+      "loss": 4.082,
+      "step": 600064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0166772255354724e-05,
+      "loss": 4.0935,
+      "step": 600576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0158386307844204e-05,
+      "loss": 4.0894,
+      "step": 601088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0150000360333684e-05,
+      "loss": 4.097,
+      "step": 601600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0141614412823164e-05,
+      "loss": 4.0785,
+      "step": 602112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0133228465312644e-05,
+      "loss": 4.0999,
+      "step": 602624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0124842517802124e-05,
+      "loss": 4.0945,
+      "step": 603136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0116456570291604e-05,
+      "loss": 4.1018,
+      "step": 603648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0108070622781084e-05,
+      "loss": 4.0957,
+      "step": 604160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0099684675270564e-05,
+      "loss": 4.0902,
+      "step": 604672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0091298727760044e-05,
+      "loss": 4.0889,
+      "step": 605184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0082912780249524e-05,
+      "loss": 4.1019,
+      "step": 605696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.0074526832739004e-05,
+      "loss": 4.0794,
+      "step": 606208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.006615726403222e-05,
+      "loss": 4.0956,
+      "step": 606720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.005777131652169e-05,
+      "loss": 4.091,
+      "step": 607232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.004938536901117e-05,
+      "loss": 4.0919,
+      "step": 607744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.004101580030439e-05,
+      "loss": 4.0901,
+      "step": 608256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.003262985279387e-05,
+      "loss": 4.0889,
+      "step": 608768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.002424390528334e-05,
+      "loss": 4.0879,
+      "step": 609280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.001585795777282e-05,
+      "loss": 4.0812,
+      "step": 609792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.00074720102623e-05,
+      "loss": 4.0884,
+      "step": 610304
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.104405879974365,
+      "eval_runtime": 294.0861,
+      "eval_samples_per_second": 1297.549,
+      "eval_steps_per_second": 40.549,
+      "step": 610560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999908606275178e-05,
+      "loss": 4.0782,
+      "step": 610816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.999070011524126e-05,
+      "loss": 4.0787,
+      "step": 611328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.998231416773074e-05,
+      "loss": 4.0935,
+      "step": 611840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.997392822022022e-05,
+      "loss": 4.0894,
+      "step": 612352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.996554227270971e-05,
+      "loss": 4.099,
+      "step": 612864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.995715632519919e-05,
+      "loss": 4.0823,
+      "step": 613376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994877037768867e-05,
+      "loss": 4.0863,
+      "step": 613888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.994038443017815e-05,
+      "loss": 4.0836,
+      "step": 614400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.993199848266763e-05,
+      "loss": 4.0815,
+      "step": 614912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.992361253515711e-05,
+      "loss": 4.0967,
+      "step": 615424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.991522658764658e-05,
+      "loss": 4.086,
+      "step": 615936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.990684064013606e-05,
+      "loss": 4.0897,
+      "step": 616448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989847107142928e-05,
+      "loss": 4.094,
+      "step": 616960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.989008512391876e-05,
+      "loss": 4.0722,
+      "step": 617472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.988169917640823e-05,
+      "loss": 4.0748,
+      "step": 617984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.987331322889771e-05,
+      "loss": 4.0731,
+      "step": 618496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.986492728138719e-05,
+      "loss": 4.0952,
+      "step": 619008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.985654133387668e-05,
+      "loss": 4.0858,
+      "step": 619520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9848171765169886e-05,
+      "loss": 4.0831,
+      "step": 620032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9839785817659366e-05,
+      "loss": 4.0964,
+      "step": 620544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9831399870148846e-05,
+      "loss": 4.0986,
+      "step": 621056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9823013922638326e-05,
+      "loss": 4.0823,
+      "step": 621568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9814627975127806e-05,
+      "loss": 4.0895,
+      "step": 622080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9806242027617286e-05,
+      "loss": 4.0835,
+      "step": 622592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9797856080106766e-05,
+      "loss": 4.0871,
+      "step": 623104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9789470132596246e-05,
+      "loss": 4.0809,
+      "step": 623616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9781084185085726e-05,
+      "loss": 4.0792,
+      "step": 624128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9772698237575205e-05,
+      "loss": 4.0881,
+      "step": 624640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9764312290064685e-05,
+      "loss": 4.0763,
+      "step": 625152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9755942721357895e-05,
+      "loss": 4.0754,
+      "step": 625664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9747556773847374e-05,
+      "loss": 4.0741,
+      "step": 626176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.973917082633686e-05,
+      "loss": 4.0793,
+      "step": 626688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.973078487882634e-05,
+      "loss": 4.0872,
+      "step": 627200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.972239893131582e-05,
+      "loss": 4.0946,
+      "step": 627712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9714012983805294e-05,
+      "loss": 4.0778,
+      "step": 628224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.970564341509851e-05,
+      "loss": 4.0792,
+      "step": 628736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.969725746758799e-05,
+      "loss": 4.0895,
+      "step": 629248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968887152007747e-05,
+      "loss": 4.0665,
+      "step": 629760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.968048557256694e-05,
+      "loss": 4.0721,
+      "step": 630272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.967209962505642e-05,
+      "loss": 4.0775,
+      "step": 630784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96637136775459e-05,
+      "loss": 4.0678,
+      "step": 631296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.965534410883912e-05,
+      "loss": 4.0784,
+      "step": 631808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.96469581613286e-05,
+      "loss": 4.0759,
+      "step": 632320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963857221381808e-05,
+      "loss": 4.0835,
+      "step": 632832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.963018626630756e-05,
+      "loss": 4.0716,
+      "step": 633344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.962180031879704e-05,
+      "loss": 4.0825,
+      "step": 633856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.961341437128652e-05,
+      "loss": 4.0844,
+      "step": 634368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9605028423776e-05,
+      "loss": 4.0762,
+      "step": 634880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.959665885506921e-05,
+      "loss": 4.0483,
+      "step": 635392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.958827290755869e-05,
+      "loss": 4.087,
+      "step": 635904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.957988696004817e-05,
+      "loss": 4.0733,
+      "step": 636416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.957150101253765e-05,
+      "loss": 4.0821,
+      "step": 636928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.956311506502713e-05,
+      "loss": 4.0716,
+      "step": 637440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.955474549632034e-05,
+      "loss": 4.0687,
+      "step": 637952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.954635954880982e-05,
+      "loss": 4.0686,
+      "step": 638464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.95379736012993e-05,
+      "loss": 4.0714,
+      "step": 638976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.952958765378878e-05,
+      "loss": 4.0849,
+      "step": 639488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9521201706278263e-05,
+      "loss": 4.0581,
+      "step": 640000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.951283213757147e-05,
+      "loss": 4.0917,
+      "step": 640512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.950446256886468e-05,
+      "loss": 4.0702,
+      "step": 641024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.949607662135416e-05,
+      "loss": 4.0534,
+      "step": 641536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.948769067384364e-05,
+      "loss": 4.0749,
+      "step": 642048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.947930472633312e-05,
+      "loss": 4.0626,
+      "step": 642560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.94709187788226e-05,
+      "loss": 4.0592,
+      "step": 643072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.946253283131208e-05,
+      "loss": 4.0753,
+      "step": 643584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.945414688380156e-05,
+      "loss": 4.0812,
+      "step": 644096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.944576093629104e-05,
+      "loss": 4.0642,
+      "step": 644608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.943737498878052e-05,
+      "loss": 4.0798,
+      "step": 645120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942898904127e-05,
+      "loss": 4.0536,
+      "step": 645632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.942060309375948e-05,
+      "loss": 4.0725,
+      "step": 646144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.941221714624896e-05,
+      "loss": 4.0739,
+      "step": 646656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.940384757754218e-05,
+      "loss": 4.0725,
+      "step": 647168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.939546163003166e-05,
+      "loss": 4.0791,
+      "step": 647680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.938707568252113e-05,
+      "loss": 4.0737,
+      "step": 648192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.937868973501061e-05,
+      "loss": 4.0931,
+      "step": 648704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9370336545107555e-05,
+      "loss": 4.0602,
+      "step": 649216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9361950597597035e-05,
+      "loss": 4.076,
+      "step": 649728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9353564650086515e-05,
+      "loss": 4.0676,
+      "step": 650240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9345178702575995e-05,
+      "loss": 4.0645,
+      "step": 650752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9336792755065475e-05,
+      "loss": 4.0801,
+      "step": 651264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9328423186358684e-05,
+      "loss": 4.0775,
+      "step": 651776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.932003723884817e-05,
+      "loss": 4.0809,
+      "step": 652288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.931165129133765e-05,
+      "loss": 4.0902,
+      "step": 652800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.930326534382713e-05,
+      "loss": 4.0642,
+      "step": 653312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9294879396316604e-05,
+      "loss": 4.0618,
+      "step": 653824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9286493448806084e-05,
+      "loss": 4.0689,
+      "step": 654336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9278107501295564e-05,
+      "loss": 4.0816,
+      "step": 654848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9269721553785044e-05,
+      "loss": 4.0652,
+      "step": 655360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.926135198507825e-05,
+      "loss": 4.0761,
+      "step": 655872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.925296603756773e-05,
+      "loss": 4.0719,
+      "step": 656384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.924458009005721e-05,
+      "loss": 4.0694,
+      "step": 656896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.923619414254669e-05,
+      "loss": 4.0719,
+      "step": 657408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.922782457383991e-05,
+      "loss": 4.0592,
+      "step": 657920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.921943862632939e-05,
+      "loss": 4.074,
+      "step": 658432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9211069057622604e-05,
+      "loss": 4.0807,
+      "step": 658944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.920268311011208e-05,
+      "loss": 4.0716,
+      "step": 659456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.919429716260156e-05,
+      "loss": 4.059,
+      "step": 659968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.918591121509104e-05,
+      "loss": 4.065,
+      "step": 660480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.917752526758052e-05,
+      "loss": 4.064,
+      "step": 660992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.916913932007e-05,
+      "loss": 4.0758,
+      "step": 661504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.916075337255948e-05,
+      "loss": 4.0751,
+      "step": 662016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.915236742504896e-05,
+      "loss": 4.0597,
+      "step": 662528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9143997856342166e-05,
+      "loss": 4.0657,
+      "step": 663040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9135611908831646e-05,
+      "loss": 4.0556,
+      "step": 663552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.912724234012486e-05,
+      "loss": 4.0663,
+      "step": 664064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911885639261434e-05,
+      "loss": 4.0635,
+      "step": 664576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.911047044510382e-05,
+      "loss": 4.0564,
+      "step": 665088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.91020844975933e-05,
+      "loss": 4.0727,
+      "step": 665600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.909369855008278e-05,
+      "loss": 4.0646,
+      "step": 666112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.908532898137599e-05,
+      "loss": 4.066,
+      "step": 666624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.907694303386547e-05,
+      "loss": 4.0543,
+      "step": 667136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.906855708635495e-05,
+      "loss": 4.0645,
+      "step": 667648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.906017113884443e-05,
+      "loss": 4.0676,
+      "step": 668160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.905178519133391e-05,
+      "loss": 4.0566,
+      "step": 668672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.904339924382339e-05,
+      "loss": 4.0752,
+      "step": 669184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.903501329631287e-05,
+      "loss": 4.0713,
+      "step": 669696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902664372760608e-05,
+      "loss": 4.0648,
+      "step": 670208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.901825778009556e-05,
+      "loss": 4.0697,
+      "step": 670720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.900987183258504e-05,
+      "loss": 4.0595,
+      "step": 671232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9001485885074527e-05,
+      "loss": 4.0744,
+      "step": 671744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8993099937564007e-05,
+      "loss": 4.0657,
+      "step": 672256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8984713990053487e-05,
+      "loss": 4.068,
+      "step": 672768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8976328042542966e-05,
+      "loss": 4.0645,
+      "step": 673280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.896794209503244e-05,
+      "loss": 4.0708,
+      "step": 673792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8959572526325656e-05,
+      "loss": 4.0729,
+      "step": 674304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8951186578815135e-05,
+      "loss": 4.0574,
+      "step": 674816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8942800631304615e-05,
+      "loss": 4.048,
+      "step": 675328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.893441468379409e-05,
+      "loss": 4.0612,
+      "step": 675840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8926045115087304e-05,
+      "loss": 4.0581,
+      "step": 676352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8917659167576784e-05,
+      "loss": 4.0683,
+      "step": 676864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8909273220066264e-05,
+      "loss": 4.0612,
+      "step": 677376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8900887272555744e-05,
+      "loss": 4.0703,
+      "step": 677888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8892501325045224e-05,
+      "loss": 4.0522,
+      "step": 678400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8884115377534704e-05,
+      "loss": 4.0775,
+      "step": 678912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8875729430024184e-05,
+      "loss": 4.0636,
+      "step": 679424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.886735986131739e-05,
+      "loss": 4.0716,
+      "step": 679936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885897391380687e-05,
+      "loss": 4.0681,
+      "step": 680448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885058796629635e-05,
+      "loss": 4.0627,
+      "step": 680960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884220201878583e-05,
+      "loss": 4.0622,
+      "step": 681472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.883383245007904e-05,
+      "loss": 4.073,
+      "step": 681984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.882544650256852e-05,
+      "loss": 4.0527,
+      "step": 682496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8817060555058e-05,
+      "loss": 4.0689,
+      "step": 683008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880867460754748e-05,
+      "loss": 4.0616,
+      "step": 683520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.880028866003696e-05,
+      "loss": 4.0667,
+      "step": 684032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.879191909133018e-05,
+      "loss": 4.0621,
+      "step": 684544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.878353314381966e-05,
+      "loss": 4.062,
+      "step": 685056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.877514719630914e-05,
+      "loss": 4.0582,
+      "step": 685568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876676124879862e-05,
+      "loss": 4.06,
+      "step": 686080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.87583753012881e-05,
+      "loss": 4.06,
+      "step": 686592
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.085296630859375,
+      "eval_runtime": 289.9737,
+      "eval_samples_per_second": 1315.95,
+      "eval_steps_per_second": 41.124,
+      "step": 686880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.874998935377758e-05,
+      "loss": 4.0513,
+      "step": 687104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.874160340626706e-05,
+      "loss": 4.0541,
+      "step": 687616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.873323383756027e-05,
+      "loss": 4.0645,
+      "step": 688128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.872484789004975e-05,
+      "loss": 4.0664,
+      "step": 688640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.871646194253923e-05,
+      "loss": 4.0734,
+      "step": 689152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.870807599502871e-05,
+      "loss": 4.0523,
+      "step": 689664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.869969004751819e-05,
+      "loss": 4.0604,
+      "step": 690176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.86913204788114e-05,
+      "loss": 4.0581,
+      "step": 690688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.868293453130088e-05,
+      "loss": 4.0522,
+      "step": 691200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.867456496259409e-05,
+      "loss": 4.0686,
+      "step": 691712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.866617901508357e-05,
+      "loss": 4.0577,
+      "step": 692224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.865779306757305e-05,
+      "loss": 4.0649,
+      "step": 692736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864942349886626e-05,
+      "loss": 4.0687,
+      "step": 693248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.864103755135574e-05,
+      "loss": 4.0481,
+      "step": 693760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.863265160384522e-05,
+      "loss": 4.0463,
+      "step": 694272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.86242656563347e-05,
+      "loss": 4.0496,
+      "step": 694784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.861587970882418e-05,
+      "loss": 4.0671,
+      "step": 695296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.860749376131366e-05,
+      "loss": 4.0648,
+      "step": 695808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.859910781380314e-05,
+      "loss": 4.05,
+      "step": 696320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.859072186629262e-05,
+      "loss": 4.0718,
+      "step": 696832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.85823359187821e-05,
+      "loss": 4.072,
+      "step": 697344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.857394997127158e-05,
+      "loss": 4.0581,
+      "step": 697856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.856556402376106e-05,
+      "loss": 4.0634,
+      "step": 698368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.855717807625054e-05,
+      "loss": 4.0595,
+      "step": 698880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.854879212874002e-05,
+      "loss": 4.0596,
+      "step": 699392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.85404061812295e-05,
+      "loss": 4.0567,
+      "step": 699904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.853202023371898e-05,
+      "loss": 4.0514,
+      "step": 700416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.852363428620846e-05,
+      "loss": 4.0608,
+      "step": 700928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.851526471750167e-05,
+      "loss": 4.0513,
+      "step": 701440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.850687876999115e-05,
+      "loss": 4.0488,
+      "step": 701952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849850920128436e-05,
+      "loss": 4.048,
+      "step": 702464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.849012325377384e-05,
+      "loss": 4.0563,
+      "step": 702976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.848173730626332e-05,
+      "loss": 4.0573,
+      "step": 703488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8473351358752805e-05,
+      "loss": 4.071,
+      "step": 704000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8464965411242285e-05,
+      "loss": 4.0518,
+      "step": 704512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8456595842535494e-05,
+      "loss": 4.0528,
+      "step": 705024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8448209895024974e-05,
+      "loss": 4.0678,
+      "step": 705536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8439823947514454e-05,
+      "loss": 4.0359,
+      "step": 706048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8431438000003934e-05,
+      "loss": 4.0501,
+      "step": 706560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.842306843129714e-05,
+      "loss": 4.0518,
+      "step": 707072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.841468248378662e-05,
+      "loss": 4.0448,
+      "step": 707584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.84062965362761e-05,
+      "loss": 4.0503,
+      "step": 708096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.839791058876558e-05,
+      "loss": 4.0535,
+      "step": 708608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.838952464125506e-05,
+      "loss": 4.0592,
+      "step": 709120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.838113869374454e-05,
+      "loss": 4.0476,
+      "step": 709632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.837275274623402e-05,
+      "loss": 4.0564,
+      "step": 710144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.83643667987235e-05,
+      "loss": 4.0568,
+      "step": 710656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.835599723001672e-05,
+      "loss": 4.0516,
+      "step": 711168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.83476112825062e-05,
+      "loss": 4.0278,
+      "step": 711680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.833922533499568e-05,
+      "loss": 4.0587,
+      "step": 712192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.833083938748516e-05,
+      "loss": 4.0503,
+      "step": 712704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.832246981877837e-05,
+      "loss": 4.0606,
+      "step": 713216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.831408387126785e-05,
+      "loss": 4.0402,
+      "step": 713728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.830569792375733e-05,
+      "loss": 4.0441,
+      "step": 714240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.829731197624681e-05,
+      "loss": 4.0418,
+      "step": 714752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.828892602873628e-05,
+      "loss": 4.0461,
+      "step": 715264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8280556460029496e-05,
+      "loss": 4.0645,
+      "step": 715776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8272170512518976e-05,
+      "loss": 4.0348,
+      "step": 716288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8263784565008456e-05,
+      "loss": 4.063,
+      "step": 716800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8255398617497936e-05,
+      "loss": 4.0489,
+      "step": 717312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8247012669987416e-05,
+      "loss": 4.0303,
+      "step": 717824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8238626722476896e-05,
+      "loss": 4.0458,
+      "step": 718336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8230240774966376e-05,
+      "loss": 4.0373,
+      "step": 718848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8221871206259585e-05,
+      "loss": 4.0396,
+      "step": 719360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8213485258749065e-05,
+      "loss": 4.0495,
+      "step": 719872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8205099311238545e-05,
+      "loss": 4.0574,
+      "step": 720384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8196713363728025e-05,
+      "loss": 4.0404,
+      "step": 720896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8188327416217505e-05,
+      "loss": 4.0546,
+      "step": 721408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8179941468706985e-05,
+      "loss": 4.0295,
+      "step": 721920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8171571900000194e-05,
+      "loss": 4.0478,
+      "step": 722432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.816318595248968e-05,
+      "loss": 4.0473,
+      "step": 722944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.815480000497916e-05,
+      "loss": 4.0522,
+      "step": 723456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.814641405746864e-05,
+      "loss": 4.049,
+      "step": 723968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.813804448876185e-05,
+      "loss": 4.0502,
+      "step": 724480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.812965854125133e-05,
+      "loss": 4.0682,
+      "step": 724992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.812127259374081e-05,
+      "loss": 4.0363,
+      "step": 725504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.811290302503402e-05,
+      "loss": 4.0548,
+      "step": 726016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.81045170775235e-05,
+      "loss": 4.0456,
+      "step": 726528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.809613113001298e-05,
+      "loss": 4.0374,
+      "step": 727040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.808774518250246e-05,
+      "loss": 4.0523,
+      "step": 727552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.807937561379567e-05,
+      "loss": 4.0547,
+      "step": 728064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.807098966628515e-05,
+      "loss": 4.0561,
+      "step": 728576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8062603718774634e-05,
+      "loss": 4.0693,
+      "step": 729088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8054217771264114e-05,
+      "loss": 4.0433,
+      "step": 729600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.8045831823753594e-05,
+      "loss": 4.0373,
+      "step": 730112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.80374622550468e-05,
+      "loss": 4.0451,
+      "step": 730624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.802907630753628e-05,
+      "loss": 4.0566,
+      "step": 731136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.802069036002576e-05,
+      "loss": 4.0427,
+      "step": 731648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.801230441251524e-05,
+      "loss": 4.0514,
+      "step": 732160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.800391846500472e-05,
+      "loss": 4.0462,
+      "step": 732672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79955325174942e-05,
+      "loss": 4.0487,
+      "step": 733184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.798714656998368e-05,
+      "loss": 4.0469,
+      "step": 733696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797877700127689e-05,
+      "loss": 4.0347,
+      "step": 734208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.797039105376637e-05,
+      "loss": 4.0494,
+      "step": 734720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.796200510625585e-05,
+      "loss": 4.0567,
+      "step": 735232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.795361915874533e-05,
+      "loss": 4.0483,
+      "step": 735744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.794523321123481e-05,
+      "loss": 4.0324,
+      "step": 736256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.79368472637243e-05,
+      "loss": 4.0436,
+      "step": 736768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.792846131621377e-05,
+      "loss": 4.0433,
+      "step": 737280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.792009174750699e-05,
+      "loss": 4.0519,
+      "step": 737792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.791170579999647e-05,
+      "loss": 4.0493,
+      "step": 738304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.790331985248594e-05,
+      "loss": 4.0343,
+      "step": 738816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.789493390497542e-05,
+      "loss": 4.0437,
+      "step": 739328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.78865479574649e-05,
+      "loss": 4.0337,
+      "step": 739840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.787816200995438e-05,
+      "loss": 4.0452,
+      "step": 740352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.786977606244386e-05,
+      "loss": 4.0385,
+      "step": 740864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.786140649373707e-05,
+      "loss": 4.0339,
+      "step": 741376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7853036925030286e-05,
+      "loss": 4.0446,
+      "step": 741888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7844650977519766e-05,
+      "loss": 4.0446,
+      "step": 742400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7836265030009246e-05,
+      "loss": 4.0387,
+      "step": 742912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7827879082498725e-05,
+      "loss": 4.0334,
+      "step": 743424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7819493134988205e-05,
+      "loss": 4.0366,
+      "step": 743936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7811107187477685e-05,
+      "loss": 4.0495,
+      "step": 744448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7802721239967165e-05,
+      "loss": 4.0293,
+      "step": 744960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7794335292456645e-05,
+      "loss": 4.0536,
+      "step": 745472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7785949344946125e-05,
+      "loss": 4.0494,
+      "step": 745984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7777563397435605e-05,
+      "loss": 4.0407,
+      "step": 746496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7769193828728814e-05,
+      "loss": 4.049,
+      "step": 747008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7760807881218294e-05,
+      "loss": 4.0387,
+      "step": 747520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7752421933707774e-05,
+      "loss": 4.0485,
+      "step": 748032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7744035986197254e-05,
+      "loss": 4.0453,
+      "step": 748544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.773566641749047e-05,
+      "loss": 4.043,
+      "step": 749056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.772729684878368e-05,
+      "loss": 4.0435,
+      "step": 749568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771891090127316e-05,
+      "loss": 4.0436,
+      "step": 750080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.771052495376264e-05,
+      "loss": 4.0471,
+      "step": 750592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.770213900625212e-05,
+      "loss": 4.0371,
+      "step": 751104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.76937530587416e-05,
+      "loss": 4.0248,
+      "step": 751616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.768536711123108e-05,
+      "loss": 4.0403,
+      "step": 752128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.767698116372056e-05,
+      "loss": 4.0279,
+      "step": 752640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766859521621004e-05,
+      "loss": 4.0498,
+      "step": 753152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.766020926869952e-05,
+      "loss": 4.0356,
+      "step": 753664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7651823321189e-05,
+      "loss": 4.0485,
+      "step": 754176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.764345375248221e-05,
+      "loss": 4.0273,
+      "step": 754688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.763506780497169e-05,
+      "loss": 4.0531,
+      "step": 755200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7626681857461175e-05,
+      "loss": 4.0416,
+      "step": 755712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7618295909950654e-05,
+      "loss": 4.0542,
+      "step": 756224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.760990996244013e-05,
+      "loss": 4.0434,
+      "step": 756736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.760152401492961e-05,
+      "loss": 4.0423,
+      "step": 757248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.759313806741909e-05,
+      "loss": 4.0345,
+      "step": 757760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.758475211990857e-05,
+      "loss": 4.0553,
+      "step": 758272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.757638255120178e-05,
+      "loss": 4.0262,
+      "step": 758784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7567996603691257e-05,
+      "loss": 4.0497,
+      "step": 759296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7559610656180737e-05,
+      "loss": 4.0388,
+      "step": 759808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7551224708670216e-05,
+      "loss": 4.041,
+      "step": 760320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.7542855139963426e-05,
+      "loss": 4.0423,
+      "step": 760832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.753446919245291e-05,
+      "loss": 4.0369,
+      "step": 761344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.752608324494239e-05,
+      "loss": 4.0392,
+      "step": 761856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.751769729743187e-05,
+      "loss": 4.039,
+      "step": 762368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.750932772872508e-05,
+      "loss": 4.0349,
+      "step": 762880
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.069877624511719,
+      "eval_runtime": 293.3818,
+      "eval_samples_per_second": 1300.664,
+      "eval_steps_per_second": 40.647,
+      "step": 763200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.750094178121456e-05,
+      "loss": 4.0241,
+      "step": 763392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.749255583370404e-05,
+      "loss": 4.0312,
+      "step": 763904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.748416988619352e-05,
+      "loss": 4.0404,
+      "step": 764416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7475783938683e-05,
+      "loss": 4.0447,
+      "step": 764928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.746739799117248e-05,
+      "loss": 4.051,
+      "step": 765440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745901204366196e-05,
+      "loss": 4.0326,
+      "step": 765952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.745062609615144e-05,
+      "loss": 4.0343,
+      "step": 766464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.744224014864092e-05,
+      "loss": 4.0399,
+      "step": 766976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.74338542011304e-05,
+      "loss": 4.0315,
+      "step": 767488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.742546825361988e-05,
+      "loss": 4.0404,
+      "step": 768000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.741708230610936e-05,
+      "loss": 4.0412,
+      "step": 768512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740869635859884e-05,
+      "loss": 4.0417,
+      "step": 769024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.740032678989206e-05,
+      "loss": 4.0445,
+      "step": 769536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.739194084238154e-05,
+      "loss": 4.0263,
+      "step": 770048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7383554894871017e-05,
+      "loss": 4.0254,
+      "step": 770560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.737516894736049e-05,
+      "loss": 4.0228,
+      "step": 771072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.736678299984997e-05,
+      "loss": 4.0501,
+      "step": 771584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.735839705233945e-05,
+      "loss": 4.0399,
+      "step": 772096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.735001110482893e-05,
+      "loss": 4.0304,
+      "step": 772608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.734162515731841e-05,
+      "loss": 4.0471,
+      "step": 773120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.733323920980789e-05,
+      "loss": 4.0515,
+      "step": 773632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.732485326229737e-05,
+      "loss": 4.0387,
+      "step": 774144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.731646731478685e-05,
+      "loss": 4.0389,
+      "step": 774656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.730808136727633e-05,
+      "loss": 4.0364,
+      "step": 775168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.729969541976581e-05,
+      "loss": 4.0422,
+      "step": 775680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7291325851059025e-05,
+      "loss": 4.0326,
+      "step": 776192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7282939903548505e-05,
+      "loss": 4.0272,
+      "step": 776704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7274553956037985e-05,
+      "loss": 4.04,
+      "step": 777216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7266168008527465e-05,
+      "loss": 4.0267,
+      "step": 777728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7257798439820674e-05,
+      "loss": 4.0239,
+      "step": 778240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724942887111388e-05,
+      "loss": 4.0324,
+      "step": 778752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.724104292360336e-05,
+      "loss": 4.0312,
+      "step": 779264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723265697609284e-05,
+      "loss": 4.0379,
+      "step": 779776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.722427102858232e-05,
+      "loss": 4.0482,
+      "step": 780288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.72158850810718e-05,
+      "loss": 4.0264,
+      "step": 780800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.720751551236501e-05,
+      "loss": 4.034,
+      "step": 781312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71991295648545e-05,
+      "loss": 4.046,
+      "step": 781824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.719074361734398e-05,
+      "loss": 4.0154,
+      "step": 782336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.718235766983346e-05,
+      "loss": 4.0273,
+      "step": 782848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.717397172232294e-05,
+      "loss": 4.0302,
+      "step": 783360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.716558577481242e-05,
+      "loss": 4.0207,
+      "step": 783872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.71571998273019e-05,
+      "loss": 4.0312,
+      "step": 784384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714881387979138e-05,
+      "loss": 4.0326,
+      "step": 784896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.714042793228085e-05,
+      "loss": 4.0367,
+      "step": 785408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.713205836357407e-05,
+      "loss": 4.0269,
+      "step": 785920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.712368879486728e-05,
+      "loss": 4.0313,
+      "step": 786432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.711530284735676e-05,
+      "loss": 4.036,
+      "step": 786944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.710691689984624e-05,
+      "loss": 4.0302,
+      "step": 787456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.709853095233572e-05,
+      "loss": 4.0066,
+      "step": 787968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70901450048252e-05,
+      "loss": 4.0349,
+      "step": 788480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7081759057314677e-05,
+      "loss": 4.0313,
+      "step": 788992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.707338948860789e-05,
+      "loss": 4.0381,
+      "step": 789504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.706500354109737e-05,
+      "loss": 4.023,
+      "step": 790016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.705661759358685e-05,
+      "loss": 4.0189,
+      "step": 790528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7048231646076326e-05,
+      "loss": 4.0241,
+      "step": 791040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7039845698565806e-05,
+      "loss": 4.0225,
+      "step": 791552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7031459751055285e-05,
+      "loss": 4.0384,
+      "step": 792064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.70230901823485e-05,
+      "loss": 4.0119,
+      "step": 792576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7014704234837975e-05,
+      "loss": 4.039,
+      "step": 793088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7006318287327454e-05,
+      "loss": 4.0321,
+      "step": 793600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6997932339816934e-05,
+      "loss": 4.0086,
+      "step": 794112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698954639230642e-05,
+      "loss": 4.0279,
+      "step": 794624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.698117682359963e-05,
+      "loss": 4.0124,
+      "step": 795136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.697279087608911e-05,
+      "loss": 4.0156,
+      "step": 795648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.696440492857859e-05,
+      "loss": 4.0243,
+      "step": 796160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.695601898106807e-05,
+      "loss": 4.0379,
+      "step": 796672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.694763303355755e-05,
+      "loss": 4.0189,
+      "step": 797184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.693924708604703e-05,
+      "loss": 4.0294,
+      "step": 797696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.693086113853651e-05,
+      "loss": 4.01,
+      "step": 798208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.692249156982972e-05,
+      "loss": 4.0251,
+      "step": 798720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.69141056223192e-05,
+      "loss": 4.0297,
+      "step": 799232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.690571967480868e-05,
+      "loss": 4.0335,
+      "step": 799744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.689733372729816e-05,
+      "loss": 4.0257,
+      "step": 800256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.688894777978764e-05,
+      "loss": 4.0267,
+      "step": 800768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.688056183227712e-05,
+      "loss": 4.0458,
+      "step": 801280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.68721758847666e-05,
+      "loss": 4.0186,
+      "step": 801792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6863789937256086e-05,
+      "loss": 4.0295,
+      "step": 802304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6855403989745566e-05,
+      "loss": 4.0242,
+      "step": 802816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6847034421038775e-05,
+      "loss": 4.0217,
+      "step": 803328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6838648473528255e-05,
+      "loss": 4.0311,
+      "step": 803840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6830262526017735e-05,
+      "loss": 4.0286,
+      "step": 804352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6821876578507214e-05,
+      "loss": 4.0371,
+      "step": 804864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.681349063099669e-05,
+      "loss": 4.0462,
+      "step": 805376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6805121062289904e-05,
+      "loss": 4.03,
+      "step": 805888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6796735114779383e-05,
+      "loss": 4.0118,
+      "step": 806400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6788349167268863e-05,
+      "loss": 4.0218,
+      "step": 806912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677997959856207e-05,
+      "loss": 4.0347,
+      "step": 807424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.677159365105155e-05,
+      "loss": 4.0238,
+      "step": 807936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.676320770354104e-05,
+      "loss": 4.0308,
+      "step": 808448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.675482175603051e-05,
+      "loss": 4.0252,
+      "step": 808960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.674643580851999e-05,
+      "loss": 4.0268,
+      "step": 809472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.673804986100947e-05,
+      "loss": 4.0247,
+      "step": 809984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672966391349895e-05,
+      "loss": 4.0145,
+      "step": 810496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.672127796598843e-05,
+      "loss": 4.0321,
+      "step": 811008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.671289201847791e-05,
+      "loss": 4.0344,
+      "step": 811520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.670452244977112e-05,
+      "loss": 4.0282,
+      "step": 812032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66961365022606e-05,
+      "loss": 4.0111,
+      "step": 812544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.668776693355381e-05,
+      "loss": 4.024,
+      "step": 813056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667938098604329e-05,
+      "loss": 4.0257,
+      "step": 813568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.667099503853278e-05,
+      "loss": 4.0301,
+      "step": 814080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.666260909102226e-05,
+      "loss": 4.0293,
+      "step": 814592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.665422314351174e-05,
+      "loss": 4.0111,
+      "step": 815104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.664583719600122e-05,
+      "loss": 4.0209,
+      "step": 815616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.66374512484907e-05,
+      "loss": 4.0136,
+      "step": 816128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.662906530098018e-05,
+      "loss": 4.0226,
+      "step": 816640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.662067935346966e-05,
+      "loss": 4.0187,
+      "step": 817152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6612326163566595e-05,
+      "loss": 4.0157,
+      "step": 817664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6603940216056075e-05,
+      "loss": 4.0259,
+      "step": 818176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6595554268545555e-05,
+      "loss": 4.021,
+      "step": 818688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6587168321035035e-05,
+      "loss": 4.0192,
+      "step": 819200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6578798752328244e-05,
+      "loss": 4.0125,
+      "step": 819712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.657041280481773e-05,
+      "loss": 4.0143,
+      "step": 820224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.656202685730721e-05,
+      "loss": 4.0278,
+      "step": 820736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.655364090979669e-05,
+      "loss": 4.0133,
+      "step": 821248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.654525496228617e-05,
+      "loss": 4.027,
+      "step": 821760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.653686901477565e-05,
+      "loss": 4.0275,
+      "step": 822272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.652848306726513e-05,
+      "loss": 4.0211,
+      "step": 822784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.652009711975461e-05,
+      "loss": 4.0267,
+      "step": 823296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.651171117224409e-05,
+      "loss": 4.0218,
+      "step": 823808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.650332522473357e-05,
+      "loss": 4.0273,
+      "step": 824320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.649493927722305e-05,
+      "loss": 4.031,
+      "step": 824832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.648656970851626e-05,
+      "loss": 4.0199,
+      "step": 825344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.647820013980947e-05,
+      "loss": 4.0225,
+      "step": 825856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646981419229895e-05,
+      "loss": 4.0226,
+      "step": 826368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.646142824478843e-05,
+      "loss": 4.028,
+      "step": 826880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6453042297277915e-05,
+      "loss": 4.0129,
+      "step": 827392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6444656349767395e-05,
+      "loss": 4.0058,
+      "step": 827904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6436286781060604e-05,
+      "loss": 4.0181,
+      "step": 828416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6427900833550084e-05,
+      "loss": 4.0085,
+      "step": 828928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6419514886039564e-05,
+      "loss": 4.028,
+      "step": 829440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6411128938529044e-05,
+      "loss": 4.0165,
+      "step": 829952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6402742991018524e-05,
+      "loss": 4.0269,
+      "step": 830464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.639437342231173e-05,
+      "loss": 4.0115,
+      "step": 830976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.638598747480121e-05,
+      "loss": 4.0284,
+      "step": 831488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.637760152729069e-05,
+      "loss": 4.0211,
+      "step": 832000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6369215579780166e-05,
+      "loss": 4.0345,
+      "step": 832512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.636082963226965e-05,
+      "loss": 4.0241,
+      "step": 833024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.635244368475913e-05,
+      "loss": 4.0224,
+      "step": 833536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.634405773724861e-05,
+      "loss": 4.0161,
+      "step": 834048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.633567178973809e-05,
+      "loss": 4.0317,
+      "step": 834560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.63273022210313e-05,
+      "loss": 4.0106,
+      "step": 835072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631891627352078e-05,
+      "loss": 4.025,
+      "step": 835584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.631053032601026e-05,
+      "loss": 4.0225,
+      "step": 836096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.630214437849974e-05,
+      "loss": 4.0191,
+      "step": 836608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.629377480979295e-05,
+      "loss": 4.0216,
+      "step": 837120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.628538886228243e-05,
+      "loss": 4.0189,
+      "step": 837632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.627700291477191e-05,
+      "loss": 4.0192,
+      "step": 838144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.626863334606512e-05,
+      "loss": 4.0183,
+      "step": 838656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.62602473985546e-05,
+      "loss": 4.0153,
+      "step": 839168
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.056442737579346,
+      "eval_runtime": 294.4579,
+      "eval_samples_per_second": 1295.91,
+      "eval_steps_per_second": 40.498,
+      "step": 839520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6251861451044087e-05,
+      "loss": 4.0081,
+      "step": 839680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6243475503533566e-05,
+      "loss": 4.0104,
+      "step": 840192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6235089556023046e-05,
+      "loss": 4.0193,
+      "step": 840704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6226703608512526e-05,
+      "loss": 4.0256,
+      "step": 841216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6218317661002006e-05,
+      "loss": 4.0306,
+      "step": 841728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6209931713491486e-05,
+      "loss": 4.0122,
+      "step": 842240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6201545765980966e-05,
+      "loss": 4.0174,
+      "step": 842752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6193192576077905e-05,
+      "loss": 4.0208,
+      "step": 843264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6184806628567384e-05,
+      "loss": 4.0108,
+      "step": 843776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6176420681056864e-05,
+      "loss": 4.0114,
+      "step": 844288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6168034733546344e-05,
+      "loss": 4.0272,
+      "step": 844800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6159648786035824e-05,
+      "loss": 4.0201,
+      "step": 845312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.615129559613277e-05,
+      "loss": 4.0276,
+      "step": 845824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.614290964862225e-05,
+      "loss": 4.0049,
+      "step": 846336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.613452370111173e-05,
+      "loss": 4.0048,
+      "step": 846848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.612613775360121e-05,
+      "loss": 4.0038,
+      "step": 847360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.611775180609069e-05,
+      "loss": 4.0266,
+      "step": 847872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.610936585858017e-05,
+      "loss": 4.0242,
+      "step": 848384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.610097991106965e-05,
+      "loss": 4.011,
+      "step": 848896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.609259396355913e-05,
+      "loss": 4.0239,
+      "step": 849408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.608420801604861e-05,
+      "loss": 4.0332,
+      "step": 849920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.607582206853809e-05,
+      "loss": 4.0163,
+      "step": 850432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.606743612102757e-05,
+      "loss": 4.021,
+      "step": 850944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.605905017351705e-05,
+      "loss": 4.0168,
+      "step": 851456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.605068060481026e-05,
+      "loss": 4.0243,
+      "step": 851968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.604229465729974e-05,
+      "loss": 4.0118,
+      "step": 852480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6033908709789225e-05,
+      "loss": 4.0087,
+      "step": 852992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6025522762278705e-05,
+      "loss": 4.0245,
+      "step": 853504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6017153193571914e-05,
+      "loss": 4.0086,
+      "step": 854016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6008767246061394e-05,
+      "loss": 4.0056,
+      "step": 854528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.6000381298550874e-05,
+      "loss": 4.0081,
+      "step": 855040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5991995351040354e-05,
+      "loss": 4.0117,
+      "step": 855552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.598362578233356e-05,
+      "loss": 4.0204,
+      "step": 856064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.597523983482304e-05,
+      "loss": 4.0293,
+      "step": 856576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.596685388731252e-05,
+      "loss": 4.0048,
+      "step": 857088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595848431860573e-05,
+      "loss": 4.0166,
+      "step": 857600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.595009837109521e-05,
+      "loss": 4.0242,
+      "step": 858112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.594171242358469e-05,
+      "loss": 3.9925,
+      "step": 858624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.593332647607418e-05,
+      "loss": 4.0058,
+      "step": 859136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.592494052856366e-05,
+      "loss": 4.0134,
+      "step": 859648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.591655458105313e-05,
+      "loss": 3.9987,
+      "step": 860160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.590816863354261e-05,
+      "loss": 4.0141,
+      "step": 860672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.589978268603209e-05,
+      "loss": 4.0147,
+      "step": 861184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.589139673852157e-05,
+      "loss": 4.012,
+      "step": 861696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.588302716981478e-05,
+      "loss": 4.0079,
+      "step": 862208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.587464122230426e-05,
+      "loss": 4.0119,
+      "step": 862720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.586625527479374e-05,
+      "loss": 4.0124,
+      "step": 863232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.585786932728322e-05,
+      "loss": 4.0118,
+      "step": 863744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.584949975857643e-05,
+      "loss": 3.9873,
+      "step": 864256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5841113811065916e-05,
+      "loss": 4.0126,
+      "step": 864768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5832727863555396e-05,
+      "loss": 4.0147,
+      "step": 865280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5824341916044876e-05,
+      "loss": 4.0169,
+      "step": 865792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5815955968534356e-05,
+      "loss": 4.0035,
+      "step": 866304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5807570021023836e-05,
+      "loss": 4.0021,
+      "step": 866816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5799200452317045e-05,
+      "loss": 4.0015,
+      "step": 867328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5790814504806525e-05,
+      "loss": 4.0012,
+      "step": 867840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5782428557296005e-05,
+      "loss": 4.0213,
+      "step": 868352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5774042609785485e-05,
+      "loss": 3.9953,
+      "step": 868864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5765656662274965e-05,
+      "loss": 4.011,
+      "step": 869376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5757287093568174e-05,
+      "loss": 4.0132,
+      "step": 869888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5748901146057654e-05,
+      "loss": 3.9936,
+      "step": 870400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5740515198547134e-05,
+      "loss": 4.0015,
+      "step": 870912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5732129251036614e-05,
+      "loss": 3.9967,
+      "step": 871424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5723743303526094e-05,
+      "loss": 3.999,
+      "step": 871936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.571535735601558e-05,
+      "loss": 3.999,
+      "step": 872448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.570697140850506e-05,
+      "loss": 4.0241,
+      "step": 872960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.569858546099454e-05,
+      "loss": 4.0017,
+      "step": 873472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.569021589228775e-05,
+      "loss": 4.0028,
+      "step": 873984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.568182994477723e-05,
+      "loss": 3.9956,
+      "step": 874496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.567346037607044e-05,
+      "loss": 4.0025,
+      "step": 875008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.566507442855992e-05,
+      "loss": 4.0096,
+      "step": 875520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.56566884810494e-05,
+      "loss": 4.01,
+      "step": 876032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.564830253353888e-05,
+      "loss": 4.0094,
+      "step": 876544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563991658602836e-05,
+      "loss": 4.0101,
+      "step": 877056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.563153063851784e-05,
+      "loss": 4.0213,
+      "step": 877568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.562316106981105e-05,
+      "loss": 4.0004,
+      "step": 878080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5614775122300534e-05,
+      "loss": 4.0099,
+      "step": 878592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5606389174790014e-05,
+      "loss": 4.0056,
+      "step": 879104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5598003227279494e-05,
+      "loss": 4.001,
+      "step": 879616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558961727976897e-05,
+      "loss": 4.0128,
+      "step": 880128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.558123133225845e-05,
+      "loss": 4.0084,
+      "step": 880640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.557284538474793e-05,
+      "loss": 4.0193,
+      "step": 881152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.556445943723741e-05,
+      "loss": 4.0234,
+      "step": 881664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5556089868530616e-05,
+      "loss": 4.0163,
+      "step": 882176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5547703921020096e-05,
+      "loss": 3.9916,
+      "step": 882688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5539317973509576e-05,
+      "loss": 4.0025,
+      "step": 883200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5530932025999056e-05,
+      "loss": 4.0141,
+      "step": 883712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.5522546078488536e-05,
+      "loss": 4.0076,
+      "step": 884224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.551417650978175e-05,
+      "loss": 4.0077,
+      "step": 884736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.550580694107497e-05,
+      "loss": 4.008,
+      "step": 885248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.549742099356444e-05,
+      "loss": 4.01,
+      "step": 885760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.548903504605392e-05,
+      "loss": 4.0029,
+      "step": 886272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.54806490985434e-05,
+      "loss": 3.998,
+      "step": 886784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.547226315103288e-05,
+      "loss": 4.0119,
+      "step": 887296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.546387720352236e-05,
+      "loss": 4.009,
+      "step": 887808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.545549125601184e-05,
+      "loss": 4.0145,
+      "step": 888320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.544712168730505e-05,
+      "loss": 3.9913,
+      "step": 888832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543873573979453e-05,
+      "loss": 4.0051,
+      "step": 889344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.543034979228401e-05,
+      "loss": 4.005,
+      "step": 889856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.542196384477349e-05,
+      "loss": 4.0124,
+      "step": 890368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.541357789726297e-05,
+      "loss": 4.0112,
+      "step": 890880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5405208328556186e-05,
+      "loss": 3.9919,
+      "step": 891392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5396822381045665e-05,
+      "loss": 4.0058,
+      "step": 891904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5388452812338875e-05,
+      "loss": 3.9892,
+      "step": 892416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5380066864828355e-05,
+      "loss": 4.0077,
+      "step": 892928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5371680917317835e-05,
+      "loss": 3.997,
+      "step": 893440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5363294969807314e-05,
+      "loss": 3.9965,
+      "step": 893952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5354909022296794e-05,
+      "loss": 4.0078,
+      "step": 894464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5346523074786274e-05,
+      "loss": 4.0032,
+      "step": 894976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5338153506079483e-05,
+      "loss": 3.9999,
+      "step": 895488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5329767558568963e-05,
+      "loss": 4.0005,
+      "step": 896000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.532138161105844e-05,
+      "loss": 3.9929,
+      "step": 896512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.531299566354792e-05,
+      "loss": 4.0121,
+      "step": 897024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.530460971603741e-05,
+      "loss": 3.9919,
+      "step": 897536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.529622376852689e-05,
+      "loss": 4.0104,
+      "step": 898048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.52878541998201e-05,
+      "loss": 4.0044,
+      "step": 898560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527946825230958e-05,
+      "loss": 4.0072,
+      "step": 899072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.527108230479906e-05,
+      "loss": 4.0049,
+      "step": 899584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.526269635728854e-05,
+      "loss": 4.0008,
+      "step": 900096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.525431040977802e-05,
+      "loss": 4.0123,
+      "step": 900608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.52459244622675e-05,
+      "loss": 4.0102,
+      "step": 901120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.523753851475697e-05,
+      "loss": 4.0021,
+      "step": 901632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.522915256724645e-05,
+      "loss": 4.0022,
+      "step": 902144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.522076661973593e-05,
+      "loss": 4.0028,
+      "step": 902656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.521239705102915e-05,
+      "loss": 4.0097,
+      "step": 903168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.520401110351863e-05,
+      "loss": 3.9946,
+      "step": 903680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.519562515600811e-05,
+      "loss": 3.9921,
+      "step": 904192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.518723920849759e-05,
+      "loss": 3.9963,
+      "step": 904704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.51788696397908e-05,
+      "loss": 3.994,
+      "step": 905216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.517048369228028e-05,
+      "loss": 4.0073,
+      "step": 905728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.516209774476976e-05,
+      "loss": 3.9989,
+      "step": 906240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.515372817606297e-05,
+      "loss": 4.0118,
+      "step": 906752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5145342228552446e-05,
+      "loss": 3.9923,
+      "step": 907264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5136956281041926e-05,
+      "loss": 4.0106,
+      "step": 907776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5128570333531406e-05,
+      "loss": 4.0072,
+      "step": 908288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5120184386020886e-05,
+      "loss": 4.0104,
+      "step": 908800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5111798438510366e-05,
+      "loss": 4.0058,
+      "step": 909312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5103412490999846e-05,
+      "loss": 4.0052,
+      "step": 909824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.5095026543489326e-05,
+      "loss": 3.9996,
+      "step": 910336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.508664059597881e-05,
+      "loss": 4.0129,
+      "step": 910848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.507825464846829e-05,
+      "loss": 3.9958,
+      "step": 911360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.50698850797615e-05,
+      "loss": 4.0021,
+      "step": 911872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.506149913225098e-05,
+      "loss": 4.0074,
+      "step": 912384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.505311318474046e-05,
+      "loss": 4.0003,
+      "step": 912896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.504474361603367e-05,
+      "loss": 4.0005,
+      "step": 913408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.503635766852315e-05,
+      "loss": 4.0071,
+      "step": 913920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.502797172101263e-05,
+      "loss": 3.9978,
+      "step": 914432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.501958577350211e-05,
+      "loss": 3.9992,
+      "step": 914944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.501119982599159e-05,
+      "loss": 3.9999,
+      "step": 915456
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.045405387878418,
+      "eval_runtime": 295.4603,
+      "eval_samples_per_second": 1291.514,
+      "eval_steps_per_second": 40.361,
+      "step": 915840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.500281387848107e-05,
+      "loss": 4.0005,
+      "step": 915968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.499442793097055e-05,
+      "loss": 3.9871,
+      "step": 916480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.498604198346003e-05,
+      "loss": 4.0048,
+      "step": 916992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.497765603594951e-05,
+      "loss": 4.0085,
+      "step": 917504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.496927008843899e-05,
+      "loss": 4.0075,
+      "step": 918016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.496088414092847e-05,
+      "loss": 3.9992,
+      "step": 918528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.495249819341795e-05,
+      "loss": 3.9996,
+      "step": 919040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.494411224590743e-05,
+      "loss": 4.0012,
+      "step": 919552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.493572629839691e-05,
+      "loss": 3.9933,
+      "step": 920064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.492734035088639e-05,
+      "loss": 3.998,
+      "step": 920576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.491895440337587e-05,
+      "loss": 4.0076,
+      "step": 921088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.491056845586535e-05,
+      "loss": 4.0037,
+      "step": 921600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.490219888715856e-05,
+      "loss": 4.0146,
+      "step": 922112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.489381293964804e-05,
+      "loss": 3.9879,
+      "step": 922624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.488542699213752e-05,
+      "loss": 3.9869,
+      "step": 923136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4877041044627e-05,
+      "loss": 3.9846,
+      "step": 923648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486865509711648e-05,
+      "loss": 4.0126,
+      "step": 924160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4860269149605965e-05,
+      "loss": 4.0019,
+      "step": 924672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4851883202095445e-05,
+      "loss": 3.9952,
+      "step": 925184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4843497254584925e-05,
+      "loss": 4.0081,
+      "step": 925696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4835111307074405e-05,
+      "loss": 4.0143,
+      "step": 926208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.482672535956388e-05,
+      "loss": 4.0023,
+      "step": 926720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.481833941205336e-05,
+      "loss": 4.0015,
+      "step": 927232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480995346454284e-05,
+      "loss": 4.0001,
+      "step": 927744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.480158389583605e-05,
+      "loss": 4.0102,
+      "step": 928256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.479321432712926e-05,
+      "loss": 3.9961,
+      "step": 928768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.478482837961874e-05,
+      "loss": 3.9895,
+      "step": 929280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.477644243210822e-05,
+      "loss": 4.0094,
+      "step": 929792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.47680564845977e-05,
+      "loss": 3.9842,
+      "step": 930304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.475967053708718e-05,
+      "loss": 3.9895,
+      "step": 930816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.47513009683804e-05,
+      "loss": 3.993,
+      "step": 931328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.474291502086988e-05,
+      "loss": 3.9935,
+      "step": 931840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.473452907335935e-05,
+      "loss": 4.0037,
+      "step": 932352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.472614312584883e-05,
+      "loss": 4.0104,
+      "step": 932864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.471775717833831e-05,
+      "loss": 3.9922,
+      "step": 933376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.470938760963152e-05,
+      "loss": 3.9955,
+      "step": 933888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4701001662121e-05,
+      "loss": 4.009,
+      "step": 934400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469261571461048e-05,
+      "loss": 3.9774,
+      "step": 934912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.468422976709996e-05,
+      "loss": 3.9946,
+      "step": 935424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.467584381958944e-05,
+      "loss": 3.9949,
+      "step": 935936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.466745787207892e-05,
+      "loss": 3.9824,
+      "step": 936448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.46590719245684e-05,
+      "loss": 3.9956,
+      "step": 936960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.465068597705788e-05,
+      "loss": 4.0032,
+      "step": 937472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.464230002954737e-05,
+      "loss": 3.9936,
+      "step": 937984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4633930460840577e-05,
+      "loss": 3.9923,
+      "step": 938496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4625544513330056e-05,
+      "loss": 4.0011,
+      "step": 939008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4617158565819536e-05,
+      "loss": 3.9919,
+      "step": 939520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4608772618309016e-05,
+      "loss": 3.9997,
+      "step": 940032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4600419428405955e-05,
+      "loss": 3.9684,
+      "step": 940544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4592033480895435e-05,
+      "loss": 3.9927,
+      "step": 941056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4583647533384915e-05,
+      "loss": 4.0034,
+      "step": 941568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4575277964678124e-05,
+      "loss": 3.9984,
+      "step": 942080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4566892017167604e-05,
+      "loss": 3.9847,
+      "step": 942592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455850606965709e-05,
+      "loss": 3.9893,
+      "step": 943104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.455012012214657e-05,
+      "loss": 3.9826,
+      "step": 943616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.454173417463605e-05,
+      "loss": 3.9838,
+      "step": 944128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.453336460592926e-05,
+      "loss": 4.0099,
+      "step": 944640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.452497865841874e-05,
+      "loss": 3.9788,
+      "step": 945152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.451659271090822e-05,
+      "loss": 3.9958,
+      "step": 945664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.45082067633977e-05,
+      "loss": 3.9988,
+      "step": 946176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449982081588718e-05,
+      "loss": 3.9781,
+      "step": 946688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.449143486837666e-05,
+      "loss": 3.9843,
+      "step": 947200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.448304892086614e-05,
+      "loss": 3.9807,
+      "step": 947712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.447466297335562e-05,
+      "loss": 3.9784,
+      "step": 948224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.44662770258451e-05,
+      "loss": 3.9862,
+      "step": 948736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.445789107833458e-05,
+      "loss": 4.0052,
+      "step": 949248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.444952150962779e-05,
+      "loss": 3.9938,
+      "step": 949760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4441135562117275e-05,
+      "loss": 3.988,
+      "step": 950272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4432749614606755e-05,
+      "loss": 3.9758,
+      "step": 950784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4424363667096235e-05,
+      "loss": 3.9876,
+      "step": 951296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.441597771958571e-05,
+      "loss": 3.9938,
+      "step": 951808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.440759177207519e-05,
+      "loss": 3.9967,
+      "step": 952320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439920582456467e-05,
+      "loss": 3.9899,
+      "step": 952832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.439081987705415e-05,
+      "loss": 3.9942,
+      "step": 953344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.438245030834736e-05,
+      "loss": 4.0066,
+      "step": 953856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.437406436083684e-05,
+      "loss": 3.9863,
+      "step": 954368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.436569479213005e-05,
+      "loss": 3.99,
+      "step": 954880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.435730884461953e-05,
+      "loss": 3.994,
+      "step": 955392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434892289710901e-05,
+      "loss": 3.9883,
+      "step": 955904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.434053694959849e-05,
+      "loss": 3.996,
+      "step": 956416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.433216738089171e-05,
+      "loss": 3.9943,
+      "step": 956928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.432378143338118e-05,
+      "loss": 3.9995,
+      "step": 957440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.431539548587066e-05,
+      "loss": 4.0109,
+      "step": 957952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.430700953836014e-05,
+      "loss": 3.9983,
+      "step": 958464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.429862359084962e-05,
+      "loss": 3.978,
+      "step": 958976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.42902376433391e-05,
+      "loss": 3.9855,
+      "step": 959488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428185169582858e-05,
+      "loss": 4.0005,
+      "step": 960000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.427348212712179e-05,
+      "loss": 3.9938,
+      "step": 960512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.426509617961127e-05,
+      "loss": 3.988,
+      "step": 961024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.425671023210075e-05,
+      "loss": 3.9956,
+      "step": 961536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.424832428459023e-05,
+      "loss": 3.9958,
+      "step": 962048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.423993833707971e-05,
+      "loss": 3.9876,
+      "step": 962560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4231568768372926e-05,
+      "loss": 3.9852,
+      "step": 963072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4223182820862406e-05,
+      "loss": 3.9964,
+      "step": 963584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4214796873351886e-05,
+      "loss": 3.9899,
+      "step": 964096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4206410925841366e-05,
+      "loss": 4.0003,
+      "step": 964608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4198041357134575e-05,
+      "loss": 3.9822,
+      "step": 965120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4189655409624055e-05,
+      "loss": 3.9836,
+      "step": 965632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4181269462113535e-05,
+      "loss": 3.9922,
+      "step": 966144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4172899893406744e-05,
+      "loss": 3.9955,
+      "step": 966656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4164513945896224e-05,
+      "loss": 3.9958,
+      "step": 967168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4156127998385704e-05,
+      "loss": 3.9806,
+      "step": 967680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4147742050875184e-05,
+      "loss": 3.9874,
+      "step": 968192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4139356103364664e-05,
+      "loss": 3.9765,
+      "step": 968704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4130970155854144e-05,
+      "loss": 3.9916,
+      "step": 969216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.412260058714736e-05,
+      "loss": 3.9801,
+      "step": 969728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.411421463963684e-05,
+      "loss": 3.9818,
+      "step": 970240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.410582869212632e-05,
+      "loss": 3.991,
+      "step": 970752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.40974427446158e-05,
+      "loss": 3.9865,
+      "step": 971264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.408905679710528e-05,
+      "loss": 3.9884,
+      "step": 971776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.408067084959476e-05,
+      "loss": 3.9875,
+      "step": 972288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.407228490208424e-05,
+      "loss": 3.9728,
+      "step": 972800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.406389895457372e-05,
+      "loss": 3.9943,
+      "step": 973312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.405551300706319e-05,
+      "loss": 3.9801,
+      "step": 973824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.404712705955267e-05,
+      "loss": 3.9887,
+      "step": 974336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403874111204215e-05,
+      "loss": 3.9907,
+      "step": 974848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.403035516453163e-05,
+      "loss": 3.9936,
+      "step": 975360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.402198559582485e-05,
+      "loss": 3.9892,
+      "step": 975872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.401359964831433e-05,
+      "loss": 3.9873,
+      "step": 976384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4005230079607544e-05,
+      "loss": 3.9956,
+      "step": 976896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.399684413209702e-05,
+      "loss": 3.9914,
+      "step": 977408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.39884581845865e-05,
+      "loss": 3.9885,
+      "step": 977920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.398007223707598e-05,
+      "loss": 3.991,
+      "step": 978432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.397168628956546e-05,
+      "loss": 3.9848,
+      "step": 978944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.396330034205494e-05,
+      "loss": 3.996,
+      "step": 979456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.395491439454442e-05,
+      "loss": 3.9753,
+      "step": 979968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3946544825837626e-05,
+      "loss": 3.9802,
+      "step": 980480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3938158878327106e-05,
+      "loss": 3.9794,
+      "step": 980992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3929772930816586e-05,
+      "loss": 3.9834,
+      "step": 981504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3921386983306066e-05,
+      "loss": 3.9921,
+      "step": 982016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.391300103579555e-05,
+      "loss": 3.9826,
+      "step": 982528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.390463146708876e-05,
+      "loss": 3.9951,
+      "step": 983040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.389624551957824e-05,
+      "loss": 3.9824,
+      "step": 983552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.388785957206772e-05,
+      "loss": 3.9923,
+      "step": 984064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.38794736245572e-05,
+      "loss": 3.9933,
+      "step": 984576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.387108767704668e-05,
+      "loss": 3.9958,
+      "step": 985088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.386271810833989e-05,
+      "loss": 3.9884,
+      "step": 985600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.385433216082937e-05,
+      "loss": 3.9892,
+      "step": 986112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.384594621331885e-05,
+      "loss": 3.9852,
+      "step": 986624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.383756026580833e-05,
+      "loss": 3.9974,
+      "step": 987136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382917431829781e-05,
+      "loss": 3.9828,
+      "step": 987648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.382080474959102e-05,
+      "loss": 3.9849,
+      "step": 988160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3812418802080507e-05,
+      "loss": 3.9926,
+      "step": 988672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3804032854569986e-05,
+      "loss": 3.9833,
+      "step": 989184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3795646907059466e-05,
+      "loss": 3.9833,
+      "step": 989696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3787260959548946e-05,
+      "loss": 3.9936,
+      "step": 990208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3778875012038426e-05,
+      "loss": 3.9812,
+      "step": 990720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3770489064527906e-05,
+      "loss": 3.9853,
+      "step": 991232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.376210311701738e-05,
+      "loss": 3.9841,
+      "step": 991744
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.036537170410156,
+      "eval_runtime": 299.0888,
+      "eval_samples_per_second": 1275.845,
+      "eval_steps_per_second": 39.871,
+      "step": 992160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3753733548310595e-05,
+      "loss": 3.9809,
+      "step": 992256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3745347600800075e-05,
+      "loss": 3.9736,
+      "step": 992768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3736961653289555e-05,
+      "loss": 3.9878,
+      "step": 993280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.372857570577903e-05,
+      "loss": 3.9973,
+      "step": 993792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3720206137072244e-05,
+      "loss": 3.9896,
+      "step": 994304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3711820189561724e-05,
+      "loss": 3.9873,
+      "step": 994816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3703434242051204e-05,
+      "loss": 3.9785,
+      "step": 995328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3695048294540684e-05,
+      "loss": 3.9898,
+      "step": 995840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.36866787258339e-05,
+      "loss": 3.9783,
+      "step": 996352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.367829277832338e-05,
+      "loss": 3.989,
+      "step": 996864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366990683081285e-05,
+      "loss": 3.988,
+      "step": 997376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.366152088330233e-05,
+      "loss": 3.9889,
+      "step": 997888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.365315131459555e-05,
+      "loss": 3.9959,
+      "step": 998400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.364476536708503e-05,
+      "loss": 3.9713,
+      "step": 998912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.36363794195745e-05,
+      "loss": 3.9734,
+      "step": 999424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.362799347206398e-05,
+      "loss": 3.9719,
+      "step": 999936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361960752455346e-05,
+      "loss": 3.9965,
+      "step": 1000448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.361123795584668e-05,
+      "loss": 3.9843,
+      "step": 1000960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.360285200833616e-05,
+      "loss": 3.982,
+      "step": 1001472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.359446606082564e-05,
+      "loss": 3.9923,
+      "step": 1001984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.358608011331512e-05,
+      "loss": 4.0008,
+      "step": 1002496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.357771054460833e-05,
+      "loss": 3.9868,
+      "step": 1003008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356932459709781e-05,
+      "loss": 3.983,
+      "step": 1003520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.356093864958729e-05,
+      "loss": 3.9863,
+      "step": 1004032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.355255270207677e-05,
+      "loss": 3.9936,
+      "step": 1004544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3544183133369976e-05,
+      "loss": 3.9832,
+      "step": 1005056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3535797185859456e-05,
+      "loss": 3.973,
+      "step": 1005568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3527411238348936e-05,
+      "loss": 3.9953,
+      "step": 1006080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3519025290838416e-05,
+      "loss": 3.9684,
+      "step": 1006592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3510639343327896e-05,
+      "loss": 3.9779,
+      "step": 1007104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.350226977462111e-05,
+      "loss": 3.9767,
+      "step": 1007616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.349388382711059e-05,
+      "loss": 3.9779,
+      "step": 1008128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.348549787960007e-05,
+      "loss": 3.9891,
+      "step": 1008640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.347711193208955e-05,
+      "loss": 3.9949,
+      "step": 1009152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.346874236338276e-05,
+      "loss": 3.9803,
+      "step": 1009664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.346037279467597e-05,
+      "loss": 3.9812,
+      "step": 1010176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.345198684716545e-05,
+      "loss": 3.9929,
+      "step": 1010688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.344360089965493e-05,
+      "loss": 3.9646,
+      "step": 1011200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.343521495214441e-05,
+      "loss": 3.9785,
+      "step": 1011712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.342682900463389e-05,
+      "loss": 3.977,
+      "step": 1012224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.341844305712337e-05,
+      "loss": 3.9736,
+      "step": 1012736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.341005710961285e-05,
+      "loss": 3.9756,
+      "step": 1013248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.340167116210233e-05,
+      "loss": 3.9917,
+      "step": 1013760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3393285214591816e-05,
+      "loss": 3.9742,
+      "step": 1014272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3384915645885025e-05,
+      "loss": 3.9835,
+      "step": 1014784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3376529698374505e-05,
+      "loss": 3.983,
+      "step": 1015296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3368143750863985e-05,
+      "loss": 3.9803,
+      "step": 1015808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3359757803353465e-05,
+      "loss": 3.983,
+      "step": 1016320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3351388234646674e-05,
+      "loss": 3.9576,
+      "step": 1016832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3343002287136154e-05,
+      "loss": 3.974,
+      "step": 1017344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3334616339625634e-05,
+      "loss": 3.9905,
+      "step": 1017856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3326230392115114e-05,
+      "loss": 3.9846,
+      "step": 1018368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3317844444604594e-05,
+      "loss": 3.969,
+      "step": 1018880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330949125470154e-05,
+      "loss": 3.9699,
+      "step": 1019392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.330110530719102e-05,
+      "loss": 3.9766,
+      "step": 1019904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.32927193596805e-05,
+      "loss": 3.9655,
+      "step": 1020416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.328433341216998e-05,
+      "loss": 3.9982,
+      "step": 1020928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.327594746465946e-05,
+      "loss": 3.9627,
+      "step": 1021440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.326756151714894e-05,
+      "loss": 3.9844,
+      "step": 1021952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325919194844215e-05,
+      "loss": 3.9842,
+      "step": 1022464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.325080600093163e-05,
+      "loss": 3.9609,
+      "step": 1022976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324242005342111e-05,
+      "loss": 3.9719,
+      "step": 1023488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.323403410591059e-05,
+      "loss": 3.9652,
+      "step": 1024000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.322564815840007e-05,
+      "loss": 3.9645,
+      "step": 1024512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.321726221088955e-05,
+      "loss": 3.9697,
+      "step": 1025024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320887626337903e-05,
+      "loss": 3.9909,
+      "step": 1025536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320049031586851e-05,
+      "loss": 3.9792,
+      "step": 1026048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.319212074716172e-05,
+      "loss": 3.9752,
+      "step": 1026560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31837347996512e-05,
+      "loss": 3.9604,
+      "step": 1027072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.317534885214068e-05,
+      "loss": 3.9735,
+      "step": 1027584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.316697928343389e-05,
+      "loss": 3.9783,
+      "step": 1028096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315859333592337e-05,
+      "loss": 3.983,
+      "step": 1028608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.315020738841285e-05,
+      "loss": 3.9744,
+      "step": 1029120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.314182144090233e-05,
+      "loss": 3.9799,
+      "step": 1029632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3133435493391805e-05,
+      "loss": 3.9913,
+      "step": 1030144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3125049545881285e-05,
+      "loss": 3.9725,
+      "step": 1030656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.31166799771745e-05,
+      "loss": 3.9747,
+      "step": 1031168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.310829402966398e-05,
+      "loss": 3.9748,
+      "step": 1031680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309990808215346e-05,
+      "loss": 3.9765,
+      "step": 1032192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309152213464294e-05,
+      "loss": 3.9808,
+      "step": 1032704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.308313618713242e-05,
+      "loss": 3.981,
+      "step": 1033216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.30747502396219e-05,
+      "loss": 3.9844,
+      "step": 1033728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.306636429211138e-05,
+      "loss": 3.9924,
+      "step": 1034240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.305797834460086e-05,
+      "loss": 3.9841,
+      "step": 1034752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.304960877589407e-05,
+      "loss": 3.9646,
+      "step": 1035264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.304122282838355e-05,
+      "loss": 3.9708,
+      "step": 1035776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.303283688087303e-05,
+      "loss": 3.9843,
+      "step": 1036288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.302445093336251e-05,
+      "loss": 3.9835,
+      "step": 1036800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.301608136465572e-05,
+      "loss": 3.977,
+      "step": 1037312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.300771179594893e-05,
+      "loss": 3.9777,
+      "step": 1037824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2999325848438415e-05,
+      "loss": 3.9851,
+      "step": 1038336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2990939900927895e-05,
+      "loss": 3.9705,
+      "step": 1038848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2982553953417375e-05,
+      "loss": 3.97,
+      "step": 1039360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2974168005906855e-05,
+      "loss": 3.9833,
+      "step": 1039872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2965782058396335e-05,
+      "loss": 3.9742,
+      "step": 1040384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2957396110885815e-05,
+      "loss": 3.9883,
+      "step": 1040896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2949010163375295e-05,
+      "loss": 3.9665,
+      "step": 1041408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2940640594668504e-05,
+      "loss": 3.9678,
+      "step": 1041920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2932254647157984e-05,
+      "loss": 3.9797,
+      "step": 1042432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2923868699647464e-05,
+      "loss": 3.9816,
+      "step": 1042944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2915482752136944e-05,
+      "loss": 3.9815,
+      "step": 1043456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.290711318343015e-05,
+      "loss": 3.9677,
+      "step": 1043968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289872723591963e-05,
+      "loss": 3.9733,
+      "step": 1044480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.289034128840911e-05,
+      "loss": 3.9612,
+      "step": 1044992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.288195534089859e-05,
+      "loss": 3.979,
+      "step": 1045504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.287358577219181e-05,
+      "loss": 3.9652,
+      "step": 1046016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.286521620348502e-05,
+      "loss": 3.9682,
+      "step": 1046528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.28568302559745e-05,
+      "loss": 3.9805,
+      "step": 1047040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.284844430846398e-05,
+      "loss": 3.9717,
+      "step": 1047552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.284005836095346e-05,
+      "loss": 3.9679,
+      "step": 1048064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.283167241344294e-05,
+      "loss": 3.975,
+      "step": 1048576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.282328646593242e-05,
+      "loss": 3.9599,
+      "step": 1049088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.28149005184219e-05,
+      "loss": 3.9783,
+      "step": 1049600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.280651457091138e-05,
+      "loss": 3.9681,
+      "step": 1050112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.279812862340086e-05,
+      "loss": 3.9735,
+      "step": 1050624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278974267589034e-05,
+      "loss": 3.9788,
+      "step": 1051136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.278138948598728e-05,
+      "loss": 3.9773,
+      "step": 1051648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.277300353847676e-05,
+      "loss": 3.9727,
+      "step": 1052160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.276463396976997e-05,
+      "loss": 3.9743,
+      "step": 1052672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.275624802225945e-05,
+      "loss": 3.9853,
+      "step": 1053184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.274786207474893e-05,
+      "loss": 3.9727,
+      "step": 1053696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273947612723841e-05,
+      "loss": 3.9747,
+      "step": 1054208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.273109017972789e-05,
+      "loss": 3.9755,
+      "step": 1054720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.272270423221737e-05,
+      "loss": 3.9752,
+      "step": 1055232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.271431828470685e-05,
+      "loss": 3.9801,
+      "step": 1055744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.270593233719633e-05,
+      "loss": 3.9643,
+      "step": 1056256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.269754638968581e-05,
+      "loss": 3.965,
+      "step": 1056768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268916044217529e-05,
+      "loss": 3.9673,
+      "step": 1057280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.268077449466477e-05,
+      "loss": 3.966,
+      "step": 1057792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.267238854715425e-05,
+      "loss": 3.9737,
+      "step": 1058304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.266400259964373e-05,
+      "loss": 3.9742,
+      "step": 1058816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.265563303093694e-05,
+      "loss": 3.9786,
+      "step": 1059328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.264724708342642e-05,
+      "loss": 3.9689,
+      "step": 1059840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.26388611359159e-05,
+      "loss": 3.9794,
+      "step": 1060352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.263047518840538e-05,
+      "loss": 3.9761,
+      "step": 1060864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.262210561969859e-05,
+      "loss": 3.9843,
+      "step": 1061376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.261371967218807e-05,
+      "loss": 3.9712,
+      "step": 1061888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.260533372467755e-05,
+      "loss": 3.9771,
+      "step": 1062400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2596964155970764e-05,
+      "loss": 3.9741,
+      "step": 1062912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.258857820846024e-05,
+      "loss": 3.9826,
+      "step": 1063424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2580192260949724e-05,
+      "loss": 3.9691,
+      "step": 1063936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2571806313439204e-05,
+      "loss": 3.9696,
+      "step": 1064448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2563420365928684e-05,
+      "loss": 3.9801,
+      "step": 1064960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2555034418418164e-05,
+      "loss": 3.9694,
+      "step": 1065472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2546648470907644e-05,
+      "loss": 3.9701,
+      "step": 1065984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2538262523397124e-05,
+      "loss": 3.983,
+      "step": 1066496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.252989295469033e-05,
+      "loss": 3.9619,
+      "step": 1067008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.252150700717981e-05,
+      "loss": 3.9758,
+      "step": 1067520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.251312105966929e-05,
+      "loss": 3.9663,
+      "step": 1068032
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.028872966766357,
+      "eval_runtime": 295.3047,
+      "eval_samples_per_second": 1292.194,
+      "eval_steps_per_second": 40.382,
+      "step": 1068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.250473511215877e-05,
+      "loss": 3.9627,
+      "step": 1068544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.249634916464825e-05,
+      "loss": 3.9629,
+      "step": 1069056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.248796321713773e-05,
+      "loss": 3.97,
+      "step": 1069568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247957726962721e-05,
+      "loss": 3.9848,
+      "step": 1070080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.247119132211669e-05,
+      "loss": 3.9778,
+      "step": 1070592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.246280537460617e-05,
+      "loss": 3.9721,
+      "step": 1071104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.245441942709565e-05,
+      "loss": 3.9619,
+      "step": 1071616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.244603347958513e-05,
+      "loss": 3.9752,
+      "step": 1072128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.243764753207461e-05,
+      "loss": 3.9695,
+      "step": 1072640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242926158456409e-05,
+      "loss": 3.9712,
+      "step": 1073152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.242087563705357e-05,
+      "loss": 3.9745,
+      "step": 1073664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.241248968954305e-05,
+      "loss": 3.9759,
+      "step": 1074176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.240412012083626e-05,
+      "loss": 3.9809,
+      "step": 1074688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.239573417332574e-05,
+      "loss": 3.9588,
+      "step": 1075200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.238734822581522e-05,
+      "loss": 3.9563,
+      "step": 1075712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.23789622783047e-05,
+      "loss": 3.9592,
+      "step": 1076224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237057633079418e-05,
+      "loss": 3.98,
+      "step": 1076736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.236219038328366e-05,
+      "loss": 3.9731,
+      "step": 1077248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.235380443577314e-05,
+      "loss": 3.9672,
+      "step": 1077760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.234541848826262e-05,
+      "loss": 3.977,
+      "step": 1078272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.23370325407521e-05,
+      "loss": 3.9884,
+      "step": 1078784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.232864659324159e-05,
+      "loss": 3.9697,
+      "step": 1079296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.232026064573107e-05,
+      "loss": 3.9722,
+      "step": 1079808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.231187469822054e-05,
+      "loss": 3.974,
+      "step": 1080320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.230348875071002e-05,
+      "loss": 3.9831,
+      "step": 1080832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.22951028031995e-05,
+      "loss": 3.9656,
+      "step": 1081344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.228671685568898e-05,
+      "loss": 3.9627,
+      "step": 1081856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.227834728698219e-05,
+      "loss": 3.9791,
+      "step": 1082368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.226996133947167e-05,
+      "loss": 3.9591,
+      "step": 1082880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.226157539196115e-05,
+      "loss": 3.9622,
+      "step": 1083392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.225318944445063e-05,
+      "loss": 3.9599,
+      "step": 1083904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.224481987574384e-05,
+      "loss": 3.9692,
+      "step": 1084416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2236433928233326e-05,
+      "loss": 3.9737,
+      "step": 1084928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.222806435952654e-05,
+      "loss": 3.9809,
+      "step": 1085440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2219678412016015e-05,
+      "loss": 3.9705,
+      "step": 1085952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.221130884330923e-05,
+      "loss": 3.9663,
+      "step": 1086464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.220292289579871e-05,
+      "loss": 3.9774,
+      "step": 1086976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.219453694828819e-05,
+      "loss": 3.9554,
+      "step": 1087488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2186151000777664e-05,
+      "loss": 3.9634,
+      "step": 1088000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2177765053267144e-05,
+      "loss": 3.9641,
+      "step": 1088512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2169379105756624e-05,
+      "loss": 3.9638,
+      "step": 1089024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2160993158246104e-05,
+      "loss": 3.959,
+      "step": 1089536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2152607210735584e-05,
+      "loss": 3.9788,
+      "step": 1090048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2144221263225064e-05,
+      "loss": 3.9596,
+      "step": 1090560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.213585169451828e-05,
+      "loss": 3.9702,
+      "step": 1091072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.212746574700776e-05,
+      "loss": 3.9688,
+      "step": 1091584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.211907979949724e-05,
+      "loss": 3.9656,
+      "step": 1092096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.211069385198672e-05,
+      "loss": 3.9692,
+      "step": 1092608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.21023079044762e-05,
+      "loss": 3.9472,
+      "step": 1093120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.209392195696568e-05,
+      "loss": 3.961,
+      "step": 1093632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.208553600945516e-05,
+      "loss": 3.975,
+      "step": 1094144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.207715006194464e-05,
+      "loss": 3.9663,
+      "step": 1094656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.206878049323785e-05,
+      "loss": 3.9575,
+      "step": 1095168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.206041092453106e-05,
+      "loss": 3.9573,
+      "step": 1095680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.205202497702054e-05,
+      "loss": 3.9653,
+      "step": 1096192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.204363902951002e-05,
+      "loss": 3.9498,
+      "step": 1096704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.20352530819995e-05,
+      "loss": 3.9819,
+      "step": 1097216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.202686713448898e-05,
+      "loss": 3.9555,
+      "step": 1097728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2018481186978464e-05,
+      "loss": 3.9651,
+      "step": 1098240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2010095239467944e-05,
+      "loss": 3.9733,
+      "step": 1098752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.200172567076115e-05,
+      "loss": 3.9501,
+      "step": 1099264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.199333972325063e-05,
+      "loss": 3.9553,
+      "step": 1099776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.198495377574011e-05,
+      "loss": 3.958,
+      "step": 1100288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.197658420703332e-05,
+      "loss": 3.9496,
+      "step": 1100800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19681982595228e-05,
+      "loss": 3.9537,
+      "step": 1101312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195981231201228e-05,
+      "loss": 3.9773,
+      "step": 1101824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.195142636450176e-05,
+      "loss": 3.9662,
+      "step": 1102336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.194305679579497e-05,
+      "loss": 3.9617,
+      "step": 1102848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.193467084828445e-05,
+      "loss": 3.9502,
+      "step": 1103360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.192628490077393e-05,
+      "loss": 3.9571,
+      "step": 1103872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.191789895326342e-05,
+      "loss": 3.9643,
+      "step": 1104384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.19095130057529e-05,
+      "loss": 3.9731,
+      "step": 1104896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.190112705824238e-05,
+      "loss": 3.9614,
+      "step": 1105408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.189274111073185e-05,
+      "loss": 3.968,
+      "step": 1105920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.188435516322133e-05,
+      "loss": 3.9752,
+      "step": 1106432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.187596921571081e-05,
+      "loss": 3.9651,
+      "step": 1106944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1867599647004027e-05,
+      "loss": 3.9584,
+      "step": 1107456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.18592136994935e-05,
+      "loss": 3.9683,
+      "step": 1107968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1850844130786716e-05,
+      "loss": 3.9637,
+      "step": 1108480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1842458183276196e-05,
+      "loss": 3.9656,
+      "step": 1108992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1834072235765675e-05,
+      "loss": 3.9681,
+      "step": 1109504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1825686288255155e-05,
+      "loss": 3.9713,
+      "step": 1110016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1817300340744635e-05,
+      "loss": 3.9799,
+      "step": 1110528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1808914393234115e-05,
+      "loss": 3.9702,
+      "step": 1111040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1800544824527324e-05,
+      "loss": 3.9514,
+      "step": 1111552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1792158877016804e-05,
+      "loss": 3.9609,
+      "step": 1112064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1783772929506284e-05,
+      "loss": 3.9734,
+      "step": 1112576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1775386981995764e-05,
+      "loss": 3.9679,
+      "step": 1113088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1767001034485244e-05,
+      "loss": 3.9648,
+      "step": 1113600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.175863146577845e-05,
+      "loss": 3.9639,
+      "step": 1114112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.175024551826793e-05,
+      "loss": 3.9694,
+      "step": 1114624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.174185957075741e-05,
+      "loss": 3.9594,
+      "step": 1115136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.173347362324689e-05,
+      "loss": 3.9622,
+      "step": 1115648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.17251040545401e-05,
+      "loss": 3.9611,
+      "step": 1116160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.171671810702959e-05,
+      "loss": 3.9641,
+      "step": 1116672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.170833215951907e-05,
+      "loss": 3.97,
+      "step": 1117184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.169996259081228e-05,
+      "loss": 3.964,
+      "step": 1117696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.169157664330176e-05,
+      "loss": 3.9505,
+      "step": 1118208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.168319069579124e-05,
+      "loss": 3.9658,
+      "step": 1118720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.167480474828072e-05,
+      "loss": 3.9673,
+      "step": 1119232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.16664188007702e-05,
+      "loss": 3.9685,
+      "step": 1119744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.165803285325968e-05,
+      "loss": 3.9585,
+      "step": 1120256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.164964690574916e-05,
+      "loss": 3.9571,
+      "step": 1120768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.164126095823864e-05,
+      "loss": 3.952,
+      "step": 1121280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.163287501072812e-05,
+      "loss": 3.9635,
+      "step": 1121792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.16244890632176e-05,
+      "loss": 3.9527,
+      "step": 1122304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.161610311570708e-05,
+      "loss": 3.9534,
+      "step": 1122816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.160771716819656e-05,
+      "loss": 3.9662,
+      "step": 1123328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159933122068604e-05,
+      "loss": 3.9622,
+      "step": 1123840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.159097803078298e-05,
+      "loss": 3.9568,
+      "step": 1124352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.158259208327246e-05,
+      "loss": 3.9602,
+      "step": 1124864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.157420613576194e-05,
+      "loss": 3.9499,
+      "step": 1125376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.156582018825142e-05,
+      "loss": 3.9657,
+      "step": 1125888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.15574342407409e-05,
+      "loss": 3.9558,
+      "step": 1126400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154904829323038e-05,
+      "loss": 3.9606,
+      "step": 1126912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.154066234571986e-05,
+      "loss": 3.9648,
+      "step": 1127424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1532276398209335e-05,
+      "loss": 3.967,
+      "step": 1127936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1523890450698815e-05,
+      "loss": 3.9617,
+      "step": 1128448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.151552088199203e-05,
+      "loss": 3.9618,
+      "step": 1128960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.150713493448151e-05,
+      "loss": 3.9693,
+      "step": 1129472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.149874898697099e-05,
+      "loss": 3.9623,
+      "step": 1129984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.149036303946047e-05,
+      "loss": 3.9565,
+      "step": 1130496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.148199347075369e-05,
+      "loss": 3.9691,
+      "step": 1131008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.147360752324316e-05,
+      "loss": 3.9623,
+      "step": 1131520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.146522157573264e-05,
+      "loss": 3.9686,
+      "step": 1132032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.145683562822212e-05,
+      "loss": 3.9522,
+      "step": 1132544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.14484496807116e-05,
+      "loss": 3.9512,
+      "step": 1133056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.144006373320108e-05,
+      "loss": 3.9541,
+      "step": 1133568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.143169416449429e-05,
+      "loss": 3.9585,
+      "step": 1134080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.142330821698377e-05,
+      "loss": 3.9532,
+      "step": 1134592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.141492226947325e-05,
+      "loss": 3.9644,
+      "step": 1135104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.140653632196273e-05,
+      "loss": 3.9656,
+      "step": 1135616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1398166753255945e-05,
+      "loss": 3.9549,
+      "step": 1136128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1389780805745425e-05,
+      "loss": 3.966,
+      "step": 1136640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1381394858234905e-05,
+      "loss": 3.9678,
+      "step": 1137152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1373008910724385e-05,
+      "loss": 3.9706,
+      "step": 1137664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1364622963213865e-05,
+      "loss": 3.9577,
+      "step": 1138176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1356237015703345e-05,
+      "loss": 3.9642,
+      "step": 1138688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1347867446996554e-05,
+      "loss": 3.9658,
+      "step": 1139200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1339481499486034e-05,
+      "loss": 3.9667,
+      "step": 1139712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1331095551975514e-05,
+      "loss": 3.957,
+      "step": 1140224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1322709604464994e-05,
+      "loss": 3.9588,
+      "step": 1140736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1314323656954474e-05,
+      "loss": 3.9663,
+      "step": 1141248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1305937709443954e-05,
+      "loss": 3.9557,
+      "step": 1141760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1297551761933434e-05,
+      "loss": 3.9576,
+      "step": 1142272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1289165814422913e-05,
+      "loss": 3.9697,
+      "step": 1142784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.128079624571613e-05,
+      "loss": 3.9533,
+      "step": 1143296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.127241029820561e-05,
+      "loss": 3.9594,
+      "step": 1143808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.126402435069509e-05,
+      "loss": 3.9576,
+      "step": 1144320
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.022188186645508,
+      "eval_runtime": 290.7323,
+      "eval_samples_per_second": 1312.517,
+      "eval_steps_per_second": 41.017,
+      "step": 1144800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.125563840318457e-05,
+      "loss": 3.9522,
+      "step": 1144832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.124725245567405e-05,
+      "loss": 3.9515,
+      "step": 1145344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.123888288696726e-05,
+      "loss": 3.9574,
+      "step": 1145856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.123049693945674e-05,
+      "loss": 3.9724,
+      "step": 1146368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.122212737074995e-05,
+      "loss": 3.9618,
+      "step": 1146880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.121374142323943e-05,
+      "loss": 3.9646,
+      "step": 1147392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.120535547572891e-05,
+      "loss": 3.9532,
+      "step": 1147904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.119696952821839e-05,
+      "loss": 3.9615,
+      "step": 1148416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.118858358070787e-05,
+      "loss": 3.9541,
+      "step": 1148928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.118019763319735e-05,
+      "loss": 3.9619,
+      "step": 1149440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.117182806449056e-05,
+      "loss": 3.9595,
+      "step": 1149952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.116344211698004e-05,
+      "loss": 3.9671,
+      "step": 1150464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.115507254827325e-05,
+      "loss": 3.9739,
+      "step": 1150976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.114668660076273e-05,
+      "loss": 3.9488,
+      "step": 1151488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.113830065325221e-05,
+      "loss": 3.9445,
+      "step": 1152000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.112991470574169e-05,
+      "loss": 3.9431,
+      "step": 1152512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1121528758231165e-05,
+      "loss": 3.9629,
+      "step": 1153024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1113142810720645e-05,
+      "loss": 3.9637,
+      "step": 1153536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1104756863210125e-05,
+      "loss": 3.9589,
+      "step": 1154048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1096370915699605e-05,
+      "loss": 3.9616,
+      "step": 1154560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1087984968189085e-05,
+      "loss": 3.9826,
+      "step": 1155072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1079599020678565e-05,
+      "loss": 3.9574,
+      "step": 1155584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.107122945197178e-05,
+      "loss": 3.9588,
+      "step": 1156096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.106284350446126e-05,
+      "loss": 3.9618,
+      "step": 1156608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.105445755695074e-05,
+      "loss": 3.9704,
+      "step": 1157120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.104607160944022e-05,
+      "loss": 3.9497,
+      "step": 1157632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.103770204073343e-05,
+      "loss": 3.954,
+      "step": 1158144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.102931609322291e-05,
+      "loss": 3.9617,
+      "step": 1158656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.102093014571239e-05,
+      "loss": 3.954,
+      "step": 1159168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.101254419820187e-05,
+      "loss": 3.9491,
+      "step": 1159680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.100415825069135e-05,
+      "loss": 3.9463,
+      "step": 1160192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.099577230318083e-05,
+      "loss": 3.9552,
+      "step": 1160704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.098738635567031e-05,
+      "loss": 3.9653,
+      "step": 1161216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.097900040815979e-05,
+      "loss": 3.9666,
+      "step": 1161728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.097061446064927e-05,
+      "loss": 3.9605,
+      "step": 1162240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0962244891942485e-05,
+      "loss": 3.9519,
+      "step": 1162752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0953858944431965e-05,
+      "loss": 3.9646,
+      "step": 1163264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0945472996921445e-05,
+      "loss": 3.9437,
+      "step": 1163776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0937087049410925e-05,
+      "loss": 3.9526,
+      "step": 1164288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0928701101900405e-05,
+      "loss": 3.9521,
+      "step": 1164800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0920315154389885e-05,
+      "loss": 3.9519,
+      "step": 1165312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0911945585683094e-05,
+      "loss": 3.9488,
+      "step": 1165824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0903559638172574e-05,
+      "loss": 3.9664,
+      "step": 1166336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0895173690662054e-05,
+      "loss": 3.9484,
+      "step": 1166848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.088678774315153e-05,
+      "loss": 3.9562,
+      "step": 1167360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.087840179564101e-05,
+      "loss": 3.9591,
+      "step": 1167872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.087001584813049e-05,
+      "loss": 3.9533,
+      "step": 1168384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0861629900619974e-05,
+      "loss": 3.9582,
+      "step": 1168896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.085326033191318e-05,
+      "loss": 3.9372,
+      "step": 1169408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.084487438440266e-05,
+      "loss": 3.9465,
+      "step": 1169920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.083648843689214e-05,
+      "loss": 3.9621,
+      "step": 1170432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.082810248938162e-05,
+      "loss": 3.9565,
+      "step": 1170944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.08197165418711e-05,
+      "loss": 3.9459,
+      "step": 1171456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.081133059436058e-05,
+      "loss": 3.9454,
+      "step": 1171968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.080294464685006e-05,
+      "loss": 3.9527,
+      "step": 1172480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.079455869933954e-05,
+      "loss": 3.942,
+      "step": 1172992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.078618913063275e-05,
+      "loss": 3.9662,
+      "step": 1173504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.077780318312223e-05,
+      "loss": 3.9458,
+      "step": 1174016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.076941723561171e-05,
+      "loss": 3.9522,
+      "step": 1174528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.076103128810119e-05,
+      "loss": 3.9631,
+      "step": 1175040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.075264534059067e-05,
+      "loss": 3.9364,
+      "step": 1175552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.074427577188389e-05,
+      "loss": 3.9418,
+      "step": 1176064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.073588982437337e-05,
+      "loss": 3.9492,
+      "step": 1176576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.072750387686285e-05,
+      "loss": 3.9372,
+      "step": 1177088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.071911792935233e-05,
+      "loss": 3.9447,
+      "step": 1177600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.071073198184181e-05,
+      "loss": 3.9619,
+      "step": 1178112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.070234603433129e-05,
+      "loss": 3.9577,
+      "step": 1178624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.069396008682077e-05,
+      "loss": 3.9502,
+      "step": 1179136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.068557413931025e-05,
+      "loss": 3.9359,
+      "step": 1179648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0677204570603456e-05,
+      "loss": 3.9468,
+      "step": 1180160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0668818623092936e-05,
+      "loss": 3.9531,
+      "step": 1180672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0660432675582416e-05,
+      "loss": 3.957,
+      "step": 1181184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.065204672807189e-05,
+      "loss": 3.953,
+      "step": 1181696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0643677159365105e-05,
+      "loss": 3.9552,
+      "step": 1182208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.063529121185459e-05,
+      "loss": 3.9645,
+      "step": 1182720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.062690526434407e-05,
+      "loss": 3.9538,
+      "step": 1183232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0618519316833545e-05,
+      "loss": 3.9411,
+      "step": 1183744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.061014974812676e-05,
+      "loss": 3.9592,
+      "step": 1184256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.060176380061624e-05,
+      "loss": 3.9524,
+      "step": 1184768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.059339423190945e-05,
+      "loss": 3.9489,
+      "step": 1185280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.058500828439893e-05,
+      "loss": 3.96,
+      "step": 1185792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.057662233688841e-05,
+      "loss": 3.9573,
+      "step": 1186304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.056823638937789e-05,
+      "loss": 3.9699,
+      "step": 1186816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.055985044186736e-05,
+      "loss": 3.9592,
+      "step": 1187328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.055148087316058e-05,
+      "loss": 3.9427,
+      "step": 1187840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.054309492565006e-05,
+      "loss": 3.9462,
+      "step": 1188352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0534708978139546e-05,
+      "loss": 3.9542,
+      "step": 1188864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.052632303062902e-05,
+      "loss": 3.9592,
+      "step": 1189376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.05179370831185e-05,
+      "loss": 3.9551,
+      "step": 1189888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.0509567514411715e-05,
+      "loss": 3.9515,
+      "step": 1190400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0501181566901188e-05,
+      "loss": 3.9584,
+      "step": 1190912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0492795619390668e-05,
+      "loss": 3.9472,
+      "step": 1191424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0484409671880148e-05,
+      "loss": 3.9502,
+      "step": 1191936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0476040103173363e-05,
+      "loss": 3.9479,
+      "step": 1192448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.046765415566284e-05,
+      "loss": 3.9559,
+      "step": 1192960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.045926820815232e-05,
+      "loss": 3.9611,
+      "step": 1193472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.04508822606418e-05,
+      "loss": 3.9523,
+      "step": 1193984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.044249631313128e-05,
+      "loss": 3.9405,
+      "step": 1194496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.043411036562076e-05,
+      "loss": 3.9521,
+      "step": 1195008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.042572441811024e-05,
+      "loss": 3.955,
+      "step": 1195520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.041733847059972e-05,
+      "loss": 3.9608,
+      "step": 1196032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0408968901892932e-05,
+      "loss": 3.9443,
+      "step": 1196544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0400582954382412e-05,
+      "loss": 3.9478,
+      "step": 1197056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.039221338567562e-05,
+      "loss": 3.9387,
+      "step": 1197568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.03838274381651e-05,
+      "loss": 3.9536,
+      "step": 1198080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.037544149065458e-05,
+      "loss": 3.9409,
+      "step": 1198592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.036705554314406e-05,
+      "loss": 3.9447,
+      "step": 1199104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0358669595633545e-05,
+      "loss": 3.9542,
+      "step": 1199616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0350283648123024e-05,
+      "loss": 3.944,
+      "step": 1200128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0341897700612504e-05,
+      "loss": 3.9464,
+      "step": 1200640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0333528131905714e-05,
+      "loss": 3.9469,
+      "step": 1201152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0325142184395194e-05,
+      "loss": 3.9417,
+      "step": 1201664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0316756236884673e-05,
+      "loss": 3.9544,
+      "step": 1202176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0308370289374153e-05,
+      "loss": 3.9426,
+      "step": 1202688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0300000720667366e-05,
+      "loss": 3.9495,
+      "step": 1203200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0291614773156846e-05,
+      "loss": 3.9482,
+      "step": 1203712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0283228825646326e-05,
+      "loss": 3.9573,
+      "step": 1204224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0274842878135806e-05,
+      "loss": 3.949,
+      "step": 1204736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0266456930625286e-05,
+      "loss": 3.9522,
+      "step": 1205248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0258070983114766e-05,
+      "loss": 3.9564,
+      "step": 1205760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0249685035604246e-05,
+      "loss": 3.9544,
+      "step": 1206272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0241315466897458e-05,
+      "loss": 3.9384,
+      "step": 1206784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0232929519386938e-05,
+      "loss": 3.9576,
+      "step": 1207296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0224543571876418e-05,
+      "loss": 3.9508,
+      "step": 1207808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0216157624365898e-05,
+      "loss": 3.9598,
+      "step": 1208320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0207771676855378e-05,
+      "loss": 3.941,
+      "step": 1208832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0199385729344855e-05,
+      "loss": 3.9417,
+      "step": 1209344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0190999781834334e-05,
+      "loss": 3.9406,
+      "step": 1209856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0182613834323814e-05,
+      "loss": 3.9465,
+      "step": 1210368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0174227886813294e-05,
+      "loss": 3.94,
+      "step": 1210880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0165841939302774e-05,
+      "loss": 3.9564,
+      "step": 1211392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0157472370595983e-05,
+      "loss": 3.9559,
+      "step": 1211904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0149086423085467e-05,
+      "loss": 3.946,
+      "step": 1212416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0140700475574947e-05,
+      "loss": 3.9493,
+      "step": 1212928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0132330906868156e-05,
+      "loss": 3.9546,
+      "step": 1213440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0123944959357636e-05,
+      "loss": 3.9587,
+      "step": 1213952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0115559011847116e-05,
+      "loss": 3.9531,
+      "step": 1214464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0107173064336596e-05,
+      "loss": 3.9528,
+      "step": 1214976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0098787116826076e-05,
+      "loss": 3.955,
+      "step": 1215488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.009040116931556e-05,
+      "loss": 3.9542,
+      "step": 1216000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.008201522180504e-05,
+      "loss": 3.9475,
+      "step": 1216512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.007362927429452e-05,
+      "loss": 3.947,
+      "step": 1217024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0065259705587728e-05,
+      "loss": 3.9563,
+      "step": 1217536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.0056873758077208e-05,
+      "loss": 3.9455,
+      "step": 1218048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.004850418937042e-05,
+      "loss": 3.9424,
+      "step": 1218560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.00401182418599e-05,
+      "loss": 3.9602,
+      "step": 1219072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.003173229434938e-05,
+      "loss": 3.9453,
+      "step": 1219584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.002334634683886e-05,
+      "loss": 3.9466,
+      "step": 1220096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.001496039932834e-05,
+      "loss": 3.9451,
+      "step": 1220608
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.000657445181782e-05,
+      "loss": 3.9434,
+      "step": 1221120
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.016406059265137,
+      "eval_runtime": 290.5031,
+      "eval_samples_per_second": 1313.552,
+      "eval_steps_per_second": 41.049,
+      "step": 1221120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99981885043073e-05,
+      "loss": 3.9351,
+      "step": 1221632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.998980255679678e-05,
+      "loss": 3.9476,
+      "step": 1222144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.998141660928626e-05,
+      "loss": 3.9581,
+      "step": 1222656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9973030661775737e-05,
+      "loss": 3.9524,
+      "step": 1223168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9964644714265217e-05,
+      "loss": 3.9542,
+      "step": 1223680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9956258766754697e-05,
+      "loss": 3.9433,
+      "step": 1224192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9947872819244177e-05,
+      "loss": 3.9487,
+      "step": 1224704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9939486871733656e-05,
+      "loss": 3.9406,
+      "step": 1225216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9931100924223136e-05,
+      "loss": 3.953,
+      "step": 1225728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9922714976712616e-05,
+      "loss": 3.9484,
+      "step": 1226240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99143290292021e-05,
+      "loss": 3.9557,
+      "step": 1226752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.990594308169158e-05,
+      "loss": 3.9544,
+      "step": 1227264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.989757351298479e-05,
+      "loss": 3.943,
+      "step": 1227776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.988918756547427e-05,
+      "loss": 3.9333,
+      "step": 1228288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.988080161796375e-05,
+      "loss": 3.9378,
+      "step": 1228800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.987241567045323e-05,
+      "loss": 3.9416,
+      "step": 1229312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.986402972294271e-05,
+      "loss": 3.9512,
+      "step": 1229824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9855643775432192e-05,
+      "loss": 3.9448,
+      "step": 1230336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9847257827921672e-05,
+      "loss": 3.9515,
+      "step": 1230848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9838871880411152e-05,
+      "loss": 3.9748,
+      "step": 1231360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9830485932900625e-05,
+      "loss": 3.9464,
+      "step": 1231872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9822099985390105e-05,
+      "loss": 3.9461,
+      "step": 1232384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9813714037879585e-05,
+      "loss": 3.95,
+      "step": 1232896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9805328090369068e-05,
+      "loss": 3.9601,
+      "step": 1233408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9796958521662277e-05,
+      "loss": 3.9401,
+      "step": 1233920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9788572574151757e-05,
+      "loss": 3.9403,
+      "step": 1234432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9780203005444973e-05,
+      "loss": 3.9527,
+      "step": 1234944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9771817057934453e-05,
+      "loss": 3.9423,
+      "step": 1235456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.976343111042393e-05,
+      "loss": 3.9331,
+      "step": 1235968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.975504516291341e-05,
+      "loss": 3.9345,
+      "step": 1236480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9746675594206626e-05,
+      "loss": 3.9441,
+      "step": 1236992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.97382896466961e-05,
+      "loss": 3.9512,
+      "step": 1237504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.972990369918558e-05,
+      "loss": 3.9605,
+      "step": 1238016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.972151775167506e-05,
+      "loss": 3.9431,
+      "step": 1238528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9713148182968275e-05,
+      "loss": 3.9421,
+      "step": 1239040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.970476223545775e-05,
+      "loss": 3.9533,
+      "step": 1239552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.969637628794723e-05,
+      "loss": 3.9319,
+      "step": 1240064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.968799034043671e-05,
+      "loss": 3.9436,
+      "step": 1240576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.967960439292619e-05,
+      "loss": 3.9389,
+      "step": 1241088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.967121844541567e-05,
+      "loss": 3.9409,
+      "step": 1241600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.966283249790515e-05,
+      "loss": 3.9346,
+      "step": 1242112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9654462929198363e-05,
+      "loss": 3.9565,
+      "step": 1242624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9646076981687843e-05,
+      "loss": 3.9373,
+      "step": 1243136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9637691034177323e-05,
+      "loss": 3.9441,
+      "step": 1243648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9629305086666803e-05,
+      "loss": 3.9467,
+      "step": 1244160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9620919139156283e-05,
+      "loss": 3.9442,
+      "step": 1244672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9612533191645763e-05,
+      "loss": 3.9487,
+      "step": 1245184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9604147244135243e-05,
+      "loss": 3.9309,
+      "step": 1245696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9595761296624723e-05,
+      "loss": 3.9317,
+      "step": 1246208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9587375349114206e-05,
+      "loss": 3.9519,
+      "step": 1246720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9579005780407415e-05,
+      "loss": 3.9448,
+      "step": 1247232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9570619832896895e-05,
+      "loss": 3.9323,
+      "step": 1247744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9562250264190105e-05,
+      "loss": 3.9348,
+      "step": 1248256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9553864316679584e-05,
+      "loss": 3.9422,
+      "step": 1248768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9545478369169068e-05,
+      "loss": 3.9323,
+      "step": 1249280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9537092421658548e-05,
+      "loss": 3.9495,
+      "step": 1249792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9528706474148028e-05,
+      "loss": 3.9331,
+      "step": 1250304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9520320526637508e-05,
+      "loss": 3.942,
+      "step": 1250816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9511950957930717e-05,
+      "loss": 3.9487,
+      "step": 1251328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9503565010420197e-05,
+      "loss": 3.9362,
+      "step": 1251840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9495179062909677e-05,
+      "loss": 3.9274,
+      "step": 1252352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9486793115399157e-05,
+      "loss": 3.9375,
+      "step": 1252864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.947842354669237e-05,
+      "loss": 3.9276,
+      "step": 1253376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.947003759918185e-05,
+      "loss": 3.9329,
+      "step": 1253888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.946165165167133e-05,
+      "loss": 3.9467,
+      "step": 1254400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.945326570416081e-05,
+      "loss": 3.9484,
+      "step": 1254912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.944487975665029e-05,
+      "loss": 3.9372,
+      "step": 1255424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9436493809139766e-05,
+      "loss": 3.9298,
+      "step": 1255936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9428107861629245e-05,
+      "loss": 3.931,
+      "step": 1256448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9419721914118725e-05,
+      "loss": 3.9405,
+      "step": 1256960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.941136872421567e-05,
+      "loss": 3.9448,
+      "step": 1257472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.940298277670515e-05,
+      "loss": 3.9429,
+      "step": 1257984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.939459682919463e-05,
+      "loss": 3.9417,
+      "step": 1258496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.938621088168411e-05,
+      "loss": 3.9506,
+      "step": 1259008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9377824934173587e-05,
+      "loss": 3.9479,
+      "step": 1259520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9369455365466803e-05,
+      "loss": 3.9283,
+      "step": 1260032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9361069417956283e-05,
+      "loss": 3.9445,
+      "step": 1260544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.935268347044576e-05,
+      "loss": 3.9451,
+      "step": 1261056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9344313901738972e-05,
+      "loss": 3.936,
+      "step": 1261568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9335927954228455e-05,
+      "loss": 3.95,
+      "step": 1262080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9327542006717935e-05,
+      "loss": 3.9469,
+      "step": 1262592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9319156059207408e-05,
+      "loss": 3.9566,
+      "step": 1263104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9310770111696888e-05,
+      "loss": 3.9485,
+      "step": 1263616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9302384164186368e-05,
+      "loss": 3.9291,
+      "step": 1264128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9293998216675848e-05,
+      "loss": 3.9386,
+      "step": 1264640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.928561226916533e-05,
+      "loss": 3.9437,
+      "step": 1265152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.927722632165481e-05,
+      "loss": 3.9466,
+      "step": 1265664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.926884037414429e-05,
+      "loss": 3.9434,
+      "step": 1266176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.92604708054375e-05,
+      "loss": 3.939,
+      "step": 1266688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.925208485792698e-05,
+      "loss": 3.9441,
+      "step": 1267200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.924369891041646e-05,
+      "loss": 3.9384,
+      "step": 1267712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.923531296290594e-05,
+      "loss": 3.9331,
+      "step": 1268224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9226927015395424e-05,
+      "loss": 3.9391,
+      "step": 1268736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9218557446688633e-05,
+      "loss": 3.9432,
+      "step": 1269248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9210171499178113e-05,
+      "loss": 3.9548,
+      "step": 1269760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9201785551667593e-05,
+      "loss": 3.9372,
+      "step": 1270272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9193399604157073e-05,
+      "loss": 3.9327,
+      "step": 1270784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9185013656646553e-05,
+      "loss": 3.9367,
+      "step": 1271296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9176644087939765e-05,
+      "loss": 3.9439,
+      "step": 1271808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9168258140429245e-05,
+      "loss": 3.9482,
+      "step": 1272320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9159872192918725e-05,
+      "loss": 3.9347,
+      "step": 1272832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9151486245408205e-05,
+      "loss": 3.9409,
+      "step": 1273344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9143100297897685e-05,
+      "loss": 3.9242,
+      "step": 1273856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9134714350387165e-05,
+      "loss": 3.9424,
+      "step": 1274368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9126328402876645e-05,
+      "loss": 3.9325,
+      "step": 1274880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.911794245536612e-05,
+      "loss": 3.9303,
+      "step": 1275392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9109589265463066e-05,
+      "loss": 3.941,
+      "step": 1275904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9101219696756276e-05,
+      "loss": 3.939,
+      "step": 1276416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9092833749245755e-05,
+      "loss": 3.9352,
+      "step": 1276928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.908444780173524e-05,
+      "loss": 3.938,
+      "step": 1277440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.907606185422472e-05,
+      "loss": 3.9318,
+      "step": 1277952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.90676759067142e-05,
+      "loss": 3.9417,
+      "step": 1278464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.905928995920368e-05,
+      "loss": 3.9292,
+      "step": 1278976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.905090401169316e-05,
+      "loss": 3.9379,
+      "step": 1279488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.904251806418264e-05,
+      "loss": 3.9418,
+      "step": 1280000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.903413211667212e-05,
+      "loss": 3.9442,
+      "step": 1280512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.902577892676906e-05,
+      "loss": 3.9399,
+      "step": 1281024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.901739297925854e-05,
+      "loss": 3.939,
+      "step": 1281536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.900900703174802e-05,
+      "loss": 3.9448,
+      "step": 1282048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.90006210842375e-05,
+      "loss": 3.9415,
+      "step": 1282560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.899223513672698e-05,
+      "loss": 3.9288,
+      "step": 1283072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.898384918921646e-05,
+      "loss": 3.9498,
+      "step": 1283584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.897546324170594e-05,
+      "loss": 3.9384,
+      "step": 1284096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8967077294195423e-05,
+      "loss": 3.9457,
+      "step": 1284608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8958691346684896e-05,
+      "loss": 3.9324,
+      "step": 1285120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8950305399174376e-05,
+      "loss": 3.9299,
+      "step": 1285632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8941919451663856e-05,
+      "loss": 3.9269,
+      "step": 1286144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.893354988295707e-05,
+      "loss": 3.9339,
+      "step": 1286656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.892516393544655e-05,
+      "loss": 3.9296,
+      "step": 1287168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.891677798793603e-05,
+      "loss": 3.9454,
+      "step": 1287680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.890839204042551e-05,
+      "loss": 3.9457,
+      "step": 1288192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.890000609291499e-05,
+      "loss": 3.9346,
+      "step": 1288704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.889162014540447e-05,
+      "loss": 3.9369,
+      "step": 1289216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.888323419789395e-05,
+      "loss": 3.9428,
+      "step": 1289728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.887484825038343e-05,
+      "loss": 3.9453,
+      "step": 1290240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.886647868167664e-05,
+      "loss": 3.9453,
+      "step": 1290752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.885809273416612e-05,
+      "loss": 3.9461,
+      "step": 1291264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.88497067866556e-05,
+      "loss": 3.9387,
+      "step": 1291776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.884132083914508e-05,
+      "loss": 3.9392,
+      "step": 1292288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.883293489163456e-05,
+      "loss": 3.9354,
+      "step": 1292800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.882456532292777e-05,
+      "loss": 3.9359,
+      "step": 1293312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8816195754220982e-05,
+      "loss": 3.9471,
+      "step": 1293824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8807809806710462e-05,
+      "loss": 3.9365,
+      "step": 1294336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8799423859199942e-05,
+      "loss": 3.9276,
+      "step": 1294848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8791037911689422e-05,
+      "loss": 3.954,
+      "step": 1295360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.878266834298263e-05,
+      "loss": 3.9331,
+      "step": 1295872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8774282395472115e-05,
+      "loss": 3.9354,
+      "step": 1296384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8765896447961595e-05,
+      "loss": 3.9341,
+      "step": 1296896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8757510500451075e-05,
+      "loss": 3.9331,
+      "step": 1297408
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.010714530944824,
+      "eval_runtime": 295.7463,
+      "eval_samples_per_second": 1290.265,
+      "eval_steps_per_second": 40.322,
+      "step": 1297440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8749124552940555e-05,
+      "loss": 3.9237,
+      "step": 1297920
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8740738605430035e-05,
+      "loss": 3.939,
+      "step": 1298432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8732352657919514e-05,
+      "loss": 3.9514,
+      "step": 1298944
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8723966710408994e-05,
+      "loss": 3.9348,
+      "step": 1299456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8715597141702204e-05,
+      "loss": 3.9522,
+      "step": 1299968
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8707211194191687e-05,
+      "loss": 3.9262,
+      "step": 1300480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8698825246681167e-05,
+      "loss": 3.9398,
+      "step": 1300992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8690439299170647e-05,
+      "loss": 3.929,
+      "step": 1301504
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8682069730463856e-05,
+      "loss": 3.9424,
+      "step": 1302016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8673700161757065e-05,
+      "loss": 3.9371,
+      "step": 1302528
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.866531421424655e-05,
+      "loss": 3.9431,
+      "step": 1303040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8656928266736028e-05,
+      "loss": 3.944,
+      "step": 1303552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8648558698029237e-05,
+      "loss": 3.9302,
+      "step": 1304064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8640172750518717e-05,
+      "loss": 3.9262,
+      "step": 1304576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8631786803008197e-05,
+      "loss": 3.9264,
+      "step": 1305088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8623400855497677e-05,
+      "loss": 3.9336,
+      "step": 1305600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8615014907987157e-05,
+      "loss": 3.9425,
+      "step": 1306112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.860662896047664e-05,
+      "loss": 3.9329,
+      "step": 1306624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.859824301296612e-05,
+      "loss": 3.9393,
+      "step": 1307136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.85898570654556e-05,
+      "loss": 3.9623,
+      "step": 1307648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.858148749674881e-05,
+      "loss": 3.9376,
+      "step": 1308160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.857310154923829e-05,
+      "loss": 3.934,
+      "step": 1308672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.856471560172777e-05,
+      "loss": 3.938,
+      "step": 1309184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.855632965421725e-05,
+      "loss": 3.9472,
+      "step": 1309696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8547943706706726e-05,
+      "loss": 3.9292,
+      "step": 1310208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8539557759196206e-05,
+      "loss": 3.9341,
+      "step": 1310720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8531171811685686e-05,
+      "loss": 3.9371,
+      "step": 1311232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8522802242978902e-05,
+      "loss": 3.9355,
+      "step": 1311744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.851441629546838e-05,
+      "loss": 3.9208,
+      "step": 1312256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8506030347957858e-05,
+      "loss": 3.9305,
+      "step": 1312768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8497660779251074e-05,
+      "loss": 3.9304,
+      "step": 1313280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8489274831740554e-05,
+      "loss": 3.9412,
+      "step": 1313792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8480888884230027e-05,
+      "loss": 3.9503,
+      "step": 1314304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8472502936719507e-05,
+      "loss": 3.9318,
+      "step": 1314816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8464149746816456e-05,
+      "loss": 3.9313,
+      "step": 1315328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8455763799305936e-05,
+      "loss": 3.9424,
+      "step": 1315840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8447377851795416e-05,
+      "loss": 3.9229,
+      "step": 1316352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8438991904284896e-05,
+      "loss": 3.9317,
+      "step": 1316864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8430605956774375e-05,
+      "loss": 3.9252,
+      "step": 1317376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.842222000926385e-05,
+      "loss": 3.9304,
+      "step": 1317888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8413834061753332e-05,
+      "loss": 3.9263,
+      "step": 1318400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8405448114242812e-05,
+      "loss": 3.9465,
+      "step": 1318912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8397062166732292e-05,
+      "loss": 3.9309,
+      "step": 1319424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8388676219221772e-05,
+      "loss": 3.9334,
+      "step": 1319936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8380290271711252e-05,
+      "loss": 3.9332,
+      "step": 1320448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8371904324200732e-05,
+      "loss": 3.9369,
+      "step": 1320960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8363518376690212e-05,
+      "loss": 3.9332,
+      "step": 1321472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.835513242917969e-05,
+      "loss": 3.9245,
+      "step": 1321984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8346762860472904e-05,
+      "loss": 3.9186,
+      "step": 1322496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8338376912962384e-05,
+      "loss": 3.9424,
+      "step": 1323008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8329990965451864e-05,
+      "loss": 3.9304,
+      "step": 1323520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8321621396745073e-05,
+      "loss": 3.9234,
+      "step": 1324032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8313235449234553e-05,
+      "loss": 3.9251,
+      "step": 1324544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8304849501724033e-05,
+      "loss": 3.9325,
+      "step": 1325056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8296463554213516e-05,
+      "loss": 3.9215,
+      "step": 1325568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8288077606702996e-05,
+      "loss": 3.9381,
+      "step": 1326080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8279691659192476e-05,
+      "loss": 3.9254,
+      "step": 1326592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8271305711681956e-05,
+      "loss": 3.9287,
+      "step": 1327104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8262919764171436e-05,
+      "loss": 3.9406,
+      "step": 1327616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8254533816660916e-05,
+      "loss": 3.9196,
+      "step": 1328128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8246147869150393e-05,
+      "loss": 3.9192,
+      "step": 1328640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8237761921639873e-05,
+      "loss": 3.9326,
+      "step": 1329152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8229375974129353e-05,
+      "loss": 3.9148,
+      "step": 1329664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8221006405422562e-05,
+      "loss": 3.9233,
+      "step": 1330176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8212620457912042e-05,
+      "loss": 3.9354,
+      "step": 1330688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8204234510401522e-05,
+      "loss": 3.941,
+      "step": 1331200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8195848562891e-05,
+      "loss": 3.9272,
+      "step": 1331712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8187478994184214e-05,
+      "loss": 3.9222,
+      "step": 1332224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8179093046673694e-05,
+      "loss": 3.9163,
+      "step": 1332736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8170707099163174e-05,
+      "loss": 3.9313,
+      "step": 1333248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8162321151652654e-05,
+      "loss": 3.9327,
+      "step": 1333760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8153951582945863e-05,
+      "loss": 3.9351,
+      "step": 1334272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.814558201423908e-05,
+      "loss": 3.9332,
+      "step": 1334784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.813719606672856e-05,
+      "loss": 3.939,
+      "step": 1335296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8128810119218036e-05,
+      "loss": 3.9401,
+      "step": 1335808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8120424171707515e-05,
+      "loss": 3.9188,
+      "step": 1336320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8112038224196995e-05,
+      "loss": 3.9329,
+      "step": 1336832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8103652276686475e-05,
+      "loss": 3.935,
+      "step": 1337344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8095266329175955e-05,
+      "loss": 3.92,
+      "step": 1337856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8086880381665435e-05,
+      "loss": 3.9439,
+      "step": 1338368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8078510812958648e-05,
+      "loss": 3.9379,
+      "step": 1338880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8070124865448128e-05,
+      "loss": 3.9431,
+      "step": 1339392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8061738917937608e-05,
+      "loss": 3.9426,
+      "step": 1339904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8053352970427088e-05,
+      "loss": 3.9185,
+      "step": 1340416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8044967022916568e-05,
+      "loss": 3.9313,
+      "step": 1340928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.803659745420978e-05,
+      "loss": 3.9308,
+      "step": 1341440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.802821150669926e-05,
+      "loss": 3.9362,
+      "step": 1341952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.801982555918874e-05,
+      "loss": 3.9339,
+      "step": 1342464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.801145599048195e-05,
+      "loss": 3.928,
+      "step": 1342976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.800307004297143e-05,
+      "loss": 3.9379,
+      "step": 1343488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.799468409546091e-05,
+      "loss": 3.9339,
+      "step": 1344000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.798629814795039e-05,
+      "loss": 3.919,
+      "step": 1344512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.79779285792436e-05,
+      "loss": 3.9293,
+      "step": 1345024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.796954263173308e-05,
+      "loss": 3.9385,
+      "step": 1345536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.796115668422256e-05,
+      "loss": 3.9425,
+      "step": 1346048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.795277073671204e-05,
+      "loss": 3.9288,
+      "step": 1346560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.794438478920152e-05,
+      "loss": 3.9172,
+      "step": 1347072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7935998841691e-05,
+      "loss": 3.9287,
+      "step": 1347584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.792761289418048e-05,
+      "loss": 3.937,
+      "step": 1348096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7919226946669965e-05,
+      "loss": 3.9367,
+      "step": 1348608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7910840999159444e-05,
+      "loss": 3.9285,
+      "step": 1349120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7902471430452654e-05,
+      "loss": 3.9276,
+      "step": 1349632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7894085482942134e-05,
+      "loss": 3.9178,
+      "step": 1350144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7885699535431613e-05,
+      "loss": 3.9306,
+      "step": 1350656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7877313587921093e-05,
+      "loss": 3.9213,
+      "step": 1351168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7868944019214306e-05,
+      "loss": 3.9254,
+      "step": 1351680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7860558071703786e-05,
+      "loss": 3.9294,
+      "step": 1352192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7852172124193266e-05,
+      "loss": 3.9289,
+      "step": 1352704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7843786176682746e-05,
+      "loss": 3.9264,
+      "step": 1353216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.783540022917222e-05,
+      "loss": 3.9304,
+      "step": 1353728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7827014281661702e-05,
+      "loss": 3.9233,
+      "step": 1354240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7818644712954918e-05,
+      "loss": 3.9277,
+      "step": 1354752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7810258765444398e-05,
+      "loss": 3.9214,
+      "step": 1355264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.780187281793387e-05,
+      "loss": 3.924,
+      "step": 1355776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.779348687042335e-05,
+      "loss": 3.9367,
+      "step": 1356288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7785117301716567e-05,
+      "loss": 3.9356,
+      "step": 1356800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7776731354206047e-05,
+      "loss": 3.9313,
+      "step": 1357312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.776836178549926e-05,
+      "loss": 3.9268,
+      "step": 1357824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.775997583798874e-05,
+      "loss": 3.933,
+      "step": 1358336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.775158989047822e-05,
+      "loss": 3.9323,
+      "step": 1358848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7743203942967693e-05,
+      "loss": 3.921,
+      "step": 1359360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7734817995457173e-05,
+      "loss": 3.9414,
+      "step": 1359872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7726432047946656e-05,
+      "loss": 3.9286,
+      "step": 1360384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7718046100436136e-05,
+      "loss": 3.9374,
+      "step": 1360896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7709660152925616e-05,
+      "loss": 3.9211,
+      "step": 1361408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7701274205415096e-05,
+      "loss": 3.9211,
+      "step": 1361920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7692904636708305e-05,
+      "loss": 3.9164,
+      "step": 1362432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7684518689197785e-05,
+      "loss": 3.9235,
+      "step": 1362944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7676132741687265e-05,
+      "loss": 3.9196,
+      "step": 1363456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7667746794176748e-05,
+      "loss": 3.935,
+      "step": 1363968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7659377225469957e-05,
+      "loss": 3.9396,
+      "step": 1364480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7650991277959437e-05,
+      "loss": 3.9307,
+      "step": 1364992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7642605330448917e-05,
+      "loss": 3.9212,
+      "step": 1365504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7634219382938397e-05,
+      "loss": 3.9328,
+      "step": 1366016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.762584981423161e-05,
+      "loss": 3.9354,
+      "step": 1366528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.761746386672109e-05,
+      "loss": 3.9377,
+      "step": 1367040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.760907791921057e-05,
+      "loss": 3.9346,
+      "step": 1367552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.760070835050378e-05,
+      "loss": 3.9316,
+      "step": 1368064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.759232240299326e-05,
+      "loss": 3.9231,
+      "step": 1368576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.758393645548274e-05,
+      "loss": 3.9314,
+      "step": 1369088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.757555050797222e-05,
+      "loss": 3.928,
+      "step": 1369600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.75671645604617e-05,
+      "loss": 3.9329,
+      "step": 1370112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7558778612951182e-05,
+      "loss": 3.9305,
+      "step": 1370624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7550392665440662e-05,
+      "loss": 3.9235,
+      "step": 1371136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.754202309673387e-05,
+      "loss": 3.9416,
+      "step": 1371648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.753363714922335e-05,
+      "loss": 3.9272,
+      "step": 1372160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.752525120171283e-05,
+      "loss": 3.9235,
+      "step": 1372672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7516881633006043e-05,
+      "loss": 3.9254,
+      "step": 1373184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.7508495685495523e-05,
+      "loss": 3.9224,
+      "step": 1373696
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 4.0063323974609375,
+      "eval_runtime": 295.7745,
+      "eval_samples_per_second": 1290.142,
+      "eval_steps_per_second": 40.318,
+      "step": 1373760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7500109737985003e-05,
+      "loss": 3.9139,
+      "step": 1374208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7491723790474483e-05,
+      "loss": 3.9302,
+      "step": 1374720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7483337842963963e-05,
+      "loss": 3.9416,
+      "step": 1375232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7474951895453443e-05,
+      "loss": 3.9266,
+      "step": 1375744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7466565947942923e-05,
+      "loss": 3.9402,
+      "step": 1376256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7458180000432403e-05,
+      "loss": 3.9212,
+      "step": 1376768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.744979405292188e-05,
+      "loss": 3.9322,
+      "step": 1377280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.744140810541136e-05,
+      "loss": 3.9158,
+      "step": 1377792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.743302215790084e-05,
+      "loss": 3.9344,
+      "step": 1378304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.742463621039032e-05,
+      "loss": 3.9286,
+      "step": 1378816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.74162502628798e-05,
+      "loss": 3.9321,
+      "step": 1379328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7407880694173012e-05,
+      "loss": 3.9389,
+      "step": 1379840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7399494746662492e-05,
+      "loss": 3.9207,
+      "step": 1380352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7391108799151972e-05,
+      "loss": 3.9185,
+      "step": 1380864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.738272285164145e-05,
+      "loss": 3.9159,
+      "step": 1381376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.737433690413093e-05,
+      "loss": 3.9282,
+      "step": 1381888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.736595095662041e-05,
+      "loss": 3.9318,
+      "step": 1382400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.735756500910989e-05,
+      "loss": 3.926,
+      "step": 1382912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.734917906159937e-05,
+      "loss": 3.9294,
+      "step": 1383424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7340809492892584e-05,
+      "loss": 3.9522,
+      "step": 1383936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7332423545382064e-05,
+      "loss": 3.9285,
+      "step": 1384448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7324037597871544e-05,
+      "loss": 3.924,
+      "step": 1384960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7315651650361024e-05,
+      "loss": 3.9346,
+      "step": 1385472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7307265702850504e-05,
+      "loss": 3.9389,
+      "step": 1385984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7298879755339984e-05,
+      "loss": 3.9229,
+      "step": 1386496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7290493807829464e-05,
+      "loss": 3.925,
+      "step": 1387008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7282107860318944e-05,
+      "loss": 3.9274,
+      "step": 1387520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.727372191280842e-05,
+      "loss": 3.9257,
+      "step": 1388032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.72653359652979e-05,
+      "loss": 3.9137,
+      "step": 1388544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.725695001778738e-05,
+      "loss": 3.924,
+      "step": 1389056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7248580449080596e-05,
+      "loss": 3.9241,
+      "step": 1389568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7240194501570073e-05,
+      "loss": 3.9287,
+      "step": 1390080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7231808554059553e-05,
+      "loss": 3.9439,
+      "step": 1390592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7223422606549033e-05,
+      "loss": 3.9212,
+      "step": 1391104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.721505303784224e-05,
+      "loss": 3.9237,
+      "step": 1391616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.720666709033172e-05,
+      "loss": 3.933,
+      "step": 1392128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.71982811428212e-05,
+      "loss": 3.9199,
+      "step": 1392640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.718989519531068e-05,
+      "loss": 3.9189,
+      "step": 1393152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7181509247800165e-05,
+      "loss": 3.9174,
+      "step": 1393664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7173123300289645e-05,
+      "loss": 3.9221,
+      "step": 1394176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7164737352779125e-05,
+      "loss": 3.914,
+      "step": 1394688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7156351405268605e-05,
+      "loss": 3.9403,
+      "step": 1395200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7147965457758085e-05,
+      "loss": 3.9183,
+      "step": 1395712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7139612267855026e-05,
+      "loss": 3.9287,
+      "step": 1396224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7131226320344506e-05,
+      "loss": 3.925,
+      "step": 1396736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7122840372833986e-05,
+      "loss": 3.923,
+      "step": 1397248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7114454425323466e-05,
+      "loss": 3.9277,
+      "step": 1397760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7106068477812946e-05,
+      "loss": 3.912,
+      "step": 1398272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7097682530302426e-05,
+      "loss": 3.9097,
+      "step": 1398784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7089296582791906e-05,
+      "loss": 3.9348,
+      "step": 1399296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7080927014085115e-05,
+      "loss": 3.9255,
+      "step": 1399808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.70725410665746e-05,
+      "loss": 3.9157,
+      "step": 1400320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.706415511906408e-05,
+      "loss": 3.9192,
+      "step": 1400832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.705576917155356e-05,
+      "loss": 3.9277,
+      "step": 1401344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.704738322404304e-05,
+      "loss": 3.9086,
+      "step": 1401856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7038997276532518e-05,
+      "loss": 3.9288,
+      "step": 1402368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7030611329021998e-05,
+      "loss": 3.9187,
+      "step": 1402880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7022225381511478e-05,
+      "loss": 3.9201,
+      "step": 1403392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.701385581280469e-05,
+      "loss": 3.9344,
+      "step": 1403904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.700546986529417e-05,
+      "loss": 3.9131,
+      "step": 1404416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.699708391778365e-05,
+      "loss": 3.9096,
+      "step": 1404928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.698871434907686e-05,
+      "loss": 3.9239,
+      "step": 1405440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.698032840156634e-05,
+      "loss": 3.9096,
+      "step": 1405952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.697194245405582e-05,
+      "loss": 3.9136,
+      "step": 1406464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.69635565065453e-05,
+      "loss": 3.9261,
+      "step": 1406976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6955170559034783e-05,
+      "loss": 3.9308,
+      "step": 1407488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6946784611524256e-05,
+      "loss": 3.9161,
+      "step": 1408000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6938398664013736e-05,
+      "loss": 3.9153,
+      "step": 1408512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6930012716503216e-05,
+      "loss": 3.905,
+      "step": 1409024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6921626768992696e-05,
+      "loss": 3.9252,
+      "step": 1409536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.691325720028591e-05,
+      "loss": 3.922,
+      "step": 1410048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.690487125277539e-05,
+      "loss": 3.9297,
+      "step": 1410560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6896501684068604e-05,
+      "loss": 3.9219,
+      "step": 1411072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6888115736558077e-05,
+      "loss": 3.9298,
+      "step": 1411584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6879729789047557e-05,
+      "loss": 3.9359,
+      "step": 1412096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6871343841537037e-05,
+      "loss": 3.9093,
+      "step": 1412608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.686295789402652e-05,
+      "loss": 3.9232,
+      "step": 1413120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.685458832531973e-05,
+      "loss": 3.9287,
+      "step": 1413632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.684620237780921e-05,
+      "loss": 3.9118,
+      "step": 1414144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.683781643029869e-05,
+      "loss": 3.9352,
+      "step": 1414656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.682943048278817e-05,
+      "loss": 3.9284,
+      "step": 1415168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.682104453527765e-05,
+      "loss": 3.9335,
+      "step": 1415680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.681265858776713e-05,
+      "loss": 3.9339,
+      "step": 1416192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6804289019060342e-05,
+      "loss": 3.909,
+      "step": 1416704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6795903071549822e-05,
+      "loss": 3.926,
+      "step": 1417216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6787517124039302e-05,
+      "loss": 3.9203,
+      "step": 1417728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6779131176528782e-05,
+      "loss": 3.93,
+      "step": 1418240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6770745229018262e-05,
+      "loss": 3.9236,
+      "step": 1418752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6762375660311474e-05,
+      "loss": 3.9224,
+      "step": 1419264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6753989712800954e-05,
+      "loss": 3.9265,
+      "step": 1419776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6745603765290434e-05,
+      "loss": 3.9253,
+      "step": 1420288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6737217817779914e-05,
+      "loss": 3.911,
+      "step": 1420800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6728831870269394e-05,
+      "loss": 3.9192,
+      "step": 1421312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6720462301562603e-05,
+      "loss": 3.9282,
+      "step": 1421824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6712076354052083e-05,
+      "loss": 3.932,
+      "step": 1422336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6703690406541567e-05,
+      "loss": 3.9245,
+      "step": 1422848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6695304459031047e-05,
+      "loss": 3.909,
+      "step": 1423360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6686918511520526e-05,
+      "loss": 3.9204,
+      "step": 1423872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6678532564010006e-05,
+      "loss": 3.927,
+      "step": 1424384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6670146616499486e-05,
+      "loss": 3.9245,
+      "step": 1424896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6661777047792695e-05,
+      "loss": 3.9195,
+      "step": 1425408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6653391100282175e-05,
+      "loss": 3.9208,
+      "step": 1425920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6645021531575388e-05,
+      "loss": 3.9083,
+      "step": 1426432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6636635584064868e-05,
+      "loss": 3.9271,
+      "step": 1426944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6628266015358077e-05,
+      "loss": 3.906,
+      "step": 1427456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6619880067847557e-05,
+      "loss": 3.9185,
+      "step": 1427968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6611494120337037e-05,
+      "loss": 3.9148,
+      "step": 1428480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.660310817282652e-05,
+      "loss": 3.9271,
+      "step": 1428992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6594722225316e-05,
+      "loss": 3.9133,
+      "step": 1429504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.658633627780548e-05,
+      "loss": 3.9249,
+      "step": 1430016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.657795033029496e-05,
+      "loss": 3.9149,
+      "step": 1430528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.656956438278444e-05,
+      "loss": 3.9156,
+      "step": 1431040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6561178435273913e-05,
+      "loss": 3.9185,
+      "step": 1431552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6552792487763397e-05,
+      "loss": 3.9136,
+      "step": 1432064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6544406540252877e-05,
+      "loss": 3.9288,
+      "step": 1432576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6536020592742356e-05,
+      "loss": 3.9265,
+      "step": 1433088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6527651024035566e-05,
+      "loss": 3.922,
+      "step": 1433600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6519265076525046e-05,
+      "loss": 3.9166,
+      "step": 1434112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6510879129014525e-05,
+      "loss": 3.924,
+      "step": 1434624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6502493181504005e-05,
+      "loss": 3.9277,
+      "step": 1435136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6494107233993485e-05,
+      "loss": 3.9084,
+      "step": 1435648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6485737665286698e-05,
+      "loss": 3.932,
+      "step": 1436160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6477351717776178e-05,
+      "loss": 3.923,
+      "step": 1436672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6468965770265658e-05,
+      "loss": 3.9256,
+      "step": 1437184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6460579822755138e-05,
+      "loss": 3.915,
+      "step": 1437696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6452193875244618e-05,
+      "loss": 3.9179,
+      "step": 1438208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6443807927734098e-05,
+      "loss": 3.9063,
+      "step": 1438720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.643543835902731e-05,
+      "loss": 3.9159,
+      "step": 1439232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.642705241151679e-05,
+      "loss": 3.9119,
+      "step": 1439744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.641866646400627e-05,
+      "loss": 3.9263,
+      "step": 1440256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.641028051649575e-05,
+      "loss": 3.9318,
+      "step": 1440768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.640189456898523e-05,
+      "loss": 3.9235,
+      "step": 1441280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.639350862147471e-05,
+      "loss": 3.9131,
+      "step": 1441792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.638512267396419e-05,
+      "loss": 3.9235,
+      "step": 1442304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.637673672645367e-05,
+      "loss": 3.9262,
+      "step": 1442816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6368367157746882e-05,
+      "loss": 3.9291,
+      "step": 1443328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.635999758904009e-05,
+      "loss": 3.9258,
+      "step": 1443840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.635161164152957e-05,
+      "loss": 3.9173,
+      "step": 1444352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.634322569401905e-05,
+      "loss": 3.9219,
+      "step": 1444864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.633483974650853e-05,
+      "loss": 3.9235,
+      "step": 1445376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6326453798998015e-05,
+      "loss": 3.9158,
+      "step": 1445888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6318067851487495e-05,
+      "loss": 3.9257,
+      "step": 1446400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6309681903976975e-05,
+      "loss": 3.9211,
+      "step": 1446912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6301295956466448e-05,
+      "loss": 3.9147,
+      "step": 1447424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6292910008955928e-05,
+      "loss": 3.9314,
+      "step": 1447936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6284540440249144e-05,
+      "loss": 3.9213,
+      "step": 1448448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6276154492738624e-05,
+      "loss": 3.9144,
+      "step": 1448960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.62677685452281e-05,
+      "loss": 3.918,
+      "step": 1449472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.625938259771758e-05,
+      "loss": 3.9117,
+      "step": 1449984
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 4.002874374389648,
+      "eval_runtime": 294.4494,
+      "eval_samples_per_second": 1295.948,
+      "eval_steps_per_second": 40.499,
+      "step": 1450080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6251029407814525e-05,
+      "loss": 3.9105,
+      "step": 1450496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6242643460304005e-05,
+      "loss": 3.918,
+      "step": 1451008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6234257512793485e-05,
+      "loss": 3.9331,
+      "step": 1451520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6225871565282968e-05,
+      "loss": 3.9163,
+      "step": 1452032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6217485617772448e-05,
+      "loss": 3.937,
+      "step": 1452544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.620909967026192e-05,
+      "loss": 3.9081,
+      "step": 1453056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.62007137227514e-05,
+      "loss": 3.924,
+      "step": 1453568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.619232777524088e-05,
+      "loss": 3.9062,
+      "step": 1454080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6183958206534097e-05,
+      "loss": 3.9252,
+      "step": 1454592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6175572259023574e-05,
+      "loss": 3.9227,
+      "step": 1455104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6167186311513054e-05,
+      "loss": 3.9197,
+      "step": 1455616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.615881674280627e-05,
+      "loss": 3.9281,
+      "step": 1456128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.615044717409948e-05,
+      "loss": 3.9111,
+      "step": 1456640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.614206122658896e-05,
+      "loss": 3.9105,
+      "step": 1457152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.613367527907844e-05,
+      "loss": 3.9101,
+      "step": 1457664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6125289331567922e-05,
+      "loss": 3.9163,
+      "step": 1458176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6116903384057395e-05,
+      "loss": 3.9209,
+      "step": 1458688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6108517436546875e-05,
+      "loss": 3.9187,
+      "step": 1459200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6100131489036355e-05,
+      "loss": 3.9213,
+      "step": 1459712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6091745541525835e-05,
+      "loss": 3.9411,
+      "step": 1460224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6083359594015315e-05,
+      "loss": 3.9227,
+      "step": 1460736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6074973646504798e-05,
+      "loss": 3.9195,
+      "step": 1461248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6066587698994278e-05,
+      "loss": 3.9252,
+      "step": 1461760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6058201751483758e-05,
+      "loss": 3.9278,
+      "step": 1462272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6049832182776967e-05,
+      "loss": 3.912,
+      "step": 1462784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6041446235266447e-05,
+      "loss": 3.9196,
+      "step": 1463296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.603307666655966e-05,
+      "loss": 3.9199,
+      "step": 1463808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.602469071904914e-05,
+      "loss": 3.9133,
+      "step": 1464320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.601630477153862e-05,
+      "loss": 3.9099,
+      "step": 1464832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.60079188240281e-05,
+      "loss": 3.9106,
+      "step": 1465344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.599953287651758e-05,
+      "loss": 3.9184,
+      "step": 1465856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.599114692900706e-05,
+      "loss": 3.9191,
+      "step": 1466368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.598276098149654e-05,
+      "loss": 3.9321,
+      "step": 1466880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.597437503398602e-05,
+      "loss": 3.9194,
+      "step": 1467392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.596602184408296e-05,
+      "loss": 3.9111,
+      "step": 1467904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.595763589657244e-05,
+      "loss": 3.9226,
+      "step": 1468416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.594924994906192e-05,
+      "loss": 3.9152,
+      "step": 1468928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.59408640015514e-05,
+      "loss": 3.9067,
+      "step": 1469440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.593247805404088e-05,
+      "loss": 3.9107,
+      "step": 1469952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.592409210653036e-05,
+      "loss": 3.9119,
+      "step": 1470464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.591570615901984e-05,
+      "loss": 3.9088,
+      "step": 1470976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5907320211509324e-05,
+      "loss": 3.9282,
+      "step": 1471488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5898934263998804e-05,
+      "loss": 3.912,
+      "step": 1472000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5890564695292013e-05,
+      "loss": 3.9209,
+      "step": 1472512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5882195126585222e-05,
+      "loss": 3.9194,
+      "step": 1473024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5873809179074702e-05,
+      "loss": 3.9121,
+      "step": 1473536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5865423231564186e-05,
+      "loss": 3.9222,
+      "step": 1474048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5857037284053666e-05,
+      "loss": 3.9024,
+      "step": 1474560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5848651336543146e-05,
+      "loss": 3.8928,
+      "step": 1475072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5840265389032625e-05,
+      "loss": 3.9294,
+      "step": 1475584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5831879441522105e-05,
+      "loss": 3.9154,
+      "step": 1476096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.582349349401158e-05,
+      "loss": 3.9114,
+      "step": 1476608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5815123925304794e-05,
+      "loss": 3.9094,
+      "step": 1477120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5806754356598007e-05,
+      "loss": 3.9157,
+      "step": 1477632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5798368409087487e-05,
+      "loss": 3.9015,
+      "step": 1478144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5789982461576967e-05,
+      "loss": 3.9174,
+      "step": 1478656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5781596514066447e-05,
+      "loss": 3.9153,
+      "step": 1479168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5773210566555927e-05,
+      "loss": 3.9113,
+      "step": 1479680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5764824619045407e-05,
+      "loss": 3.9216,
+      "step": 1480192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5756438671534883e-05,
+      "loss": 3.9036,
+      "step": 1480704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5748052724024363e-05,
+      "loss": 3.9006,
+      "step": 1481216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.573968315531758e-05,
+      "loss": 3.9196,
+      "step": 1481728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5731297207807052e-05,
+      "loss": 3.8945,
+      "step": 1482240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5722911260296532e-05,
+      "loss": 3.9009,
+      "step": 1482752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5714525312786016e-05,
+      "loss": 3.9199,
+      "step": 1483264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.570615574407923e-05,
+      "loss": 3.9236,
+      "step": 1483776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5697769796568705e-05,
+      "loss": 3.9053,
+      "step": 1484288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5689383849058185e-05,
+      "loss": 3.9128,
+      "step": 1484800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5680997901547665e-05,
+      "loss": 3.8934,
+      "step": 1485312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.567262833284088e-05,
+      "loss": 3.917,
+      "step": 1485824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5664242385330357e-05,
+      "loss": 3.9133,
+      "step": 1486336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5655856437819837e-05,
+      "loss": 3.9203,
+      "step": 1486848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5647470490309317e-05,
+      "loss": 3.9123,
+      "step": 1487360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5639084542798797e-05,
+      "loss": 3.9209,
+      "step": 1487872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5630698595288277e-05,
+      "loss": 3.9284,
+      "step": 1488384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5622312647777757e-05,
+      "loss": 3.9006,
+      "step": 1488896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5613926700267237e-05,
+      "loss": 3.917,
+      "step": 1489408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.560555713156045e-05,
+      "loss": 3.9226,
+      "step": 1489920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.559717118404993e-05,
+      "loss": 3.8985,
+      "step": 1490432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.558878523653941e-05,
+      "loss": 3.9282,
+      "step": 1490944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.558039928902889e-05,
+      "loss": 3.9201,
+      "step": 1491456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.557201334151837e-05,
+      "loss": 3.9218,
+      "step": 1491968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5563643772811578e-05,
+      "loss": 3.9303,
+      "step": 1492480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.555525782530106e-05,
+      "loss": 3.9044,
+      "step": 1492992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.554687187779054e-05,
+      "loss": 3.9105,
+      "step": 1493504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.553848593028002e-05,
+      "loss": 3.9128,
+      "step": 1494016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.553011636157323e-05,
+      "loss": 3.9213,
+      "step": 1494528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.552173041406271e-05,
+      "loss": 3.9147,
+      "step": 1495040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.5513360845355923e-05,
+      "loss": 3.9152,
+      "step": 1495552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5504974897845403e-05,
+      "loss": 3.914,
+      "step": 1496064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5496588950334883e-05,
+      "loss": 3.9221,
+      "step": 1496576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5488203002824363e-05,
+      "loss": 3.9037,
+      "step": 1497088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5479817055313843e-05,
+      "loss": 3.9095,
+      "step": 1497600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5471431107803323e-05,
+      "loss": 3.9203,
+      "step": 1498112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5463045160292803e-05,
+      "loss": 3.922,
+      "step": 1498624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5454659212782283e-05,
+      "loss": 3.9167,
+      "step": 1499136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5446273265271763e-05,
+      "loss": 3.9007,
+      "step": 1499648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5437903696564975e-05,
+      "loss": 3.908,
+      "step": 1500160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5429517749054455e-05,
+      "loss": 3.915,
+      "step": 1500672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5421131801543935e-05,
+      "loss": 3.9191,
+      "step": 1501184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5412745854033415e-05,
+      "loss": 3.9162,
+      "step": 1501696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5404376285326624e-05,
+      "loss": 3.9106,
+      "step": 1502208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5396006716619837e-05,
+      "loss": 3.9026,
+      "step": 1502720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5387620769109316e-05,
+      "loss": 3.9158,
+      "step": 1503232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5379234821598796e-05,
+      "loss": 3.9001,
+      "step": 1503744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5370848874088276e-05,
+      "loss": 3.9084,
+      "step": 1504256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5362462926577756e-05,
+      "loss": 3.9027,
+      "step": 1504768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.535409335787097e-05,
+      "loss": 3.9179,
+      "step": 1505280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.534570741036045e-05,
+      "loss": 3.9066,
+      "step": 1505792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.533732146284993e-05,
+      "loss": 3.9201,
+      "step": 1506304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.532893551533941e-05,
+      "loss": 3.9018,
+      "step": 1506816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.532054956782889e-05,
+      "loss": 3.9101,
+      "step": 1507328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5312163620318362e-05,
+      "loss": 3.9076,
+      "step": 1507840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5303777672807845e-05,
+      "loss": 3.9091,
+      "step": 1508352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5295391725297325e-05,
+      "loss": 3.9151,
+      "step": 1508864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5287005777786805e-05,
+      "loss": 3.9218,
+      "step": 1509376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.527865258788375e-05,
+      "loss": 3.9144,
+      "step": 1509888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.527026664037323e-05,
+      "loss": 3.9088,
+      "step": 1510400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.526188069286271e-05,
+      "loss": 3.9161,
+      "step": 1510912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.525349474535219e-05,
+      "loss": 3.9173,
+      "step": 1511424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5245108797841667e-05,
+      "loss": 3.9018,
+      "step": 1511936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.523675560793861e-05,
+      "loss": 3.9238,
+      "step": 1512448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.522836966042809e-05,
+      "loss": 3.9093,
+      "step": 1512960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.521998371291757e-05,
+      "loss": 3.9203,
+      "step": 1513472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.521159776540705e-05,
+      "loss": 3.9095,
+      "step": 1513984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.520321181789653e-05,
+      "loss": 3.9092,
+      "step": 1514496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.519482587038601e-05,
+      "loss": 3.894,
+      "step": 1515008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5186439922875488e-05,
+      "loss": 3.9117,
+      "step": 1515520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5178053975364968e-05,
+      "loss": 3.9034,
+      "step": 1516032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5169684406658184e-05,
+      "loss": 3.9151,
+      "step": 1516544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.516129845914766e-05,
+      "loss": 3.9194,
+      "step": 1517056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.515291251163714e-05,
+      "loss": 3.9184,
+      "step": 1517568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.514452656412662e-05,
+      "loss": 3.9028,
+      "step": 1518080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.51361406166161e-05,
+      "loss": 3.9167,
+      "step": 1518592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.512775466910558e-05,
+      "loss": 3.9195,
+      "step": 1519104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.511936872159506e-05,
+      "loss": 3.9191,
+      "step": 1519616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.511098277408454e-05,
+      "loss": 3.9187,
+      "step": 1520128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.510261320537775e-05,
+      "loss": 3.911,
+      "step": 1520640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5094227257867232e-05,
+      "loss": 3.9103,
+      "step": 1521152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5085841310356712e-05,
+      "loss": 3.9175,
+      "step": 1521664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5077455362846192e-05,
+      "loss": 3.906,
+      "step": 1522176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5069069415335672e-05,
+      "loss": 3.9181,
+      "step": 1522688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5060683467825152e-05,
+      "loss": 3.9064,
+      "step": 1523200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.505231389911836e-05,
+      "loss": 3.9119,
+      "step": 1523712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.504392795160784e-05,
+      "loss": 3.9225,
+      "step": 1524224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5035558382901054e-05,
+      "loss": 3.9152,
+      "step": 1524736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5027172435390534e-05,
+      "loss": 3.9049,
+      "step": 1525248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5018786487880014e-05,
+      "loss": 3.9061,
+      "step": 1525760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.5010400540369494e-05,
+      "loss": 3.9068,
+      "step": 1526272
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9988300800323486,
+      "eval_runtime": 291.1966,
+      "eval_samples_per_second": 1310.424,
+      "eval_steps_per_second": 40.952,
+      "step": 1526400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5002014592858974e-05,
+      "loss": 3.8958,
+      "step": 1526784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4993628645348454e-05,
+      "loss": 3.9107,
+      "step": 1527296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4985242697837934e-05,
+      "loss": 3.9206,
+      "step": 1527808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4976856750327414e-05,
+      "loss": 3.9113,
+      "step": 1528320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4968470802816893e-05,
+      "loss": 3.9268,
+      "step": 1528832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4960084855306373e-05,
+      "loss": 3.9077,
+      "step": 1529344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4951698907795853e-05,
+      "loss": 3.9119,
+      "step": 1529856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4943312960285333e-05,
+      "loss": 3.9032,
+      "step": 1530368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4934927012774813e-05,
+      "loss": 3.9113,
+      "step": 1530880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4926541065264293e-05,
+      "loss": 3.9155,
+      "step": 1531392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4918155117753773e-05,
+      "loss": 3.9139,
+      "step": 1531904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4909769170243253e-05,
+      "loss": 3.9208,
+      "step": 1532416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4901399601536466e-05,
+      "loss": 3.9022,
+      "step": 1532928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4893013654025942e-05,
+      "loss": 3.9023,
+      "step": 1533440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4884627706515422e-05,
+      "loss": 3.9032,
+      "step": 1533952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4876241759004902e-05,
+      "loss": 3.9059,
+      "step": 1534464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4867855811494385e-05,
+      "loss": 3.9145,
+      "step": 1534976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4859469863983865e-05,
+      "loss": 3.9093,
+      "step": 1535488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4851083916473345e-05,
+      "loss": 3.9147,
+      "step": 1536000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4842697968962822e-05,
+      "loss": 3.933,
+      "step": 1536512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4834312021452302e-05,
+      "loss": 3.9177,
+      "step": 1537024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4825942452745514e-05,
+      "loss": 3.9121,
+      "step": 1537536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4817556505234994e-05,
+      "loss": 3.9137,
+      "step": 1538048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4809170557724474e-05,
+      "loss": 3.9174,
+      "step": 1538560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4800784610213954e-05,
+      "loss": 3.9049,
+      "step": 1539072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4792398662703434e-05,
+      "loss": 3.9123,
+      "step": 1539584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4784012715192914e-05,
+      "loss": 3.9101,
+      "step": 1540096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4775626767682394e-05,
+      "loss": 3.906,
+      "step": 1540608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4767257198975603e-05,
+      "loss": 3.9021,
+      "step": 1541120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4758871251465087e-05,
+      "loss": 3.9056,
+      "step": 1541632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4750485303954567e-05,
+      "loss": 3.9072,
+      "step": 1542144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4742099356444046e-05,
+      "loss": 3.9108,
+      "step": 1542656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4733713408933526e-05,
+      "loss": 3.9244,
+      "step": 1543168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4725327461423003e-05,
+      "loss": 3.9126,
+      "step": 1543680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4716974271519948e-05,
+      "loss": 3.9033,
+      "step": 1544192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4708588324009428e-05,
+      "loss": 3.9129,
+      "step": 1544704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4700202376498908e-05,
+      "loss": 3.9122,
+      "step": 1545216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4691816428988388e-05,
+      "loss": 3.8958,
+      "step": 1545728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4683446860281597e-05,
+      "loss": 3.901,
+      "step": 1546240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4675060912771077e-05,
+      "loss": 3.9064,
+      "step": 1546752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4666674965260557e-05,
+      "loss": 3.8983,
+      "step": 1547264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.465828901775004e-05,
+      "loss": 3.9229,
+      "step": 1547776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.464990307023952e-05,
+      "loss": 3.9046,
+      "step": 1548288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4641517122729e-05,
+      "loss": 3.9093,
+      "step": 1548800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4633131175218477e-05,
+      "loss": 3.9122,
+      "step": 1549312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4624745227707957e-05,
+      "loss": 3.9106,
+      "step": 1549824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4616359280197437e-05,
+      "loss": 3.9119,
+      "step": 1550336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4607973332686917e-05,
+      "loss": 3.8974,
+      "step": 1550848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4599587385176397e-05,
+      "loss": 3.8817,
+      "step": 1551360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.459120143766588e-05,
+      "loss": 3.9257,
+      "step": 1551872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.458281549015536e-05,
+      "loss": 3.9041,
+      "step": 1552384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4574429542644836e-05,
+      "loss": 3.9061,
+      "step": 1552896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4566043595134316e-05,
+      "loss": 3.9037,
+      "step": 1553408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4557657647623796e-05,
+      "loss": 3.9025,
+      "step": 1553920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.454928807891701e-05,
+      "loss": 3.8969,
+      "step": 1554432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.454091851021022e-05,
+      "loss": 3.9063,
+      "step": 1554944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.45325325626997e-05,
+      "loss": 3.9068,
+      "step": 1555456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.452414661518918e-05,
+      "loss": 3.9034,
+      "step": 1555968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4515760667678658e-05,
+      "loss": 3.9201,
+      "step": 1556480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4507374720168138e-05,
+      "loss": 3.8929,
+      "step": 1556992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4498988772657618e-05,
+      "loss": 3.8922,
+      "step": 1557504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4490602825147098e-05,
+      "loss": 3.9155,
+      "step": 1558016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.448221687763658e-05,
+      "loss": 3.8846,
+      "step": 1558528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.447384730892979e-05,
+      "loss": 3.8906,
+      "step": 1559040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.446546136141927e-05,
+      "loss": 3.9142,
+      "step": 1559552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.445707541390875e-05,
+      "loss": 3.9158,
+      "step": 1560064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.444868946639823e-05,
+      "loss": 3.8983,
+      "step": 1560576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.444030351888771e-05,
+      "loss": 3.9061,
+      "step": 1561088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.443191757137719e-05,
+      "loss": 3.8894,
+      "step": 1561600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4423548002670402e-05,
+      "loss": 3.908,
+      "step": 1562112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4415162055159882e-05,
+      "loss": 3.9029,
+      "step": 1562624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.440679248645309e-05,
+      "loss": 3.9103,
+      "step": 1563136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4398422917746304e-05,
+      "loss": 3.9041,
+      "step": 1563648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390036970235784e-05,
+      "loss": 3.9135,
+      "step": 1564160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4381651022725264e-05,
+      "loss": 3.9212,
+      "step": 1564672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4373265075214744e-05,
+      "loss": 3.8968,
+      "step": 1565184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4364879127704224e-05,
+      "loss": 3.9066,
+      "step": 1565696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4356509558997433e-05,
+      "loss": 3.9137,
+      "step": 1566208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4348123611486913e-05,
+      "loss": 3.8917,
+      "step": 1566720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4339737663976396e-05,
+      "loss": 3.9184,
+      "step": 1567232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4331351716465876e-05,
+      "loss": 3.9182,
+      "step": 1567744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4322965768955356e-05,
+      "loss": 3.9149,
+      "step": 1568256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4314579821444836e-05,
+      "loss": 3.9193,
+      "step": 1568768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4306193873934313e-05,
+      "loss": 3.8978,
+      "step": 1569280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4297807926423792e-05,
+      "loss": 3.9006,
+      "step": 1569792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4289421978913272e-05,
+      "loss": 3.9068,
+      "step": 1570304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4281052410206488e-05,
+      "loss": 3.9163,
+      "step": 1570816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4272666462695965e-05,
+      "loss": 3.9094,
+      "step": 1571328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4264280515185445e-05,
+      "loss": 3.9072,
+      "step": 1571840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4255894567674925e-05,
+      "loss": 3.9043,
+      "step": 1572352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4247508620164405e-05,
+      "loss": 3.9146,
+      "step": 1572864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4239139051457617e-05,
+      "loss": 3.9006,
+      "step": 1573376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4230753103947097e-05,
+      "loss": 3.9,
+      "step": 1573888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4222367156436577e-05,
+      "loss": 3.9094,
+      "step": 1574400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4213981208926057e-05,
+      "loss": 3.9169,
+      "step": 1574912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4205595261415537e-05,
+      "loss": 3.9093,
+      "step": 1575424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4197209313905017e-05,
+      "loss": 3.8961,
+      "step": 1575936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4188839745198226e-05,
+      "loss": 3.9011,
+      "step": 1576448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4180453797687706e-05,
+      "loss": 3.9064,
+      "step": 1576960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.417206785017719e-05,
+      "loss": 3.9148,
+      "step": 1577472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.416368190266667e-05,
+      "loss": 3.9056,
+      "step": 1577984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4155295955156146e-05,
+      "loss": 3.9068,
+      "step": 1578496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.414692638644936e-05,
+      "loss": 3.8967,
+      "step": 1579008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.413854043893884e-05,
+      "loss": 3.905,
+      "step": 1579520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4130154491428318e-05,
+      "loss": 3.8953,
+      "step": 1580032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4121768543917798e-05,
+      "loss": 3.9052,
+      "step": 1580544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4113382596407278e-05,
+      "loss": 3.8893,
+      "step": 1581056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4104996648896758e-05,
+      "loss": 3.919,
+      "step": 1581568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4096610701386238e-05,
+      "loss": 3.9008,
+      "step": 1582080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4088241132679447e-05,
+      "loss": 3.9052,
+      "step": 1582592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4079855185168927e-05,
+      "loss": 3.8998,
+      "step": 1583104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.407146923765841e-05,
+      "loss": 3.902,
+      "step": 1583616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.406308329014789e-05,
+      "loss": 3.901,
+      "step": 1584128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.405469734263737e-05,
+      "loss": 3.9026,
+      "step": 1584640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4046311395126847e-05,
+      "loss": 3.9089,
+      "step": 1585152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.403794182642006e-05,
+      "loss": 3.9133,
+      "step": 1585664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.402955587890954e-05,
+      "loss": 3.9054,
+      "step": 1586176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.402116993139902e-05,
+      "loss": 3.9047,
+      "step": 1586688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.40127839838885e-05,
+      "loss": 3.9065,
+      "step": 1587200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.400439803637798e-05,
+      "loss": 3.9132,
+      "step": 1587712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.399601208886746e-05,
+      "loss": 3.8902,
+      "step": 1588224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3987642520160672e-05,
+      "loss": 3.9157,
+      "step": 1588736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.397925657265015e-05,
+      "loss": 3.9019,
+      "step": 1589248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3970870625139628e-05,
+      "loss": 3.9123,
+      "step": 1589760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.396248467762911e-05,
+      "loss": 3.8994,
+      "step": 1590272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.395409873011859e-05,
+      "loss": 3.9033,
+      "step": 1590784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.39457291614118e-05,
+      "loss": 3.8913,
+      "step": 1591296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.393734321390128e-05,
+      "loss": 3.9008,
+      "step": 1591808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.392895726639076e-05,
+      "loss": 3.9001,
+      "step": 1592320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.392057131888024e-05,
+      "loss": 3.9039,
+      "step": 1592832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.391218537136972e-05,
+      "loss": 3.9102,
+      "step": 1593344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3903799423859204e-05,
+      "loss": 3.914,
+      "step": 1593856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3895429855152413e-05,
+      "loss": 3.8956,
+      "step": 1594368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3887043907641893e-05,
+      "loss": 3.9082,
+      "step": 1594880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3878657960131373e-05,
+      "loss": 3.9151,
+      "step": 1595392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3870272012620853e-05,
+      "loss": 3.9123,
+      "step": 1595904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.386188606511033e-05,
+      "loss": 3.9073,
+      "step": 1596416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3853516496403545e-05,
+      "loss": 3.9096,
+      "step": 1596928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3845130548893025e-05,
+      "loss": 3.9014,
+      "step": 1597440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3836744601382502e-05,
+      "loss": 3.9123,
+      "step": 1597952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3828358653871982e-05,
+      "loss": 3.8974,
+      "step": 1598464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.381997270636146e-05,
+      "loss": 3.9123,
+      "step": 1598976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3811603137654674e-05,
+      "loss": 3.901,
+      "step": 1599488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3803217190144154e-05,
+      "loss": 3.904,
+      "step": 1600000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3794831242633634e-05,
+      "loss": 3.9141,
+      "step": 1600512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3786445295123114e-05,
+      "loss": 3.9085,
+      "step": 1601024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3778059347612594e-05,
+      "loss": 3.9006,
+      "step": 1601536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3769689778905803e-05,
+      "loss": 3.8954,
+      "step": 1602048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3761303831395283e-05,
+      "loss": 3.9024,
+      "step": 1602560
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9965615272521973,
+      "eval_runtime": 288.1511,
+      "eval_samples_per_second": 1324.274,
+      "eval_steps_per_second": 41.385,
+      "step": 1602720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3752917883884766e-05,
+      "loss": 3.8971,
+      "step": 1603072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3744531936374246e-05,
+      "loss": 3.9008,
+      "step": 1603584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3736145988863726e-05,
+      "loss": 3.916,
+      "step": 1604096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3727760041353206e-05,
+      "loss": 3.9054,
+      "step": 1604608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3719374093842683e-05,
+      "loss": 3.9188,
+      "step": 1605120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3710988146332163e-05,
+      "loss": 3.9017,
+      "step": 1605632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3702618577625375e-05,
+      "loss": 3.8996,
+      "step": 1606144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3694249008918588e-05,
+      "loss": 3.897,
+      "step": 1606656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3685863061408068e-05,
+      "loss": 3.901,
+      "step": 1607168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3677477113897548e-05,
+      "loss": 3.9096,
+      "step": 1607680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3669091166387028e-05,
+      "loss": 3.9111,
+      "step": 1608192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3660705218876508e-05,
+      "loss": 3.9098,
+      "step": 1608704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.365233565016972e-05,
+      "loss": 3.901,
+      "step": 1609216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.36439497026592e-05,
+      "loss": 3.8953,
+      "step": 1609728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.363556375514868e-05,
+      "loss": 3.8963,
+      "step": 1610240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3627177807638157e-05,
+      "loss": 3.8968,
+      "step": 1610752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3618791860127636e-05,
+      "loss": 3.9061,
+      "step": 1611264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3610405912617116e-05,
+      "loss": 3.9031,
+      "step": 1611776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3602019965106596e-05,
+      "loss": 3.9086,
+      "step": 1612288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3593634017596076e-05,
+      "loss": 3.9227,
+      "step": 1612800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.358528082769302e-05,
+      "loss": 3.9153,
+      "step": 1613312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.35768948801825e-05,
+      "loss": 3.9002,
+      "step": 1613824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.356850893267198e-05,
+      "loss": 3.9078,
+      "step": 1614336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3560122985161458e-05,
+      "loss": 3.9094,
+      "step": 1614848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3551737037650938e-05,
+      "loss": 3.9004,
+      "step": 1615360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.354335109014042e-05,
+      "loss": 3.9061,
+      "step": 1615872
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.35349651426299e-05,
+      "loss": 3.9017,
+      "step": 1616384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.352657919511938e-05,
+      "loss": 3.9005,
+      "step": 1616896
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.351819324760886e-05,
+      "loss": 3.8984,
+      "step": 1617408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3509807300098338e-05,
+      "loss": 3.894,
+      "step": 1617920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3501421352587818e-05,
+      "loss": 3.9009,
+      "step": 1618432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3493035405077297e-05,
+      "loss": 3.9036,
+      "step": 1618944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.348464945756678e-05,
+      "loss": 3.9138,
+      "step": 1619456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.347627988885999e-05,
+      "loss": 3.905,
+      "step": 1619968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.346789394134947e-05,
+      "loss": 3.8945,
+      "step": 1620480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3459524372642682e-05,
+      "loss": 3.9105,
+      "step": 1620992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.345113842513216e-05,
+      "loss": 3.9068,
+      "step": 1621504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3442768856425375e-05,
+      "loss": 3.8844,
+      "step": 1622016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3434382908914855e-05,
+      "loss": 3.8994,
+      "step": 1622528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3425996961404335e-05,
+      "loss": 3.8968,
+      "step": 1623040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.341761101389381e-05,
+      "loss": 3.8944,
+      "step": 1623552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.340922506638329e-05,
+      "loss": 3.9134,
+      "step": 1624064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.340083911887277e-05,
+      "loss": 3.8981,
+      "step": 1624576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.339245317136225e-05,
+      "loss": 3.9007,
+      "step": 1625088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.338406722385173e-05,
+      "loss": 3.9019,
+      "step": 1625600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3375681276341214e-05,
+      "loss": 3.9019,
+      "step": 1626112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3367311707634424e-05,
+      "loss": 3.9087,
+      "step": 1626624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3358925760123903e-05,
+      "loss": 3.8909,
+      "step": 1627136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3350539812613383e-05,
+      "loss": 3.8718,
+      "step": 1627648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3342153865102863e-05,
+      "loss": 3.9195,
+      "step": 1628160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3333767917592343e-05,
+      "loss": 3.8962,
+      "step": 1628672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3325381970081823e-05,
+      "loss": 3.9054,
+      "step": 1629184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3316996022571303e-05,
+      "loss": 3.8923,
+      "step": 1629696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3308610075060783e-05,
+      "loss": 3.8958,
+      "step": 1630208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3300224127550263e-05,
+      "loss": 3.8916,
+      "step": 1630720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3291854558843472e-05,
+      "loss": 3.8954,
+      "step": 1631232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3283468611332952e-05,
+      "loss": 3.9035,
+      "step": 1631744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3275082663822436e-05,
+      "loss": 3.8931,
+      "step": 1632256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3266696716311916e-05,
+      "loss": 3.9136,
+      "step": 1632768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3258327147605125e-05,
+      "loss": 3.8889,
+      "step": 1633280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3249941200094605e-05,
+      "loss": 3.8833,
+      "step": 1633792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3241571631387814e-05,
+      "loss": 3.9072,
+      "step": 1634304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3233185683877297e-05,
+      "loss": 3.883,
+      "step": 1634816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3224799736366777e-05,
+      "loss": 3.8811,
+      "step": 1635328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3216413788856257e-05,
+      "loss": 3.9066,
+      "step": 1635840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3208027841345737e-05,
+      "loss": 3.9057,
+      "step": 1636352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3199658272638946e-05,
+      "loss": 3.8933,
+      "step": 1636864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3191272325128426e-05,
+      "loss": 3.8991,
+      "step": 1637376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3182886377617906e-05,
+      "loss": 3.8814,
+      "step": 1637888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.317450043010739e-05,
+      "loss": 3.9014,
+      "step": 1638400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.31661308614006e-05,
+      "loss": 3.8945,
+      "step": 1638912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3157744913890078e-05,
+      "loss": 3.9067,
+      "step": 1639424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3149358966379558e-05,
+      "loss": 3.8968,
+      "step": 1639936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3140973018869038e-05,
+      "loss": 3.9043,
+      "step": 1640448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3132587071358518e-05,
+      "loss": 3.9126,
+      "step": 1640960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3124201123847998e-05,
+      "loss": 3.8861,
+      "step": 1641472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3115815176337478e-05,
+      "loss": 3.9021,
+      "step": 1641984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3107429228826958e-05,
+      "loss": 3.9053,
+      "step": 1642496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.309905966012017e-05,
+      "loss": 3.8884,
+      "step": 1643008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3090673712609647e-05,
+      "loss": 3.908,
+      "step": 1643520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3082287765099127e-05,
+      "loss": 3.9095,
+      "step": 1644032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.3073901817588607e-05,
+      "loss": 3.9062,
+      "step": 1644544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.306551587007809e-05,
+      "loss": 3.9196,
+      "step": 1645056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.305712992256757e-05,
+      "loss": 3.8904,
+      "step": 1645568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.304874397505705e-05,
+      "loss": 3.8921,
+      "step": 1646080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.304035802754653e-05,
+      "loss": 3.9001,
+      "step": 1646592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.303198845883974e-05,
+      "loss": 3.9049,
+      "step": 1647104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.302360251132922e-05,
+      "loss": 3.9076,
+      "step": 1647616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.30152165638187e-05,
+      "loss": 3.8999,
+      "step": 1648128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.3006846995111912e-05,
+      "loss": 3.8939,
+      "step": 1648640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.299846104760139e-05,
+      "loss": 3.9091,
+      "step": 1649152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.299007510009087e-05,
+      "loss": 3.8946,
+      "step": 1649664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.298168915258035e-05,
+      "loss": 3.8935,
+      "step": 1650176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2973303205069828e-05,
+      "loss": 3.9021,
+      "step": 1650688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2964933636363044e-05,
+      "loss": 3.9106,
+      "step": 1651200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2956547688852524e-05,
+      "loss": 3.9027,
+      "step": 1651712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2948161741342004e-05,
+      "loss": 3.8862,
+      "step": 1652224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.293977579383148e-05,
+      "loss": 3.8942,
+      "step": 1652736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2931406225124693e-05,
+      "loss": 3.9007,
+      "step": 1653248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2923020277614173e-05,
+      "loss": 3.9035,
+      "step": 1653760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2914634330103653e-05,
+      "loss": 3.9014,
+      "step": 1654272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2906248382593133e-05,
+      "loss": 3.8991,
+      "step": 1654784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2897862435082613e-05,
+      "loss": 3.888,
+      "step": 1655296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2889476487572093e-05,
+      "loss": 3.8944,
+      "step": 1655808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2881090540061573e-05,
+      "loss": 3.8941,
+      "step": 1656320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2872720971354782e-05,
+      "loss": 3.8967,
+      "step": 1656832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2864335023844262e-05,
+      "loss": 3.8815,
+      "step": 1657344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2855949076333745e-05,
+      "loss": 3.9121,
+      "step": 1657856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2847563128823225e-05,
+      "loss": 3.8912,
+      "step": 1658368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2839193560116434e-05,
+      "loss": 3.8985,
+      "step": 1658880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2830807612605914e-05,
+      "loss": 3.8876,
+      "step": 1659392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2822421665095394e-05,
+      "loss": 3.9018,
+      "step": 1659904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2814035717584874e-05,
+      "loss": 3.8924,
+      "step": 1660416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2805649770074354e-05,
+      "loss": 3.8923,
+      "step": 1660928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2797263822563834e-05,
+      "loss": 3.9022,
+      "step": 1661440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2788877875053314e-05,
+      "loss": 3.9072,
+      "step": 1661952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2780491927542794e-05,
+      "loss": 3.8935,
+      "step": 1662464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2772122358836006e-05,
+      "loss": 3.9049,
+      "step": 1662976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2763736411325483e-05,
+      "loss": 3.8927,
+      "step": 1663488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2755350463814963e-05,
+      "loss": 3.9074,
+      "step": 1664000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.274698089510818e-05,
+      "loss": 3.8879,
+      "step": 1664512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2738594947597655e-05,
+      "loss": 3.9098,
+      "step": 1665024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2730209000087135e-05,
+      "loss": 3.8936,
+      "step": 1665536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2721823052576615e-05,
+      "loss": 3.9058,
+      "step": 1666048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2713453483869828e-05,
+      "loss": 3.8957,
+      "step": 1666560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2705067536359308e-05,
+      "loss": 3.8957,
+      "step": 1667072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2696681588848788e-05,
+      "loss": 3.882,
+      "step": 1667584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2688295641338268e-05,
+      "loss": 3.8958,
+      "step": 1668096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2679909693827748e-05,
+      "loss": 3.8934,
+      "step": 1668608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2671523746317227e-05,
+      "loss": 3.9007,
+      "step": 1669120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2663137798806707e-05,
+      "loss": 3.8977,
+      "step": 1669632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2654784608903652e-05,
+      "loss": 3.9079,
+      "step": 1670144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.264639866139313e-05,
+      "loss": 3.8911,
+      "step": 1670656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.263801271388261e-05,
+      "loss": 3.8996,
+      "step": 1671168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.262962676637209e-05,
+      "loss": 3.905,
+      "step": 1671680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.262124081886157e-05,
+      "loss": 3.9059,
+      "step": 1672192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.261285487135105e-05,
+      "loss": 3.8998,
+      "step": 1672704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.260446892384053e-05,
+      "loss": 3.9022,
+      "step": 1673216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.259608297633001e-05,
+      "loss": 3.8931,
+      "step": 1673728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.258769702881949e-05,
+      "loss": 3.9106,
+      "step": 1674240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.25793274601127e-05,
+      "loss": 3.889,
+      "step": 1674752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.257094151260218e-05,
+      "loss": 3.9044,
+      "step": 1675264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.256255556509166e-05,
+      "loss": 3.8935,
+      "step": 1675776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2554169617581138e-05,
+      "loss": 3.898,
+      "step": 1676288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.254578367007062e-05,
+      "loss": 3.9055,
+      "step": 1676800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.2537414101363833e-05,
+      "loss": 3.8979,
+      "step": 1677312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.252902815385331e-05,
+      "loss": 3.8989,
+      "step": 1677824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.252064220634279e-05,
+      "loss": 3.8841,
+      "step": 1678336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.251225625883227e-05,
+      "loss": 3.8961,
+      "step": 1678848
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.993511199951172,
+      "eval_runtime": 290.4524,
+      "eval_samples_per_second": 1313.782,
+      "eval_steps_per_second": 41.057,
+      "step": 1679040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.250387031132175e-05,
+      "loss": 3.8885,
+      "step": 1679360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.249548436381123e-05,
+      "loss": 3.8896,
+      "step": 1679872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.248709841630071e-05,
+      "loss": 3.9071,
+      "step": 1680384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2478712468790193e-05,
+      "loss": 3.8997,
+      "step": 1680896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.247032652127967e-05,
+      "loss": 3.9091,
+      "step": 1681408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.246194057376915e-05,
+      "loss": 3.8946,
+      "step": 1681920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.245355462625863e-05,
+      "loss": 3.8941,
+      "step": 1682432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.244516867874811e-05,
+      "loss": 3.8915,
+      "step": 1682944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.243678273123759e-05,
+      "loss": 3.8942,
+      "step": 1683456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.242839678372707e-05,
+      "loss": 3.9035,
+      "step": 1683968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2420010836216553e-05,
+      "loss": 3.8984,
+      "step": 1684480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.241162488870603e-05,
+      "loss": 3.9017,
+      "step": 1684992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2403255319999242e-05,
+      "loss": 3.8938,
+      "step": 1685504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2394869372488722e-05,
+      "loss": 3.8914,
+      "step": 1686016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23864834249782e-05,
+      "loss": 3.8909,
+      "step": 1686528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.237809747746768e-05,
+      "loss": 3.8873,
+      "step": 1687040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2369711529957162e-05,
+      "loss": 3.8997,
+      "step": 1687552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.236132558244664e-05,
+      "loss": 3.898,
+      "step": 1688064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.235293963493612e-05,
+      "loss": 3.8985,
+      "step": 1688576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.23445536874256e-05,
+      "loss": 3.9134,
+      "step": 1689088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.233616773991508e-05,
+      "loss": 3.9113,
+      "step": 1689600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.232779817120829e-05,
+      "loss": 3.897,
+      "step": 1690112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.231941222369777e-05,
+      "loss": 3.8954,
+      "step": 1690624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2311026276187254e-05,
+      "loss": 3.9041,
+      "step": 1691136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.230264032867673e-05,
+      "loss": 3.8981,
+      "step": 1691648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.229425438116621e-05,
+      "loss": 3.8975,
+      "step": 1692160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.228586843365569e-05,
+      "loss": 3.8978,
+      "step": 1692672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2277498864948903e-05,
+      "loss": 3.8943,
+      "step": 1693184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.226911291743838e-05,
+      "loss": 3.8912,
+      "step": 1693696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2260726969927863e-05,
+      "loss": 3.8907,
+      "step": 1694208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2252341022417343e-05,
+      "loss": 3.8875,
+      "step": 1694720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2243955074906823e-05,
+      "loss": 3.8975,
+      "step": 1695232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2235569127396303e-05,
+      "loss": 3.9064,
+      "step": 1695744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2227183179885783e-05,
+      "loss": 3.9008,
+      "step": 1696256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2218797232375263e-05,
+      "loss": 3.8897,
+      "step": 1696768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.221042766366847e-05,
+      "loss": 3.9002,
+      "step": 1697280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2202041716157955e-05,
+      "loss": 3.9004,
+      "step": 1697792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2193655768647435e-05,
+      "loss": 3.8801,
+      "step": 1698304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.218526982113691e-05,
+      "loss": 3.8938,
+      "step": 1698816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.217688387362639e-05,
+      "loss": 3.8906,
+      "step": 1699328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.216849792611587e-05,
+      "loss": 3.8836,
+      "step": 1699840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.216011197860535e-05,
+      "loss": 3.9043,
+      "step": 1700352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2151742409898564e-05,
+      "loss": 3.8915,
+      "step": 1700864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2143356462388044e-05,
+      "loss": 3.897,
+      "step": 1701376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2134970514877524e-05,
+      "loss": 3.8944,
+      "step": 1701888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2126584567367004e-05,
+      "loss": 3.8949,
+      "step": 1702400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2118214998660213e-05,
+      "loss": 3.9036,
+      "step": 1702912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2109845429953425e-05,
+      "loss": 3.8864,
+      "step": 1703424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.210145948244291e-05,
+      "loss": 3.8614,
+      "step": 1703936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2093073534932385e-05,
+      "loss": 3.9099,
+      "step": 1704448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2084687587421865e-05,
+      "loss": 3.8918,
+      "step": 1704960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2076301639911345e-05,
+      "loss": 3.8999,
+      "step": 1705472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2067915692400825e-05,
+      "loss": 3.887,
+      "step": 1705984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2059529744890305e-05,
+      "loss": 3.8896,
+      "step": 1706496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2051143797379785e-05,
+      "loss": 3.8809,
+      "step": 1707008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2042774228672998e-05,
+      "loss": 3.8896,
+      "step": 1707520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2034388281162477e-05,
+      "loss": 3.899,
+      "step": 1708032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2026002333651957e-05,
+      "loss": 3.8825,
+      "step": 1708544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2017616386141437e-05,
+      "loss": 3.9044,
+      "step": 1709056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2009230438630917e-05,
+      "loss": 3.8872,
+      "step": 1709568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2000844491120394e-05,
+      "loss": 3.8767,
+      "step": 1710080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1992458543609877e-05,
+      "loss": 3.8954,
+      "step": 1710592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1984072596099357e-05,
+      "loss": 3.8794,
+      "step": 1711104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1975703027392566e-05,
+      "loss": 3.8734,
+      "step": 1711616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1967317079882046e-05,
+      "loss": 3.9006,
+      "step": 1712128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1958931132371526e-05,
+      "loss": 3.8979,
+      "step": 1712640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1950545184861006e-05,
+      "loss": 3.8874,
+      "step": 1713152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.194217561615422e-05,
+      "loss": 3.8906,
+      "step": 1713664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.19337896686437e-05,
+      "loss": 3.8778,
+      "step": 1714176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.192540372113318e-05,
+      "loss": 3.8931,
+      "step": 1714688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.191701777362266e-05,
+      "loss": 3.891,
+      "step": 1715200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1908648204915868e-05,
+      "loss": 3.8945,
+      "step": 1715712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.190027863620908e-05,
+      "loss": 3.8918,
+      "step": 1716224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1891892688698563e-05,
+      "loss": 3.8966,
+      "step": 1716736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.188350674118804e-05,
+      "loss": 3.9076,
+      "step": 1717248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.187512079367752e-05,
+      "loss": 3.882,
+      "step": 1717760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1866734846167e-05,
+      "loss": 3.8979,
+      "step": 1718272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.185834889865648e-05,
+      "loss": 3.8944,
+      "step": 1718784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1849979329949692e-05,
+      "loss": 3.8839,
+      "step": 1719296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1841593382439172e-05,
+      "loss": 3.9002,
+      "step": 1719808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1833207434928652e-05,
+      "loss": 3.9068,
+      "step": 1720320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1824821487418132e-05,
+      "loss": 3.8967,
+      "step": 1720832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1816435539907612e-05,
+      "loss": 3.9138,
+      "step": 1721344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1808049592397092e-05,
+      "loss": 3.8839,
+      "step": 1721856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.17996800236903e-05,
+      "loss": 3.887,
+      "step": 1722368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1791294076179785e-05,
+      "loss": 3.8942,
+      "step": 1722880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1782908128669265e-05,
+      "loss": 3.9013,
+      "step": 1723392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1774522181158745e-05,
+      "loss": 3.8929,
+      "step": 1723904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1766152612451954e-05,
+      "loss": 3.8958,
+      "step": 1724416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1757766664941434e-05,
+      "loss": 3.8899,
+      "step": 1724928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1749380717430914e-05,
+      "loss": 3.8989,
+      "step": 1725440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1740994769920393e-05,
+      "loss": 3.8865,
+      "step": 1725952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1732608822409873e-05,
+      "loss": 3.8867,
+      "step": 1726464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1724222874899353e-05,
+      "loss": 3.8961,
+      "step": 1726976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1715853306192566e-05,
+      "loss": 3.9061,
+      "step": 1727488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1707467358682046e-05,
+      "loss": 3.8945,
+      "step": 1728000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1699081411171522e-05,
+      "loss": 3.8786,
+      "step": 1728512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1690695463661002e-05,
+      "loss": 3.891,
+      "step": 1729024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1682309516150486e-05,
+      "loss": 3.8902,
+      "step": 1729536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1673939947443695e-05,
+      "loss": 3.899,
+      "step": 1730048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1665553999933175e-05,
+      "loss": 3.893,
+      "step": 1730560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1657168052422655e-05,
+      "loss": 3.8901,
+      "step": 1731072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1648798483715867e-05,
+      "loss": 3.8847,
+      "step": 1731584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1640412536205347e-05,
+      "loss": 3.8824,
+      "step": 1732096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1632026588694827e-05,
+      "loss": 3.8923,
+      "step": 1732608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1623640641184307e-05,
+      "loss": 3.8864,
+      "step": 1733120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1615254693673787e-05,
+      "loss": 3.8757,
+      "step": 1733632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1606868746163267e-05,
+      "loss": 3.9056,
+      "step": 1734144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1598482798652747e-05,
+      "loss": 3.8855,
+      "step": 1734656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1590096851142223e-05,
+      "loss": 3.8894,
+      "step": 1735168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1581710903631703e-05,
+      "loss": 3.883,
+      "step": 1735680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1573324956121187e-05,
+      "loss": 3.8939,
+      "step": 1736192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1564939008610667e-05,
+      "loss": 3.8908,
+      "step": 1736704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1556553061100147e-05,
+      "loss": 3.8853,
+      "step": 1737216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1548183492393356e-05,
+      "loss": 3.8961,
+      "step": 1737728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1539797544882836e-05,
+      "loss": 3.897,
+      "step": 1738240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1531427976176048e-05,
+      "loss": 3.8897,
+      "step": 1738752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1523042028665528e-05,
+      "loss": 3.8994,
+      "step": 1739264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.151467245995874e-05,
+      "loss": 3.8858,
+      "step": 1739776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.150628651244822e-05,
+      "loss": 3.8979,
+      "step": 1740288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1497900564937697e-05,
+      "loss": 3.8856,
+      "step": 1740800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1489514617427177e-05,
+      "loss": 3.8998,
+      "step": 1741312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1481128669916657e-05,
+      "loss": 3.8861,
+      "step": 1741824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.147274272240614e-05,
+      "loss": 3.8976,
+      "step": 1742336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.146435677489562e-05,
+      "loss": 3.8952,
+      "step": 1742848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.14559708273851e-05,
+      "loss": 3.8893,
+      "step": 1743360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.144758487987458e-05,
+      "loss": 3.8778,
+      "step": 1743872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.143921531116779e-05,
+      "loss": 3.8887,
+      "step": 1744384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.143082936365727e-05,
+      "loss": 3.8846,
+      "step": 1744896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.142244341614675e-05,
+      "loss": 3.8981,
+      "step": 1745408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1414057468636233e-05,
+      "loss": 3.8865,
+      "step": 1745920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1405687899929442e-05,
+      "loss": 3.9022,
+      "step": 1746432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1397301952418922e-05,
+      "loss": 3.882,
+      "step": 1746944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.13889160049084e-05,
+      "loss": 3.8947,
+      "step": 1747456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1380530057397878e-05,
+      "loss": 3.8993,
+      "step": 1747968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1372160488691094e-05,
+      "loss": 3.9011,
+      "step": 1748480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1363774541180574e-05,
+      "loss": 3.8957,
+      "step": 1748992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1355388593670054e-05,
+      "loss": 3.8951,
+      "step": 1749504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.134700264615953e-05,
+      "loss": 3.8869,
+      "step": 1750016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.133861669864901e-05,
+      "loss": 3.9024,
+      "step": 1750528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.133023075113849e-05,
+      "loss": 3.8828,
+      "step": 1751040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.132184480362797e-05,
+      "loss": 3.8982,
+      "step": 1751552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.131345885611745e-05,
+      "loss": 3.8877,
+      "step": 1752064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1305072908606934e-05,
+      "loss": 3.8978,
+      "step": 1752576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1296686961096414e-05,
+      "loss": 3.8936,
+      "step": 1753088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.128830101358589e-05,
+      "loss": 3.8908,
+      "step": 1753600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1279931444879103e-05,
+      "loss": 3.8935,
+      "step": 1754112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1271545497368583e-05,
+      "loss": 3.8805,
+      "step": 1754624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1263159549858063e-05,
+      "loss": 3.8922,
+      "step": 1755136
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.991323232650757,
+      "eval_runtime": 294.5522,
+      "eval_samples_per_second": 1295.495,
+      "eval_steps_per_second": 40.485,
+      "step": 1755360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1254773602347543e-05,
+      "loss": 3.8852,
+      "step": 1755648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1246387654837023e-05,
+      "loss": 3.8818,
+      "step": 1756160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1238001707326503e-05,
+      "loss": 3.8973,
+      "step": 1756672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1229615759815983e-05,
+      "loss": 3.893,
+      "step": 1757184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1221229812305462e-05,
+      "loss": 3.9028,
+      "step": 1757696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1212843864794942e-05,
+      "loss": 3.8907,
+      "step": 1758208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.120445791728442e-05,
+      "loss": 3.8877,
+      "step": 1758720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1196071969773902e-05,
+      "loss": 3.8859,
+      "step": 1759232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1187686022263382e-05,
+      "loss": 3.8872,
+      "step": 1759744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1179300074752862e-05,
+      "loss": 3.8984,
+      "step": 1760256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1170914127242342e-05,
+      "loss": 3.8899,
+      "step": 1760768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1162528179731822e-05,
+      "loss": 3.8955,
+      "step": 1761280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.115415861102503e-05,
+      "loss": 3.892,
+      "step": 1761792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.114577266351451e-05,
+      "loss": 3.8833,
+      "step": 1762304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1137386716003995e-05,
+      "loss": 3.8802,
+      "step": 1762816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1129000768493474e-05,
+      "loss": 3.88,
+      "step": 1763328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.112061482098295e-05,
+      "loss": 3.8952,
+      "step": 1763840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.111222887347243e-05,
+      "loss": 3.8914,
+      "step": 1764352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.110384292596191e-05,
+      "loss": 3.8874,
+      "step": 1764864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.109545697845139e-05,
+      "loss": 3.9073,
+      "step": 1765376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.108707103094087e-05,
+      "loss": 3.9049,
+      "step": 1765888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1078701462234083e-05,
+      "loss": 3.893,
+      "step": 1766400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1070315514723563e-05,
+      "loss": 3.894,
+      "step": 1766912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1061929567213043e-05,
+      "loss": 3.8876,
+      "step": 1767424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1053543619702523e-05,
+      "loss": 3.8929,
+      "step": 1767936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1045157672192003e-05,
+      "loss": 3.8926,
+      "step": 1768448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.103677172468148e-05,
+      "loss": 3.8914,
+      "step": 1768960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.102838577717096e-05,
+      "loss": 3.8919,
+      "step": 1769472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1020016208464176e-05,
+      "loss": 3.8831,
+      "step": 1769984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1011630260953656e-05,
+      "loss": 3.8844,
+      "step": 1770496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1003244313443132e-05,
+      "loss": 3.8763,
+      "step": 1771008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0994874744736345e-05,
+      "loss": 3.8963,
+      "step": 1771520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0986488797225825e-05,
+      "loss": 3.8996,
+      "step": 1772032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0978102849715305e-05,
+      "loss": 3.8962,
+      "step": 1772544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0969716902204784e-05,
+      "loss": 3.8874,
+      "step": 1773056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0961347333497997e-05,
+      "loss": 3.8855,
+      "step": 1773568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0952961385987477e-05,
+      "loss": 3.8961,
+      "step": 1774080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0944575438476953e-05,
+      "loss": 3.8789,
+      "step": 1774592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0936189490966433e-05,
+      "loss": 3.8822,
+      "step": 1775104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0927803543455913e-05,
+      "loss": 3.8861,
+      "step": 1775616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0919417595945397e-05,
+      "loss": 3.8822,
+      "step": 1776128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0911031648434877e-05,
+      "loss": 3.8927,
+      "step": 1776640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0902645700924357e-05,
+      "loss": 3.8865,
+      "step": 1777152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0894259753413837e-05,
+      "loss": 3.8906,
+      "step": 1777664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0885873805903313e-05,
+      "loss": 3.8852,
+      "step": 1778176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0877487858392793e-05,
+      "loss": 3.8902,
+      "step": 1778688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0869101910882273e-05,
+      "loss": 3.8965,
+      "step": 1779200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0860715963371753e-05,
+      "loss": 3.882,
+      "step": 1779712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0852330015861236e-05,
+      "loss": 3.8549,
+      "step": 1780224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0843944068350716e-05,
+      "loss": 3.9046,
+      "step": 1780736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0835558120840196e-05,
+      "loss": 3.885,
+      "step": 1781248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0827204930937135e-05,
+      "loss": 3.8936,
+      "step": 1781760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0818818983426618e-05,
+      "loss": 3.8803,
+      "step": 1782272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0810433035916098e-05,
+      "loss": 3.8841,
+      "step": 1782784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0802047088405578e-05,
+      "loss": 3.8724,
+      "step": 1783296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0793661140895058e-05,
+      "loss": 3.885,
+      "step": 1783808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0785275193384538e-05,
+      "loss": 3.896,
+      "step": 1784320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0776889245874018e-05,
+      "loss": 3.8741,
+      "step": 1784832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0768503298363494e-05,
+      "loss": 3.9029,
+      "step": 1785344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0760133729656707e-05,
+      "loss": 3.8822,
+      "step": 1785856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.075174778214619e-05,
+      "loss": 3.8701,
+      "step": 1786368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.074336183463567e-05,
+      "loss": 3.8854,
+      "step": 1786880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0734975887125147e-05,
+      "loss": 3.8765,
+      "step": 1787392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0726589939614627e-05,
+      "loss": 3.8688,
+      "step": 1787904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0718203992104106e-05,
+      "loss": 3.8946,
+      "step": 1788416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.070983442339732e-05,
+      "loss": 3.8913,
+      "step": 1788928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.070146485469053e-05,
+      "loss": 3.8801,
+      "step": 1789440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.069307890718001e-05,
+      "loss": 3.8871,
+      "step": 1789952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.068469295966949e-05,
+      "loss": 3.8677,
+      "step": 1790464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0676307012158968e-05,
+      "loss": 3.8884,
+      "step": 1790976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0667921064648448e-05,
+      "loss": 3.8824,
+      "step": 1791488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0659535117137928e-05,
+      "loss": 3.8886,
+      "step": 1792000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0651165548431144e-05,
+      "loss": 3.8918,
+      "step": 1792512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.064277960092062e-05,
+      "loss": 3.8873,
+      "step": 1793024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0634410032213833e-05,
+      "loss": 3.9048,
+      "step": 1793536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0626024084703313e-05,
+      "loss": 3.8752,
+      "step": 1794048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.061763813719279e-05,
+      "loss": 3.8907,
+      "step": 1794560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0609252189682273e-05,
+      "loss": 3.8867,
+      "step": 1795072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0600866242171753e-05,
+      "loss": 3.8783,
+      "step": 1795584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0592480294661233e-05,
+      "loss": 3.8928,
+      "step": 1796096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0584094347150712e-05,
+      "loss": 3.8997,
+      "step": 1796608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0575708399640192e-05,
+      "loss": 3.8913,
+      "step": 1797120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0567322452129672e-05,
+      "loss": 3.9089,
+      "step": 1797632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.055893650461915e-05,
+      "loss": 3.8785,
+      "step": 1798144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.055055055710863e-05,
+      "loss": 3.8804,
+      "step": 1798656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0542164609598112e-05,
+      "loss": 3.8868,
+      "step": 1799168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.053379504089132e-05,
+      "loss": 3.8979,
+      "step": 1799680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.05254090933808e-05,
+      "loss": 3.8832,
+      "step": 1800192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0517039524674014e-05,
+      "loss": 3.8913,
+      "step": 1800704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0508653577163494e-05,
+      "loss": 3.8847,
+      "step": 1801216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0500267629652974e-05,
+      "loss": 3.8914,
+      "step": 1801728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0491881682142454e-05,
+      "loss": 3.8855,
+      "step": 1802240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0483495734631934e-05,
+      "loss": 3.8777,
+      "step": 1802752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0475109787121414e-05,
+      "loss": 3.89,
+      "step": 1803264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0466723839610894e-05,
+      "loss": 3.8989,
+      "step": 1803776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0458337892100373e-05,
+      "loss": 3.8899,
+      "step": 1804288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0449968323393583e-05,
+      "loss": 3.8711,
+      "step": 1804800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0441582375883066e-05,
+      "loss": 3.885,
+      "step": 1805312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0433196428372546e-05,
+      "loss": 3.883,
+      "step": 1805824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0424810480862026e-05,
+      "loss": 3.8967,
+      "step": 1806336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0416424533351502e-05,
+      "loss": 3.8918,
+      "step": 1806848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0408054964644715e-05,
+      "loss": 3.882,
+      "step": 1807360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0399669017134195e-05,
+      "loss": 3.8821,
+      "step": 1807872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0391283069623675e-05,
+      "loss": 3.8771,
+      "step": 1808384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0382897122113155e-05,
+      "loss": 3.8858,
+      "step": 1808896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0374511174602635e-05,
+      "loss": 3.881,
+      "step": 1809408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0366125227092115e-05,
+      "loss": 3.8701,
+      "step": 1809920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0357755658385327e-05,
+      "loss": 3.8963,
+      "step": 1810432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0349369710874804e-05,
+      "loss": 3.8846,
+      "step": 1810944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0340983763364284e-05,
+      "loss": 3.8794,
+      "step": 1811456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0332597815853767e-05,
+      "loss": 3.88,
+      "step": 1811968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0324211868343247e-05,
+      "loss": 3.8829,
+      "step": 1812480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0315842299636456e-05,
+      "loss": 3.8859,
+      "step": 1812992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0307456352125936e-05,
+      "loss": 3.8776,
+      "step": 1813504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0299070404615416e-05,
+      "loss": 3.8911,
+      "step": 1814016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0290684457104896e-05,
+      "loss": 3.8926,
+      "step": 1814528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0282298509594376e-05,
+      "loss": 3.8828,
+      "step": 1815040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.027392894088759e-05,
+      "loss": 3.8904,
+      "step": 1815552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.026554299337707e-05,
+      "loss": 3.8812,
+      "step": 1816064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0257157045866548e-05,
+      "loss": 3.8946,
+      "step": 1816576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0248771098356028e-05,
+      "loss": 3.8821,
+      "step": 1817088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0240385150845508e-05,
+      "loss": 3.8875,
+      "step": 1817600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0231999203334985e-05,
+      "loss": 3.8833,
+      "step": 1818112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0223613255824468e-05,
+      "loss": 3.8923,
+      "step": 1818624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0215227308313948e-05,
+      "loss": 3.8931,
+      "step": 1819136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0206857739607157e-05,
+      "loss": 3.8807,
+      "step": 1819648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0198471792096637e-05,
+      "loss": 3.8662,
+      "step": 1820160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.019010222338985e-05,
+      "loss": 3.8864,
+      "step": 1820672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.018171627587933e-05,
+      "loss": 3.8757,
+      "step": 1821184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.017333032836881e-05,
+      "loss": 3.8888,
+      "step": 1821696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.016494438085829e-05,
+      "loss": 3.8835,
+      "step": 1822208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.015655843334777e-05,
+      "loss": 3.8945,
+      "step": 1822720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.014817248583725e-05,
+      "loss": 3.8739,
+      "step": 1823232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.013978653832673e-05,
+      "loss": 3.8946,
+      "step": 1823744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.013140059081621e-05,
+      "loss": 3.8896,
+      "step": 1824256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.012301464330569e-05,
+      "loss": 3.8974,
+      "step": 1824768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0114645074598902e-05,
+      "loss": 3.8897,
+      "step": 1825280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0106259127088382e-05,
+      "loss": 3.8864,
+      "step": 1825792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.009787317957786e-05,
+      "loss": 3.8843,
+      "step": 1826304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0089487232067338e-05,
+      "loss": 3.8975,
+      "step": 1826816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.008111766336055e-05,
+      "loss": 3.8743,
+      "step": 1827328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.007273171585003e-05,
+      "loss": 3.8927,
+      "step": 1827840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0064345768339514e-05,
+      "loss": 3.8866,
+      "step": 1828352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0055976199632723e-05,
+      "loss": 3.8897,
+      "step": 1828864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0047590252122203e-05,
+      "loss": 3.8849,
+      "step": 1829376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0039204304611683e-05,
+      "loss": 3.8884,
+      "step": 1829888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0030818357101163e-05,
+      "loss": 3.8863,
+      "step": 1830400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0022432409590643e-05,
+      "loss": 3.8778,
+      "step": 1830912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0014062840883855e-05,
+      "loss": 3.8832,
+      "step": 1831424
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9887776374816895,
+      "eval_runtime": 318.7565,
+      "eval_samples_per_second": 1197.124,
+      "eval_steps_per_second": 37.411,
+      "step": 1831680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005676893373335e-05,
+      "loss": 3.8776,
+      "step": 1831936
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9997290945862812e-05,
+      "loss": 3.8784,
+      "step": 1832448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9988904998352292e-05,
+      "loss": 3.8912,
+      "step": 1832960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9980535429645504e-05,
+      "loss": 3.8868,
+      "step": 1833472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9972149482134984e-05,
+      "loss": 3.8994,
+      "step": 1833984
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9963763534624464e-05,
+      "loss": 3.8819,
+      "step": 1834496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9955377587113944e-05,
+      "loss": 3.886,
+      "step": 1835008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9946991639603424e-05,
+      "loss": 3.8819,
+      "step": 1835520
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9938605692092904e-05,
+      "loss": 3.8788,
+      "step": 1836032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9930219744582384e-05,
+      "loss": 3.894,
+      "step": 1836544
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9921833797071864e-05,
+      "loss": 3.885,
+      "step": 1837056
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9913464228365077e-05,
+      "loss": 3.8891,
+      "step": 1837568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9905094659658286e-05,
+      "loss": 3.8927,
+      "step": 1838080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9896708712147766e-05,
+      "loss": 3.8711,
+      "step": 1838592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9888322764637246e-05,
+      "loss": 3.8727,
+      "step": 1839104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9879936817126726e-05,
+      "loss": 3.8741,
+      "step": 1839616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9871550869616205e-05,
+      "loss": 3.8912,
+      "step": 1840128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9863164922105685e-05,
+      "loss": 3.8845,
+      "step": 1840640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.985477897459517e-05,
+      "loss": 3.8834,
+      "step": 1841152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9846409405888378e-05,
+      "loss": 3.9002,
+      "step": 1841664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9838023458377858e-05,
+      "loss": 3.898,
+      "step": 1842176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9829637510867338e-05,
+      "loss": 3.8844,
+      "step": 1842688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9821251563356814e-05,
+      "loss": 3.8897,
+      "step": 1843200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9812865615846298e-05,
+      "loss": 3.8853,
+      "step": 1843712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.980449604713951e-05,
+      "loss": 3.8888,
+      "step": 1844224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.979611009962899e-05,
+      "loss": 3.8828,
+      "step": 1844736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9787724152118467e-05,
+      "loss": 3.8798,
+      "step": 1845248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9779338204607947e-05,
+      "loss": 3.8889,
+      "step": 1845760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9770952257097427e-05,
+      "loss": 3.8767,
+      "step": 1846272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9762566309586907e-05,
+      "loss": 3.8771,
+      "step": 1846784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.975419674088012e-05,
+      "loss": 3.8771,
+      "step": 1847296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.97458107933696e-05,
+      "loss": 3.882,
+      "step": 1847808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.973742484585908e-05,
+      "loss": 3.8883,
+      "step": 1848320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.972903889834856e-05,
+      "loss": 3.8957,
+      "step": 1848832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.972065295083804e-05,
+      "loss": 3.8801,
+      "step": 1849344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.971228338213125e-05,
+      "loss": 3.8832,
+      "step": 1849856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.970389743462073e-05,
+      "loss": 3.8919,
+      "step": 1850368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.969551148711021e-05,
+      "loss": 3.869,
+      "step": 1850880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.968712553959969e-05,
+      "loss": 3.8749,
+      "step": 1851392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.967873959208917e-05,
+      "loss": 3.8814,
+      "step": 1851904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9670353644578648e-05,
+      "loss": 3.8729,
+      "step": 1852416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9661967697068128e-05,
+      "loss": 3.8837,
+      "step": 1852928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9653581749557608e-05,
+      "loss": 3.8826,
+      "step": 1853440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9645212180850824e-05,
+      "loss": 3.8881,
+      "step": 1853952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9636842612144033e-05,
+      "loss": 3.8768,
+      "step": 1854464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9628456664633513e-05,
+      "loss": 3.8863,
+      "step": 1854976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9620070717122993e-05,
+      "loss": 3.8887,
+      "step": 1855488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.961168476961247e-05,
+      "loss": 3.8769,
+      "step": 1856000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9603298822101952e-05,
+      "loss": 3.8519,
+      "step": 1856512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9594929253395165e-05,
+      "loss": 3.8939,
+      "step": 1857024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9586543305884645e-05,
+      "loss": 3.8787,
+      "step": 1857536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.957815735837412e-05,
+      "loss": 3.8872,
+      "step": 1858048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.95697714108636e-05,
+      "loss": 3.8776,
+      "step": 1858560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.956138546335308e-05,
+      "loss": 3.8752,
+      "step": 1859072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.955299951584256e-05,
+      "loss": 3.8723,
+      "step": 1859584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9544629947135774e-05,
+      "loss": 3.8761,
+      "step": 1860096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9536243999625254e-05,
+      "loss": 3.8911,
+      "step": 1860608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9527858052114734e-05,
+      "loss": 3.8647,
+      "step": 1861120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9519472104604214e-05,
+      "loss": 3.8986,
+      "step": 1861632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9511086157093694e-05,
+      "loss": 3.8772,
+      "step": 1862144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9502716588386906e-05,
+      "loss": 3.8604,
+      "step": 1862656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9494330640876386e-05,
+      "loss": 3.8803,
+      "step": 1863168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9485944693365866e-05,
+      "loss": 3.8695,
+      "step": 1863680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9477558745855346e-05,
+      "loss": 3.8657,
+      "step": 1864192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9469172798344826e-05,
+      "loss": 3.8831,
+      "step": 1864704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9460803229638035e-05,
+      "loss": 3.8871,
+      "step": 1865216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9452417282127515e-05,
+      "loss": 3.8712,
+      "step": 1865728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9444031334617e-05,
+      "loss": 3.8864,
+      "step": 1866240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9435645387106478e-05,
+      "loss": 3.8616,
+      "step": 1866752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9427259439595955e-05,
+      "loss": 3.8804,
+      "step": 1867264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9418889870889167e-05,
+      "loss": 3.8802,
+      "step": 1867776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9410503923378647e-05,
+      "loss": 3.8801,
+      "step": 1868288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9402117975868124e-05,
+      "loss": 3.8859,
+      "step": 1868800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9393732028357607e-05,
+      "loss": 3.8804,
+      "step": 1869312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9385346080847087e-05,
+      "loss": 3.9004,
+      "step": 1869824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.93769765121403e-05,
+      "loss": 3.8709,
+      "step": 1870336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9368590564629776e-05,
+      "loss": 3.8832,
+      "step": 1870848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9360204617119256e-05,
+      "loss": 3.8772,
+      "step": 1871360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9351818669608736e-05,
+      "loss": 3.8741,
+      "step": 1871872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9343432722098216e-05,
+      "loss": 3.8892,
+      "step": 1872384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.933506315339143e-05,
+      "loss": 3.8869,
+      "step": 1872896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.932667720588091e-05,
+      "loss": 3.8904,
+      "step": 1873408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.931829125837039e-05,
+      "loss": 3.9003,
+      "step": 1873920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.930990531085987e-05,
+      "loss": 3.8752,
+      "step": 1874432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.930151936334935e-05,
+      "loss": 3.873,
+      "step": 1874944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.929313341583883e-05,
+      "loss": 3.8778,
+      "step": 1875456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9284747468328308e-05,
+      "loss": 3.8938,
+      "step": 1875968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9276361520817788e-05,
+      "loss": 3.8767,
+      "step": 1876480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.926800833091473e-05,
+      "loss": 3.8838,
+      "step": 1876992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.925962238340421e-05,
+      "loss": 3.8841,
+      "step": 1877504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.925123643589369e-05,
+      "loss": 3.8813,
+      "step": 1878016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.924285048838317e-05,
+      "loss": 3.8819,
+      "step": 1878528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9234464540872653e-05,
+      "loss": 3.8711,
+      "step": 1879040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.922607859336213e-05,
+      "loss": 3.8845,
+      "step": 1879552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.921769264585161e-05,
+      "loss": 3.8925,
+      "step": 1880064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.920930669834109e-05,
+      "loss": 3.8851,
+      "step": 1880576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9200937129634302e-05,
+      "loss": 3.8698,
+      "step": 1881088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.919255118212378e-05,
+      "loss": 3.8768,
+      "step": 1881600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9184197992220724e-05,
+      "loss": 3.877,
+      "step": 1882112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9175812044710204e-05,
+      "loss": 3.8862,
+      "step": 1882624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9167426097199684e-05,
+      "loss": 3.8868,
+      "step": 1883136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9159040149689164e-05,
+      "loss": 3.8724,
+      "step": 1883648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9150654202178643e-05,
+      "loss": 3.879,
+      "step": 1884160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9142268254668123e-05,
+      "loss": 3.8683,
+      "step": 1884672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9133882307157603e-05,
+      "loss": 3.8802,
+      "step": 1885184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9125496359647083e-05,
+      "loss": 3.8752,
+      "step": 1885696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9117110412136563e-05,
+      "loss": 3.8696,
+      "step": 1886208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9108724464626043e-05,
+      "loss": 3.8878,
+      "step": 1886720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9100338517115523e-05,
+      "loss": 3.8758,
+      "step": 1887232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9091952569605003e-05,
+      "loss": 3.8765,
+      "step": 1887744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9083583000898216e-05,
+      "loss": 3.8681,
+      "step": 1888256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9075213432191428e-05,
+      "loss": 3.8761,
+      "step": 1888768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9066827484680905e-05,
+      "loss": 3.8819,
+      "step": 1889280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9058441537170385e-05,
+      "loss": 3.8696,
+      "step": 1889792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9050055589659865e-05,
+      "loss": 3.8866,
+      "step": 1890304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9041669642149345e-05,
+      "loss": 3.8855,
+      "step": 1890816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9033283694638825e-05,
+      "loss": 3.8782,
+      "step": 1891328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.902493050473577e-05,
+      "loss": 3.8857,
+      "step": 1891840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.901654455722525e-05,
+      "loss": 3.8739,
+      "step": 1892352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9008158609714726e-05,
+      "loss": 3.8879,
+      "step": 1892864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8999772662204206e-05,
+      "loss": 3.881,
+      "step": 1893376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8991386714693686e-05,
+      "loss": 3.8819,
+      "step": 1893888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.898300076718317e-05,
+      "loss": 3.8776,
+      "step": 1894400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.897461481967265e-05,
+      "loss": 3.8839,
+      "step": 1894912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.896622887216213e-05,
+      "loss": 3.8889,
+      "step": 1895424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.895784292465161e-05,
+      "loss": 3.8711,
+      "step": 1895936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8949456977141086e-05,
+      "loss": 3.8645,
+      "step": 1896448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8941071029630566e-05,
+      "loss": 3.8749,
+      "step": 1896960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8932685082120046e-05,
+      "loss": 3.8737,
+      "step": 1897472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.892433189221699e-05,
+      "loss": 3.8842,
+      "step": 1897984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.891594594470647e-05,
+      "loss": 3.8752,
+      "step": 1898496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.890755999719595e-05,
+      "loss": 3.889,
+      "step": 1899008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.889917404968543e-05,
+      "loss": 3.8694,
+      "step": 1899520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8890788102174907e-05,
+      "loss": 3.8924,
+      "step": 1900032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8882402154664387e-05,
+      "loss": 3.8811,
+      "step": 1900544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.887401620715387e-05,
+      "loss": 3.8887,
+      "step": 1901056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.886563025964335e-05,
+      "loss": 3.8847,
+      "step": 1901568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.885726069093656e-05,
+      "loss": 3.8787,
+      "step": 1902080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884887474342604e-05,
+      "loss": 3.8773,
+      "step": 1902592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.884048879591552e-05,
+      "loss": 3.8889,
+      "step": 1903104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8832102848405e-05,
+      "loss": 3.8703,
+      "step": 1903616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8823733279698212e-05,
+      "loss": 3.8859,
+      "step": 1904128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8815347332187692e-05,
+      "loss": 3.8796,
+      "step": 1904640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8806961384677172e-05,
+      "loss": 3.8815,
+      "step": 1905152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.879859181597038e-05,
+      "loss": 3.8786,
+      "step": 1905664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.879020586845986e-05,
+      "loss": 3.8831,
+      "step": 1906176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.878181992094934e-05,
+      "loss": 3.878,
+      "step": 1906688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8773433973438824e-05,
+      "loss": 3.879,
+      "step": 1907200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8765048025928304e-05,
+      "loss": 3.876,
+      "step": 1907712
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9861364364624023,
+      "eval_runtime": 308.0895,
+      "eval_samples_per_second": 1238.572,
+      "eval_steps_per_second": 38.706,
+      "step": 1908000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8756662078417784e-05,
+      "loss": 3.8695,
+      "step": 1908224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.874827613090726e-05,
+      "loss": 3.8707,
+      "step": 1908736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.873989018339674e-05,
+      "loss": 3.8833,
+      "step": 1909248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.873150423588622e-05,
+      "loss": 3.8824,
+      "step": 1909760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.87231182883757e-05,
+      "loss": 3.8952,
+      "step": 1910272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.871473234086518e-05,
+      "loss": 3.8739,
+      "step": 1910784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8706346393354664e-05,
+      "loss": 3.88,
+      "step": 1911296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8697960445844144e-05,
+      "loss": 3.8772,
+      "step": 1911808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.868957449833362e-05,
+      "loss": 3.8726,
+      "step": 1912320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.86811885508231e-05,
+      "loss": 3.8898,
+      "step": 1912832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.867280260331258e-05,
+      "loss": 3.8765,
+      "step": 1913344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.866441665580206e-05,
+      "loss": 3.8838,
+      "step": 1913856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8656047087095273e-05,
+      "loss": 3.8903,
+      "step": 1914368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8647661139584753e-05,
+      "loss": 3.8668,
+      "step": 1914880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8639275192074232e-05,
+      "loss": 3.8664,
+      "step": 1915392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8630889244563712e-05,
+      "loss": 3.869,
+      "step": 1915904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8622503297053192e-05,
+      "loss": 3.8866,
+      "step": 1916416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8614117349542672e-05,
+      "loss": 3.8852,
+      "step": 1916928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.860573140203215e-05,
+      "loss": 3.8718,
+      "step": 1917440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8597345454521632e-05,
+      "loss": 3.8941,
+      "step": 1917952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8588959507011112e-05,
+      "loss": 3.8927,
+      "step": 1918464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8580573559500592e-05,
+      "loss": 3.8799,
+      "step": 1918976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8572187611990072e-05,
+      "loss": 3.8852,
+      "step": 1919488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8563801664479552e-05,
+      "loss": 3.8815,
+      "step": 1920000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.855543209577276e-05,
+      "loss": 3.8827,
+      "step": 1920512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.854704614826224e-05,
+      "loss": 3.8786,
+      "step": 1921024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8538660200751724e-05,
+      "loss": 3.8735,
+      "step": 1921536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8530274253241204e-05,
+      "loss": 3.8836,
+      "step": 1922048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8521904684534414e-05,
+      "loss": 3.8736,
+      "step": 1922560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8513535115827623e-05,
+      "loss": 3.8703,
+      "step": 1923072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8505149168317103e-05,
+      "loss": 3.8722,
+      "step": 1923584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8496763220806586e-05,
+      "loss": 3.878,
+      "step": 1924096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8488377273296066e-05,
+      "loss": 3.8811,
+      "step": 1924608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8479991325785546e-05,
+      "loss": 3.8911,
+      "step": 1925120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8471605378275026e-05,
+      "loss": 3.8749,
+      "step": 1925632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8463235809568235e-05,
+      "loss": 3.8753,
+      "step": 1926144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8454849862057715e-05,
+      "loss": 3.8893,
+      "step": 1926656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8446463914547195e-05,
+      "loss": 3.8597,
+      "step": 1927168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8438077967036678e-05,
+      "loss": 3.8735,
+      "step": 1927680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8429708398329887e-05,
+      "loss": 3.8756,
+      "step": 1928192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8421322450819367e-05,
+      "loss": 3.8674,
+      "step": 1928704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8412936503308847e-05,
+      "loss": 3.875,
+      "step": 1929216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8404550555798327e-05,
+      "loss": 3.8781,
+      "step": 1929728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8396164608287804e-05,
+      "loss": 3.8828,
+      "step": 1930240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8387778660777287e-05,
+      "loss": 3.8719,
+      "step": 1930752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8379392713266767e-05,
+      "loss": 3.8811,
+      "step": 1931264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8371006765756247e-05,
+      "loss": 3.8793,
+      "step": 1931776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8362620818245727e-05,
+      "loss": 3.8732,
+      "step": 1932288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8354234870735207e-05,
+      "loss": 3.8512,
+      "step": 1932800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8345848923224687e-05,
+      "loss": 3.8836,
+      "step": 1933312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8337462975714163e-05,
+      "loss": 3.8747,
+      "step": 1933824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.832909340700738e-05,
+      "loss": 3.8841,
+      "step": 1934336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.832070745949686e-05,
+      "loss": 3.8662,
+      "step": 1934848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.831232151198634e-05,
+      "loss": 3.8705,
+      "step": 1935360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8303951943279548e-05,
+      "loss": 3.8654,
+      "step": 1935872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8295565995769028e-05,
+      "loss": 3.869,
+      "step": 1936384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8287180048258508e-05,
+      "loss": 3.889,
+      "step": 1936896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8278794100747988e-05,
+      "loss": 3.8612,
+      "step": 1937408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8270408153237468e-05,
+      "loss": 3.8909,
+      "step": 1937920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.826203858453068e-05,
+      "loss": 3.8753,
+      "step": 1938432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.825365263702016e-05,
+      "loss": 3.8565,
+      "step": 1938944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8245266689509637e-05,
+      "loss": 3.8702,
+      "step": 1939456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8236880741999117e-05,
+      "loss": 3.8635,
+      "step": 1939968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8228494794488597e-05,
+      "loss": 3.8645,
+      "step": 1940480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.822010884697808e-05,
+      "loss": 3.875,
+      "step": 1940992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.821172289946756e-05,
+      "loss": 3.8842,
+      "step": 1941504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.820333695195704e-05,
+      "loss": 3.8654,
+      "step": 1942016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.819496738325025e-05,
+      "loss": 3.8809,
+      "step": 1942528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.818658143573973e-05,
+      "loss": 3.857,
+      "step": 1943040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.817819548822921e-05,
+      "loss": 3.8759,
+      "step": 1943552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.816980954071869e-05,
+      "loss": 3.8742,
+      "step": 1944064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.81614399720119e-05,
+      "loss": 3.8788,
+      "step": 1944576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8153086782108843e-05,
+      "loss": 3.8739,
+      "step": 1945088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8144700834598323e-05,
+      "loss": 3.8778,
+      "step": 1945600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8136314887087803e-05,
+      "loss": 3.8932,
+      "step": 1946112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8127928939577283e-05,
+      "loss": 3.8653,
+      "step": 1946624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8119542992066763e-05,
+      "loss": 3.8819,
+      "step": 1947136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8111157044556243e-05,
+      "loss": 3.8729,
+      "step": 1947648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8102771097045723e-05,
+      "loss": 3.8675,
+      "step": 1948160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8094385149535203e-05,
+      "loss": 3.8805,
+      "step": 1948672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8085999202024683e-05,
+      "loss": 3.8837,
+      "step": 1949184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8077613254514163e-05,
+      "loss": 3.8837,
+      "step": 1949696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8069243685807375e-05,
+      "loss": 3.8977,
+      "step": 1950208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8060874117100585e-05,
+      "loss": 3.8719,
+      "step": 1950720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8052488169590064e-05,
+      "loss": 3.8659,
+      "step": 1951232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8044102222079544e-05,
+      "loss": 3.8704,
+      "step": 1951744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8035716274569024e-05,
+      "loss": 3.8843,
+      "step": 1952256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8027330327058504e-05,
+      "loss": 3.8727,
+      "step": 1952768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8018944379547988e-05,
+      "loss": 3.8818,
+      "step": 1953280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8010558432037464e-05,
+      "loss": 3.876,
+      "step": 1953792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8002172484526944e-05,
+      "loss": 3.8784,
+      "step": 1954304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7993786537016424e-05,
+      "loss": 3.8781,
+      "step": 1954816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7985416968309637e-05,
+      "loss": 3.8654,
+      "step": 1955328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7977031020799117e-05,
+      "loss": 3.8786,
+      "step": 1955840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7968645073288597e-05,
+      "loss": 3.8879,
+      "step": 1956352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7960259125778076e-05,
+      "loss": 3.8808,
+      "step": 1956864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7951873178267556e-05,
+      "loss": 3.8611,
+      "step": 1957376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7943487230757036e-05,
+      "loss": 3.8745,
+      "step": 1957888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7935101283246516e-05,
+      "loss": 3.8716,
+      "step": 1958400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7926715335735996e-05,
+      "loss": 3.8813,
+      "step": 1958912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7918362145832938e-05,
+      "loss": 3.8825,
+      "step": 1959424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7909976198322418e-05,
+      "loss": 3.8678,
+      "step": 1959936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7901590250811898e-05,
+      "loss": 3.872,
+      "step": 1960448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7893204303301378e-05,
+      "loss": 3.8652,
+      "step": 1960960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7884818355790858e-05,
+      "loss": 3.8741,
+      "step": 1961472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.787644878708407e-05,
+      "loss": 3.868,
+      "step": 1961984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.786806283957355e-05,
+      "loss": 3.865,
+      "step": 1962496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.785967689206303e-05,
+      "loss": 3.8768,
+      "step": 1963008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.785129094455251e-05,
+      "loss": 3.8752,
+      "step": 1963520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.784290499704199e-05,
+      "loss": 3.871,
+      "step": 1964032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.783451904953147e-05,
+      "loss": 3.8672,
+      "step": 1964544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7826133102020947e-05,
+      "loss": 3.8683,
+      "step": 1965056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.781776353331416e-05,
+      "loss": 3.8805,
+      "step": 1965568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7809377585803642e-05,
+      "loss": 3.8619,
+      "step": 1966080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.780099163829312e-05,
+      "loss": 3.8839,
+      "step": 1966592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.77926056907826e-05,
+      "loss": 3.8795,
+      "step": 1967104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.778421974327208e-05,
+      "loss": 3.8721,
+      "step": 1967616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.777585017456529e-05,
+      "loss": 3.8814,
+      "step": 1968128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.776746422705477e-05,
+      "loss": 3.8716,
+      "step": 1968640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.775907827954425e-05,
+      "loss": 3.8807,
+      "step": 1969152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.775069233203373e-05,
+      "loss": 3.8762,
+      "step": 1969664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.774230638452321e-05,
+      "loss": 3.8751,
+      "step": 1970176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.773392043701269e-05,
+      "loss": 3.8757,
+      "step": 1970688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.77255508683059e-05,
+      "loss": 3.8761,
+      "step": 1971200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.771716492079538e-05,
+      "loss": 3.8789,
+      "step": 1971712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7708778973284864e-05,
+      "loss": 3.8699,
+      "step": 1972224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7700393025774344e-05,
+      "loss": 3.8594,
+      "step": 1972736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7692007078263823e-05,
+      "loss": 3.873,
+      "step": 1973248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7683637509557033e-05,
+      "loss": 3.8629,
+      "step": 1973760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7675251562046513e-05,
+      "loss": 3.8833,
+      "step": 1974272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7666865614535992e-05,
+      "loss": 3.8695,
+      "step": 1974784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7658479667025472e-05,
+      "loss": 3.8831,
+      "step": 1975296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7650093719514952e-05,
+      "loss": 3.8622,
+      "step": 1975808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7641707772004432e-05,
+      "loss": 3.8856,
+      "step": 1976320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7633321824493912e-05,
+      "loss": 3.8758,
+      "step": 1976832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7624952255787125e-05,
+      "loss": 3.8869,
+      "step": 1977344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.76165663082766e-05,
+      "loss": 3.8781,
+      "step": 1977856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.760818036076608e-05,
+      "loss": 3.8765,
+      "step": 1978368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7599794413255565e-05,
+      "loss": 3.8699,
+      "step": 1978880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7591408465745045e-05,
+      "loss": 3.8893,
+      "step": 1979392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7583038897038254e-05,
+      "loss": 3.8617,
+      "step": 1979904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7574669328331466e-05,
+      "loss": 3.8844,
+      "step": 1980416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7566283380820946e-05,
+      "loss": 3.8729,
+      "step": 1980928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7557897433310426e-05,
+      "loss": 3.8759,
+      "step": 1981440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7549511485799906e-05,
+      "loss": 3.8766,
+      "step": 1981952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7541125538289386e-05,
+      "loss": 3.873,
+      "step": 1982464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7532739590778866e-05,
+      "loss": 3.8743,
+      "step": 1982976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7524353643268346e-05,
+      "loss": 3.8737,
+      "step": 1983488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7515967695757826e-05,
+      "loss": 3.8682,
+      "step": 1984000
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9844439029693604,
+      "eval_runtime": 286.0624,
+      "eval_samples_per_second": 1333.943,
+      "eval_steps_per_second": 41.687,
+      "step": 1984320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7507598127051035e-05,
+      "loss": 3.8644,
+      "step": 1984512
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749921217954052e-05,
+      "loss": 3.8685,
+      "step": 1985024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7490826232029998e-05,
+      "loss": 3.8774,
+      "step": 1985536
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7482440284519478e-05,
+      "loss": 3.8793,
+      "step": 1986048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7474054337008955e-05,
+      "loss": 3.8899,
+      "step": 1986560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7465668389498435e-05,
+      "loss": 3.8705,
+      "step": 1987072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7457282441987915e-05,
+      "loss": 3.8718,
+      "step": 1987584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7448896494477395e-05,
+      "loss": 3.8768,
+      "step": 1988096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7440510546966875e-05,
+      "loss": 3.8665,
+      "step": 1988608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.743215735706382e-05,
+      "loss": 3.8772,
+      "step": 1989120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.74237714095533e-05,
+      "loss": 3.8774,
+      "step": 1989632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7415385462042776e-05,
+      "loss": 3.8795,
+      "step": 1990144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.740701589333599e-05,
+      "loss": 3.8825,
+      "step": 1990656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7398629945825472e-05,
+      "loss": 3.8626,
+      "step": 1991168
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7390243998314952e-05,
+      "loss": 3.8625,
+      "step": 1991680
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.738185805080443e-05,
+      "loss": 3.8597,
+      "step": 1992192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.737347210329391e-05,
+      "loss": 3.8853,
+      "step": 1992704
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.736508615578339e-05,
+      "loss": 3.8756,
+      "step": 1993216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.735670020827287e-05,
+      "loss": 3.8678,
+      "step": 1993728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.734831426076235e-05,
+      "loss": 3.8848,
+      "step": 1994240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7339928313251828e-05,
+      "loss": 3.8899,
+      "step": 1994752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.733154236574131e-05,
+      "loss": 3.8762,
+      "step": 1995264
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7323156418230788e-05,
+      "loss": 3.8791,
+      "step": 1995776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7314770470720268e-05,
+      "loss": 3.8766,
+      "step": 1996288
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7306384523209748e-05,
+      "loss": 3.8815,
+      "step": 1996800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7298014954502957e-05,
+      "loss": 3.8726,
+      "step": 1997312
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7289629006992437e-05,
+      "loss": 3.8668,
+      "step": 1997824
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7281259438285653e-05,
+      "loss": 3.8795,
+      "step": 1998336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7272873490775133e-05,
+      "loss": 3.8674,
+      "step": 1998848
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.726448754326461e-05,
+      "loss": 3.8629,
+      "step": 1999360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.725610159575409e-05,
+      "loss": 3.872,
+      "step": 1999872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.724771564824357e-05,
+      "loss": 3.8704,
+      "step": 2000384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.723932970073305e-05,
+      "loss": 3.8788,
+      "step": 2000896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.723094375322253e-05,
+      "loss": 3.8866,
+      "step": 2001408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7222557805712013e-05,
+      "loss": 3.8693,
+      "step": 2001920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7214188237005222e-05,
+      "loss": 3.8757,
+      "step": 2002432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7205802289494702e-05,
+      "loss": 3.8842,
+      "step": 2002944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7197416341984182e-05,
+      "loss": 3.8555,
+      "step": 2003456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.718904677327739e-05,
+      "loss": 3.8686,
+      "step": 2003968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7180660825766874e-05,
+      "loss": 3.8712,
+      "step": 2004480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7172274878256354e-05,
+      "loss": 3.8616,
+      "step": 2004992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7163888930745834e-05,
+      "loss": 3.8712,
+      "step": 2005504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7155502983235314e-05,
+      "loss": 3.8736,
+      "step": 2006016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7147133414528523e-05,
+      "loss": 3.8777,
+      "step": 2006528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7138747467018003e-05,
+      "loss": 3.8673,
+      "step": 2007040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7130361519507483e-05,
+      "loss": 3.8706,
+      "step": 2007552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7121975571996966e-05,
+      "loss": 3.877,
+      "step": 2008064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7113589624486443e-05,
+      "loss": 3.8715,
+      "step": 2008576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7105203676975923e-05,
+      "loss": 3.8467,
+      "step": 2009088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7096817729465403e-05,
+      "loss": 3.8781,
+      "step": 2009600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7088431781954883e-05,
+      "loss": 3.873,
+      "step": 2010112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7080062213248095e-05,
+      "loss": 3.8788,
+      "step": 2010624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7071676265737575e-05,
+      "loss": 3.8653,
+      "step": 2011136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7063290318227055e-05,
+      "loss": 3.8616,
+      "step": 2011648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7054904370716535e-05,
+      "loss": 3.8633,
+      "step": 2012160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7046534802009744e-05,
+      "loss": 3.8665,
+      "step": 2012672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7038148854499224e-05,
+      "loss": 3.8808,
+      "step": 2013184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7029762906988704e-05,
+      "loss": 3.8559,
+      "step": 2013696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7021376959478184e-05,
+      "loss": 3.8815,
+      "step": 2014208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7012991011967667e-05,
+      "loss": 3.8754,
+      "step": 2014720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7004621443260877e-05,
+      "loss": 3.851,
+      "step": 2015232
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6996235495750357e-05,
+      "loss": 3.8693,
+      "step": 2015744
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6987849548239836e-05,
+      "loss": 3.8556,
+      "step": 2016256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6979463600729316e-05,
+      "loss": 3.858,
+      "step": 2016768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.697109403202253e-05,
+      "loss": 3.8677,
+      "step": 2017280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.696270808451201e-05,
+      "loss": 3.8821,
+      "step": 2017792
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.695432213700149e-05,
+      "loss": 3.86,
+      "step": 2018304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.694593618949097e-05,
+      "loss": 3.8735,
+      "step": 2018816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6937550241980445e-05,
+      "loss": 3.8539,
+      "step": 2019328
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6929164294469925e-05,
+      "loss": 3.8669,
+      "step": 2019840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6920794725763138e-05,
+      "loss": 3.8724,
+      "step": 2020352
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6912408778252618e-05,
+      "loss": 3.875,
+      "step": 2020864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6904022830742098e-05,
+      "loss": 3.8688,
+      "step": 2021376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6895636883231578e-05,
+      "loss": 3.8705,
+      "step": 2021888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6887250935721058e-05,
+      "loss": 3.8887,
+      "step": 2022400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6878881367014267e-05,
+      "loss": 3.8625,
+      "step": 2022912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.687049541950375e-05,
+      "loss": 3.8733,
+      "step": 2023424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.686210947199323e-05,
+      "loss": 3.8678,
+      "step": 2023936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.685372352448271e-05,
+      "loss": 3.866,
+      "step": 2024448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.684533757697219e-05,
+      "loss": 3.877,
+      "step": 2024960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.683695162946167e-05,
+      "loss": 3.8725,
+      "step": 2025472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.682858206075488e-05,
+      "loss": 3.8815,
+      "step": 2025984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.682019611324436e-05,
+      "loss": 3.8892,
+      "step": 2026496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6811810165733842e-05,
+      "loss": 3.8758,
+      "step": 2027008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.6803424218223322e-05,
+      "loss": 3.857,
+      "step": 2027520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.67950382707128e-05,
+      "loss": 3.8654,
+      "step": 2028032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.678665232320228e-05,
+      "loss": 3.881,
+      "step": 2028544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.677828275449549e-05,
+      "loss": 3.8693,
+      "step": 2029056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.676989680698497e-05,
+      "loss": 3.8746,
+      "step": 2029568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.676151085947445e-05,
+      "loss": 3.8719,
+      "step": 2030080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.675312491196393e-05,
+      "loss": 3.874,
+      "step": 2030592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.674473896445341e-05,
+      "loss": 3.8716,
+      "step": 2031104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6736369395746624e-05,
+      "loss": 3.8613,
+      "step": 2031616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.67279834482361e-05,
+      "loss": 3.8782,
+      "step": 2032128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.671959750072558e-05,
+      "loss": 3.8808,
+      "step": 2032640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.671121155321506e-05,
+      "loss": 3.8773,
+      "step": 2033152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6702841984508273e-05,
+      "loss": 3.8569,
+      "step": 2033664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6694456036997752e-05,
+      "loss": 3.8694,
+      "step": 2034176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6686070089487232e-05,
+      "loss": 3.8708,
+      "step": 2034688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6677684141976712e-05,
+      "loss": 3.876,
+      "step": 2035200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6669298194466192e-05,
+      "loss": 3.8767,
+      "step": 2035712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6660912246955672e-05,
+      "loss": 3.8612,
+      "step": 2036224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6652526299445152e-05,
+      "loss": 3.8702,
+      "step": 2036736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6644140351934632e-05,
+      "loss": 3.8584,
+      "step": 2037248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6635770783227845e-05,
+      "loss": 3.8689,
+      "step": 2037760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6627401214521054e-05,
+      "loss": 3.8658,
+      "step": 2038272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6619015267010534e-05,
+      "loss": 3.8622,
+      "step": 2038784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6610629319500014e-05,
+      "loss": 3.8734,
+      "step": 2039296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6602243371989497e-05,
+      "loss": 3.8691,
+      "step": 2039808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6593857424478977e-05,
+      "loss": 3.8658,
+      "step": 2040320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6585487855772186e-05,
+      "loss": 3.8637,
+      "step": 2040832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6577101908261666e-05,
+      "loss": 3.8604,
+      "step": 2041344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6568715960751146e-05,
+      "loss": 3.8755,
+      "step": 2041856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6560330013240626e-05,
+      "loss": 3.8618,
+      "step": 2042368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6551944065730106e-05,
+      "loss": 3.8727,
+      "step": 2042880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6543558118219586e-05,
+      "loss": 3.875,
+      "step": 2043392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6535172170709066e-05,
+      "loss": 3.8697,
+      "step": 2043904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.652680260200228e-05,
+      "loss": 3.8747,
+      "step": 2044416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6518416654491755e-05,
+      "loss": 3.8708,
+      "step": 2044928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6510030706981235e-05,
+      "loss": 3.873,
+      "step": 2045440
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6501644759470715e-05,
+      "loss": 3.8785,
+      "step": 2045952
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6493258811960198e-05,
+      "loss": 3.869,
+      "step": 2046464
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6484889243253407e-05,
+      "loss": 3.8707,
+      "step": 2046976
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6476503295742887e-05,
+      "loss": 3.8714,
+      "step": 2047488
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6468117348232367e-05,
+      "loss": 3.8771,
+      "step": 2048000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6459731400721847e-05,
+      "loss": 3.8622,
+      "step": 2048512
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6451345453211327e-05,
+      "loss": 3.8553,
+      "step": 2049024
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.644297588450454e-05,
+      "loss": 3.8665,
+      "step": 2049536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.643458993699402e-05,
+      "loss": 3.8589,
+      "step": 2050048
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.64262039894835e-05,
+      "loss": 3.8767,
+      "step": 2050560
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.641781804197298e-05,
+      "loss": 3.8654,
+      "step": 2051072
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.640943209446246e-05,
+      "loss": 3.8795,
+      "step": 2051584
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6401046146951936e-05,
+      "loss": 3.8612,
+      "step": 2052096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6392676578245152e-05,
+      "loss": 3.8782,
+      "step": 2052608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6384290630734632e-05,
+      "loss": 3.8707,
+      "step": 2053120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.637590468322411e-05,
+      "loss": 3.8855,
+      "step": 2053632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6367518735713588e-05,
+      "loss": 3.8748,
+      "step": 2054144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.63591491670068e-05,
+      "loss": 3.8729,
+      "step": 2054656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.635076321949628e-05,
+      "loss": 3.8656,
+      "step": 2055168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.634237727198576e-05,
+      "loss": 3.8809,
+      "step": 2055680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.633399132447524e-05,
+      "loss": 3.8613,
+      "step": 2056192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.632560537696472e-05,
+      "loss": 3.8736,
+      "step": 2056704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.63172194294542e-05,
+      "loss": 3.8731,
+      "step": 2057216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.630884986074741e-05,
+      "loss": 3.8702,
+      "step": 2057728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.630046391323689e-05,
+      "loss": 3.8715,
+      "step": 2058240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.629207796572637e-05,
+      "loss": 3.8713,
+      "step": 2058752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6283692018215853e-05,
+      "loss": 3.8696,
+      "step": 2059264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6275322449509062e-05,
+      "loss": 3.8683,
+      "step": 2059776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.6266936501998542e-05,
+      "loss": 3.8638,
+      "step": 2060288
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.983138084411621,
+      "eval_runtime": 286.1047,
+      "eval_samples_per_second": 1333.746,
+      "eval_steps_per_second": 41.681,
+      "step": 2060640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6258550554488022e-05,
+      "loss": 3.8525,
+      "step": 2060800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6250164606977502e-05,
+      "loss": 3.8622,
+      "step": 2061312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6241778659466982e-05,
+      "loss": 3.8719,
+      "step": 2061824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6233392711956462e-05,
+      "loss": 3.8766,
+      "step": 2062336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6225006764445942e-05,
+      "loss": 3.884,
+      "step": 2062848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.621662081693542e-05,
+      "loss": 3.864,
+      "step": 2063360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.62082348694249e-05,
+      "loss": 3.8682,
+      "step": 2063872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.619984892191438e-05,
+      "loss": 3.8714,
+      "step": 2064384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.619146297440386e-05,
+      "loss": 3.8644,
+      "step": 2064896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.618307702689334e-05,
+      "loss": 3.8672,
+      "step": 2065408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.617469107938282e-05,
+      "loss": 3.8792,
+      "step": 2065920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.61663051318723e-05,
+      "loss": 3.8736,
+      "step": 2066432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6157935563165514e-05,
+      "loss": 3.8807,
+      "step": 2066944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6149549615654994e-05,
+      "loss": 3.8577,
+      "step": 2067456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.614116366814447e-05,
+      "loss": 3.8587,
+      "step": 2067968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.613277772063395e-05,
+      "loss": 3.8597,
+      "step": 2068480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.612439177312343e-05,
+      "loss": 3.8765,
+      "step": 2068992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6116005825612914e-05,
+      "loss": 3.8748,
+      "step": 2069504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6107619878102394e-05,
+      "loss": 3.8625,
+      "step": 2070016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6099233930591874e-05,
+      "loss": 3.8791,
+      "step": 2070528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6090847983081354e-05,
+      "loss": 3.8877,
+      "step": 2071040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6082478414374563e-05,
+      "loss": 3.8684,
+      "step": 2071552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6074092466864043e-05,
+      "loss": 3.8758,
+      "step": 2072064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6065706519353523e-05,
+      "loss": 3.8726,
+      "step": 2072576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6057320571843003e-05,
+      "loss": 3.8781,
+      "step": 2073088
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6048934624332482e-05,
+      "loss": 3.8653,
+      "step": 2073600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6040565055625695e-05,
+      "loss": 3.8625,
+      "step": 2074112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6032179108115175e-05,
+      "loss": 3.8774,
+      "step": 2074624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.602379316060465e-05,
+      "loss": 3.8634,
+      "step": 2075136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.601540721309413e-05,
+      "loss": 3.8602,
+      "step": 2075648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6007021265583615e-05,
+      "loss": 3.8641,
+      "step": 2076160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5998651696876827e-05,
+      "loss": 3.8658,
+      "step": 2076672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5990265749366304e-05,
+      "loss": 3.8745,
+      "step": 2077184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5981879801855784e-05,
+      "loss": 3.8842,
+      "step": 2077696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5973493854345264e-05,
+      "loss": 3.8594,
+      "step": 2078208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5965124285638476e-05,
+      "loss": 3.8712,
+      "step": 2078720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5956738338127956e-05,
+      "loss": 3.8794,
+      "step": 2079232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5948352390617436e-05,
+      "loss": 3.8499,
+      "step": 2079744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5939966443106916e-05,
+      "loss": 3.8622,
+      "step": 2080256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5931580495596396e-05,
+      "loss": 3.8708,
+      "step": 2080768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5923194548085876e-05,
+      "loss": 3.8545,
+      "step": 2081280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5914808600575356e-05,
+      "loss": 3.8702,
+      "step": 2081792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5906422653064833e-05,
+      "loss": 3.8724,
+      "step": 2082304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5898036705554316e-05,
+      "loss": 3.8705,
+      "step": 2082816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5889650758043796e-05,
+      "loss": 3.863,
+      "step": 2083328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5881264810533276e-05,
+      "loss": 3.8708,
+      "step": 2083840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5872895241826485e-05,
+      "loss": 3.8685,
+      "step": 2084352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5864509294315965e-05,
+      "loss": 3.8695,
+      "step": 2084864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5856123346805445e-05,
+      "loss": 3.8423,
+      "step": 2085376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5847737399294925e-05,
+      "loss": 3.87,
+      "step": 2085888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5839351451784408e-05,
+      "loss": 3.8714,
+      "step": 2086400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5830981883077617e-05,
+      "loss": 3.8741,
+      "step": 2086912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5822595935567097e-05,
+      "loss": 3.8596,
+      "step": 2087424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5814209988056577e-05,
+      "loss": 3.8601,
+      "step": 2087936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5805824040546057e-05,
+      "loss": 3.8589,
+      "step": 2088448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5797438093035537e-05,
+      "loss": 3.8596,
+      "step": 2088960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578906852432875e-05,
+      "loss": 3.8782,
+      "step": 2089472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.578068257681823e-05,
+      "loss": 3.8543,
+      "step": 2089984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.577229662930771e-05,
+      "loss": 3.8703,
+      "step": 2090496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5763910681797186e-05,
+      "loss": 3.8721,
+      "step": 2091008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5755524734286666e-05,
+      "loss": 3.8525,
+      "step": 2091520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5747138786776146e-05,
+      "loss": 3.8582,
+      "step": 2092032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5738769218069362e-05,
+      "loss": 3.8556,
+      "step": 2092544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.573038327055884e-05,
+      "loss": 3.8561,
+      "step": 2093056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5721997323048318e-05,
+      "loss": 3.8554,
+      "step": 2093568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5713611375537798e-05,
+      "loss": 3.8797,
+      "step": 2094080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5705225428027278e-05,
+      "loss": 3.8596,
+      "step": 2094592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5696855859320487e-05,
+      "loss": 3.8606,
+      "step": 2095104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.568846991180997e-05,
+      "loss": 3.852,
+      "step": 2095616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.568008396429945e-05,
+      "loss": 3.8607,
+      "step": 2096128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.567169801678893e-05,
+      "loss": 3.8676,
+      "step": 2096640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.566332844808214e-05,
+      "loss": 3.8693,
+      "step": 2097152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5654958879375352e-05,
+      "loss": 3.8667,
+      "step": 2097664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5646572931864832e-05,
+      "loss": 3.8686,
+      "step": 2098176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5638186984354312e-05,
+      "loss": 3.8802,
+      "step": 2098688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5629801036843792e-05,
+      "loss": 3.8596,
+      "step": 2099200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5621415089333272e-05,
+      "loss": 3.8688,
+      "step": 2099712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5613029141822752e-05,
+      "loss": 3.8646,
+      "step": 2100224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5604643194312232e-05,
+      "loss": 3.8622,
+      "step": 2100736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5596257246801712e-05,
+      "loss": 3.8723,
+      "step": 2101248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5587871299291192e-05,
+      "loss": 3.8674,
+      "step": 2101760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5579501730584404e-05,
+      "loss": 3.8782,
+      "step": 2102272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5571115783073884e-05,
+      "loss": 3.8817,
+      "step": 2102784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5562729835563364e-05,
+      "loss": 3.8762,
+      "step": 2103296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.555434388805284e-05,
+      "loss": 3.8509,
+      "step": 2103808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.554595794054232e-05,
+      "loss": 3.8625,
+      "step": 2104320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.55375719930318e-05,
+      "loss": 3.8709,
+      "step": 2104832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5529186045521284e-05,
+      "loss": 3.8686,
+      "step": 2105344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5520800098010764e-05,
+      "loss": 3.8682,
+      "step": 2105856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5512430529303973e-05,
+      "loss": 3.8709,
+      "step": 2106368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5504044581793453e-05,
+      "loss": 3.8704,
+      "step": 2106880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5495658634282933e-05,
+      "loss": 3.8656,
+      "step": 2107392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5487272686772413e-05,
+      "loss": 3.8594,
+      "step": 2107904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5478886739261893e-05,
+      "loss": 3.871,
+      "step": 2108416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5470517170555105e-05,
+      "loss": 3.8724,
+      "step": 2108928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5462131223044585e-05,
+      "loss": 3.8761,
+      "step": 2109440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5453745275534065e-05,
+      "loss": 3.8513,
+      "step": 2109952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5445359328023545e-05,
+      "loss": 3.8633,
+      "step": 2110464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5436973380513022e-05,
+      "loss": 3.8646,
+      "step": 2110976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5428587433002502e-05,
+      "loss": 3.8722,
+      "step": 2111488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5420201485491985e-05,
+      "loss": 3.8726,
+      "step": 2112000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5411815537981465e-05,
+      "loss": 3.8564,
+      "step": 2112512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5403462348078407e-05,
+      "loss": 3.8664,
+      "step": 2113024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5395076400567887e-05,
+      "loss": 3.8516,
+      "step": 2113536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5386690453057367e-05,
+      "loss": 3.8703,
+      "step": 2114048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5378304505546847e-05,
+      "loss": 3.8588,
+      "step": 2114560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5369918558036326e-05,
+      "loss": 3.857,
+      "step": 2115072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5361532610525806e-05,
+      "loss": 3.8703,
+      "step": 2115584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535316304181902e-05,
+      "loss": 3.8646,
+      "step": 2116096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5344777094308495e-05,
+      "loss": 3.8594,
+      "step": 2116608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5336391146797975e-05,
+      "loss": 3.8621,
+      "step": 2117120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5328005199287455e-05,
+      "loss": 3.8556,
+      "step": 2117632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.531961925177694e-05,
+      "loss": 3.8743,
+      "step": 2118144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5311249683070148e-05,
+      "loss": 3.8544,
+      "step": 2118656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5302863735559628e-05,
+      "loss": 3.871,
+      "step": 2119168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5294477788049108e-05,
+      "loss": 3.8656,
+      "step": 2119680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5286091840538588e-05,
+      "loss": 3.8697,
+      "step": 2120192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5277705893028068e-05,
+      "loss": 3.8687,
+      "step": 2120704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5269319945517548e-05,
+      "loss": 3.8643,
+      "step": 2121216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.526095037681076e-05,
+      "loss": 3.8743,
+      "step": 2121728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.525256442930024e-05,
+      "loss": 3.8703,
+      "step": 2122240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.524417848178972e-05,
+      "loss": 3.8649,
+      "step": 2122752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.52357925342792e-05,
+      "loss": 3.8641,
+      "step": 2123264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5227406586768678e-05,
+      "loss": 3.8649,
+      "step": 2123776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5219020639258158e-05,
+      "loss": 3.8731,
+      "step": 2124288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5210634691747638e-05,
+      "loss": 3.8578,
+      "step": 2124800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5202248744237118e-05,
+      "loss": 3.8554,
+      "step": 2125312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5193879175530329e-05,
+      "loss": 3.8582,
+      "step": 2125824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5185493228019809e-05,
+      "loss": 3.857,
+      "step": 2126336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5177107280509289e-05,
+      "loss": 3.8702,
+      "step": 2126848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.516872133299877e-05,
+      "loss": 3.8627,
+      "step": 2127360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.516035176429198e-05,
+      "loss": 3.8777,
+      "step": 2127872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5151965816781461e-05,
+      "loss": 3.8553,
+      "step": 2128384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5143579869270941e-05,
+      "loss": 3.8746,
+      "step": 2128896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135193921760421e-05,
+      "loss": 3.8687,
+      "step": 2129408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5126824353053632e-05,
+      "loss": 3.8744,
+      "step": 2129920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5118454784346844e-05,
+      "loss": 3.8695,
+      "step": 2130432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5110085215640055e-05,
+      "loss": 3.8687,
+      "step": 2130944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5101699268129535e-05,
+      "loss": 3.864,
+      "step": 2131456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5093313320619015e-05,
+      "loss": 3.8758,
+      "step": 2131968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5084927373108495e-05,
+      "loss": 3.861,
+      "step": 2132480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5076541425597973e-05,
+      "loss": 3.8656,
+      "step": 2132992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5068155478087453e-05,
+      "loss": 3.8714,
+      "step": 2133504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5059785909380666e-05,
+      "loss": 3.8653,
+      "step": 2134016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5051399961870147e-05,
+      "loss": 3.8622,
+      "step": 2134528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5043014014359624e-05,
+      "loss": 3.8726,
+      "step": 2135040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5034628066849104e-05,
+      "loss": 3.8617,
+      "step": 2135552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5026242119338586e-05,
+      "loss": 3.8659,
+      "step": 2136064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5017856171828066e-05,
+      "loss": 3.8615,
+      "step": 2136576
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.981598138809204,
+      "eval_runtime": 293.4084,
+      "eval_samples_per_second": 1300.546,
+      "eval_steps_per_second": 40.643,
+      "step": 2136960
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5009470224317545e-05,
+      "loss": 3.8609,
+      "step": 2137088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5001084276807025e-05,
+      "loss": 3.8537,
+      "step": 2137600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4992698329296505e-05,
+      "loss": 3.8688,
+      "step": 2138112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4984328760589716e-05,
+      "loss": 3.8697,
+      "step": 2138624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4975942813079196e-05,
+      "loss": 3.8739,
+      "step": 2139136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4967556865568678e-05,
+      "loss": 3.867,
+      "step": 2139648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4959170918058154e-05,
+      "loss": 3.8645,
+      "step": 2140160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4950801349351367e-05,
+      "loss": 3.8676,
+      "step": 2140672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4942415401840848e-05,
+      "loss": 3.8581,
+      "step": 2141184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4934029454330328e-05,
+      "loss": 3.8646,
+      "step": 2141696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4925643506819805e-05,
+      "loss": 3.8698,
+      "step": 2142208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4917257559309287e-05,
+      "loss": 3.8678,
+      "step": 2142720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.490890436940623e-05,
+      "loss": 3.8791,
+      "step": 2143232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.490051842189571e-05,
+      "loss": 3.8531,
+      "step": 2143744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.489213247438519e-05,
+      "loss": 3.8532,
+      "step": 2144256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.488374652687467e-05,
+      "loss": 3.8498,
+      "step": 2144768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.487536057936415e-05,
+      "loss": 3.8781,
+      "step": 2145280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4866974631853628e-05,
+      "loss": 3.8672,
+      "step": 2145792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4858588684343108e-05,
+      "loss": 3.8612,
+      "step": 2146304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4850202736832588e-05,
+      "loss": 3.8737,
+      "step": 2146816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.484181678932207e-05,
+      "loss": 3.8798,
+      "step": 2147328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.483343084181155e-05,
+      "loss": 3.8672,
+      "step": 2147840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.482504489430103e-05,
+      "loss": 3.8685,
+      "step": 2148352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.481665894679051e-05,
+      "loss": 3.8674,
+      "step": 2148864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4808272999279988e-05,
+      "loss": 3.8767,
+      "step": 2149376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4799919809376931e-05,
+      "loss": 3.863,
+      "step": 2149888
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4791533861866411e-05,
+      "loss": 3.8579,
+      "step": 2150400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4783147914355891e-05,
+      "loss": 3.8755,
+      "step": 2150912
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4774778345649102e-05,
+      "loss": 3.8503,
+      "step": 2151424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4766392398138582e-05,
+      "loss": 3.8571,
+      "step": 2151936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4758006450628062e-05,
+      "loss": 3.8591,
+      "step": 2152448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4749620503117542e-05,
+      "loss": 3.8614,
+      "step": 2152960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4741234555607023e-05,
+      "loss": 3.871,
+      "step": 2153472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4732848608096503e-05,
+      "loss": 3.8772,
+      "step": 2153984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4724462660585983e-05,
+      "loss": 3.8613,
+      "step": 2154496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4716093091879194e-05,
+      "loss": 3.8645,
+      "step": 2155008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4707707144368674e-05,
+      "loss": 3.8742,
+      "step": 2155520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4699321196858154e-05,
+      "loss": 3.8461,
+      "step": 2156032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4690935249347634e-05,
+      "loss": 3.8636,
+      "step": 2156544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4682549301837112e-05,
+      "loss": 3.8616,
+      "step": 2157056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4674163354326592e-05,
+      "loss": 3.8509,
+      "step": 2157568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4665777406816072e-05,
+      "loss": 3.8637,
+      "step": 2158080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4657391459305552e-05,
+      "loss": 3.872,
+      "step": 2158592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4649005511795034e-05,
+      "loss": 3.8605,
+      "step": 2159104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4640619564284514e-05,
+      "loss": 3.859,
+      "step": 2159616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.463223361677399e-05,
+      "loss": 3.8696,
+      "step": 2160128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4623847669263472e-05,
+      "loss": 3.8623,
+      "step": 2160640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4615478100556684e-05,
+      "loss": 3.8675,
+      "step": 2161152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4607092153046164e-05,
+      "loss": 3.8355,
+      "step": 2161664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4598706205535643e-05,
+      "loss": 3.8606,
+      "step": 2162176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4590320258025122e-05,
+      "loss": 3.8713,
+      "step": 2162688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4581950689318335e-05,
+      "loss": 3.8653,
+      "step": 2163200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4573564741807813e-05,
+      "loss": 3.8536,
+      "step": 2163712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4565178794297293e-05,
+      "loss": 3.8568,
+      "step": 2164224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4556792846786773e-05,
+      "loss": 3.8516,
+      "step": 2164736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4548423278079987e-05,
+      "loss": 3.8532,
+      "step": 2165248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4540037330569464e-05,
+      "loss": 3.881,
+      "step": 2165760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4531651383058944e-05,
+      "loss": 3.8517,
+      "step": 2166272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4523265435548425e-05,
+      "loss": 3.8643,
+      "step": 2166784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4514879488037905e-05,
+      "loss": 3.8672,
+      "step": 2167296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4506509919331116e-05,
+      "loss": 3.8483,
+      "step": 2167808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4498123971820596e-05,
+      "loss": 3.8542,
+      "step": 2168320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4489738024310076e-05,
+      "loss": 3.8518,
+      "step": 2168832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4481352076799556e-05,
+      "loss": 3.8495,
+      "step": 2169344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4472966129289036e-05,
+      "loss": 3.8557,
+      "step": 2169856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4464580181778518e-05,
+      "loss": 3.8725,
+      "step": 2170368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4456194234267994e-05,
+      "loss": 3.862,
+      "step": 2170880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4447808286757474e-05,
+      "loss": 3.8568,
+      "step": 2171392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4439438718050688e-05,
+      "loss": 3.8451,
+      "step": 2171904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4431052770540168e-05,
+      "loss": 3.8592,
+      "step": 2172416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4422683201833379e-05,
+      "loss": 3.8632,
+      "step": 2172928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4414313633126588e-05,
+      "loss": 3.8686,
+      "step": 2173440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.440592768561607e-05,
+      "loss": 3.8611,
+      "step": 2173952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.439754173810555e-05,
+      "loss": 3.8644,
+      "step": 2174464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.438915579059503e-05,
+      "loss": 3.8764,
+      "step": 2174976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.438076984308451e-05,
+      "loss": 3.856,
+      "step": 2175488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.437240027437772e-05,
+      "loss": 3.861,
+      "step": 2176000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.43640143268672e-05,
+      "loss": 3.8651,
+      "step": 2176512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.435562837935668e-05,
+      "loss": 3.8592,
+      "step": 2177024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.434724243184616e-05,
+      "loss": 3.8652,
+      "step": 2177536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4338856484335642e-05,
+      "loss": 3.8651,
+      "step": 2178048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4330470536825119e-05,
+      "loss": 3.8712,
+      "step": 2178560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4322084589314599e-05,
+      "loss": 3.8813,
+      "step": 2179072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4313715020607813e-05,
+      "loss": 3.8694,
+      "step": 2179584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4305329073097293e-05,
+      "loss": 3.8475,
+      "step": 2180096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4296943125586771e-05,
+      "loss": 3.8566,
+      "step": 2180608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4288557178076251e-05,
+      "loss": 3.8705,
+      "step": 2181120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4280171230565731e-05,
+      "loss": 3.8669,
+      "step": 2181632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.427178528305521e-05,
+      "loss": 3.8596,
+      "step": 2182144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4263415714348422e-05,
+      "loss": 3.8678,
+      "step": 2182656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4255029766837902e-05,
+      "loss": 3.868,
+      "step": 2183168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4246643819327382e-05,
+      "loss": 3.8587,
+      "step": 2183680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4238257871816863e-05,
+      "loss": 3.8562,
+      "step": 2184192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4229871924306343e-05,
+      "loss": 3.8685,
+      "step": 2184704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4221485976795823e-05,
+      "loss": 3.8615,
+      "step": 2185216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4213100029285301e-05,
+      "loss": 3.8727,
+      "step": 2185728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4204714081774781e-05,
+      "loss": 3.8547,
+      "step": 2186240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4196344513067994e-05,
+      "loss": 3.8561,
+      "step": 2186752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4187974944361205e-05,
+      "loss": 3.8653,
+      "step": 2187264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4179588996850685e-05,
+      "loss": 3.8667,
+      "step": 2187776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4171203049340165e-05,
+      "loss": 3.8686,
+      "step": 2188288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4162817101829644e-05,
+      "loss": 3.8518,
+      "step": 2188800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4154431154319123e-05,
+      "loss": 3.8594,
+      "step": 2189312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4146045206808603e-05,
+      "loss": 3.8495,
+      "step": 2189824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4137659259298083e-05,
+      "loss": 3.8636,
+      "step": 2190336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4129273311787564e-05,
+      "loss": 3.8527,
+      "step": 2190848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4120903743080773e-05,
+      "loss": 3.8533,
+      "step": 2191360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4112517795570255e-05,
+      "loss": 3.8638,
+      "step": 2191872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4104131848059735e-05,
+      "loss": 3.8587,
+      "step": 2192384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4095745900549215e-05,
+      "loss": 3.8623,
+      "step": 2192896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4087359953038695e-05,
+      "loss": 3.8604,
+      "step": 2193408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4078974005528175e-05,
+      "loss": 3.846,
+      "step": 2193920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4070604436821386e-05,
+      "loss": 3.8671,
+      "step": 2194432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4062218489310866e-05,
+      "loss": 3.8548,
+      "step": 2194944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4053832541800346e-05,
+      "loss": 3.8618,
+      "step": 2195456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4045446594289827e-05,
+      "loss": 3.8634,
+      "step": 2195968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4037060646779304e-05,
+      "loss": 3.8663,
+      "step": 2196480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4028691078072518e-05,
+      "loss": 3.861,
+      "step": 2196992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4020305130561998e-05,
+      "loss": 3.86,
+      "step": 2197504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4011919183051478e-05,
+      "loss": 3.8684,
+      "step": 2198016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4003533235540956e-05,
+      "loss": 3.8648,
+      "step": 2198528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3995147288030436e-05,
+      "loss": 3.8612,
+      "step": 2199040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3986761340519916e-05,
+      "loss": 3.8635,
+      "step": 2199552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3978375393009396e-05,
+      "loss": 3.8593,
+      "step": 2200064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3969989445498876e-05,
+      "loss": 3.8695,
+      "step": 2200576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3961619876792087e-05,
+      "loss": 3.8472,
+      "step": 2201088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3953233929281567e-05,
+      "loss": 3.8544,
+      "step": 2201600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3944847981771048e-05,
+      "loss": 3.8508,
+      "step": 2202112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3936478413064257e-05,
+      "loss": 3.8571,
+      "step": 2202624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3928092465553737e-05,
+      "loss": 3.8669,
+      "step": 2203136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3919706518043219e-05,
+      "loss": 3.8546,
+      "step": 2203648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3911320570532699e-05,
+      "loss": 3.8716,
+      "step": 2204160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3902934623022179e-05,
+      "loss": 3.8574,
+      "step": 2204672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.389456505431539e-05,
+      "loss": 3.8681,
+      "step": 2205184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.388617910680487e-05,
+      "loss": 3.8658,
+      "step": 2205696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.387779315929435e-05,
+      "loss": 3.8691,
+      "step": 2206208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.386940721178383e-05,
+      "loss": 3.8645,
+      "step": 2206720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.386103764307704e-05,
+      "loss": 3.8642,
+      "step": 2207232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.385265169556652e-05,
+      "loss": 3.8629,
+      "step": 2207744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3844265748056002e-05,
+      "loss": 3.872,
+      "step": 2208256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3835879800545482e-05,
+      "loss": 3.8574,
+      "step": 2208768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3827493853034959e-05,
+      "loss": 3.8607,
+      "step": 2209280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.381910790552444e-05,
+      "loss": 3.8693,
+      "step": 2209792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.381072195801392e-05,
+      "loss": 3.861,
+      "step": 2210304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.38023360105034e-05,
+      "loss": 3.8567,
+      "step": 2210816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3793966441796611e-05,
+      "loss": 3.871,
+      "step": 2211328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.378558049428609e-05,
+      "loss": 3.8555,
+      "step": 2211840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3777210925579303e-05,
+      "loss": 3.8599,
+      "step": 2212352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.3768824978068782e-05,
+      "loss": 3.8567,
+      "step": 2212864
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.980386734008789,
+      "eval_runtime": 296.7656,
+      "eval_samples_per_second": 1285.833,
+      "eval_steps_per_second": 40.183,
+      "step": 2213280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3760439030558262e-05,
+      "loss": 3.8601,
+      "step": 2213376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3752053083047741e-05,
+      "loss": 3.8516,
+      "step": 2213888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3743667135537221e-05,
+      "loss": 3.8635,
+      "step": 2214400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3735281188026703e-05,
+      "loss": 3.8723,
+      "step": 2214912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3726895240516183e-05,
+      "loss": 3.8675,
+      "step": 2215424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3718509293005663e-05,
+      "loss": 3.8644,
+      "step": 2215936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3710123345495141e-05,
+      "loss": 3.8556,
+      "step": 2216448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3701737397984621e-05,
+      "loss": 3.8637,
+      "step": 2216960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3693351450474101e-05,
+      "loss": 3.8545,
+      "step": 2217472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3684965502963581e-05,
+      "loss": 3.8644,
+      "step": 2217984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3676579555453061e-05,
+      "loss": 3.8642,
+      "step": 2218496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3668193607942543e-05,
+      "loss": 3.8655,
+      "step": 2219008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3659824039235752e-05,
+      "loss": 3.8721,
+      "step": 2219520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3651438091725233e-05,
+      "loss": 3.8495,
+      "step": 2220032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3643052144214713e-05,
+      "loss": 3.8491,
+      "step": 2220544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3634666196704193e-05,
+      "loss": 3.8484,
+      "step": 2221056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3626280249193672e-05,
+      "loss": 3.8742,
+      "step": 2221568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3617894301683152e-05,
+      "loss": 3.8602,
+      "step": 2222080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3609508354172632e-05,
+      "loss": 3.8584,
+      "step": 2222592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3601122406662112e-05,
+      "loss": 3.8703,
+      "step": 2223104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3592736459151591e-05,
+      "loss": 3.8756,
+      "step": 2223616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3584350511641073e-05,
+      "loss": 3.863,
+      "step": 2224128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3575964564130553e-05,
+      "loss": 3.8618,
+      "step": 2224640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.356757861662003e-05,
+      "loss": 3.8626,
+      "step": 2225152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3559209047913244e-05,
+      "loss": 3.8717,
+      "step": 2225664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3550839479206453e-05,
+      "loss": 3.8601,
+      "step": 2226176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3542453531695935e-05,
+      "loss": 3.8532,
+      "step": 2226688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3534067584185415e-05,
+      "loss": 3.8716,
+      "step": 2227200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3525681636674894e-05,
+      "loss": 3.8451,
+      "step": 2227712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3517312067968105e-05,
+      "loss": 3.8547,
+      "step": 2228224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3508926120457585e-05,
+      "loss": 3.8548,
+      "step": 2228736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3500540172947065e-05,
+      "loss": 3.8559,
+      "step": 2229248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3492154225436545e-05,
+      "loss": 3.8676,
+      "step": 2229760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3483768277926027e-05,
+      "loss": 3.8737,
+      "step": 2230272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3475382330415503e-05,
+      "loss": 3.858,
+      "step": 2230784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3467012761708716e-05,
+      "loss": 3.858,
+      "step": 2231296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3458643193001927e-05,
+      "loss": 3.8702,
+      "step": 2231808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3450257245491407e-05,
+      "loss": 3.8445,
+      "step": 2232320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3441871297980888e-05,
+      "loss": 3.8582,
+      "step": 2232832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3433485350470368e-05,
+      "loss": 3.8566,
+      "step": 2233344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3425099402959848e-05,
+      "loss": 3.8536,
+      "step": 2233856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3416713455449326e-05,
+      "loss": 3.8551,
+      "step": 2234368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3408327507938806e-05,
+      "loss": 3.8693,
+      "step": 2234880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3399941560428286e-05,
+      "loss": 3.8541,
+      "step": 2235392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3391555612917766e-05,
+      "loss": 3.8626,
+      "step": 2235904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3383169665407246e-05,
+      "loss": 3.8603,
+      "step": 2236416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3374800096700457e-05,
+      "loss": 3.8593,
+      "step": 2236928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3366414149189937e-05,
+      "loss": 3.8605,
+      "step": 2237440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3358028201679419e-05,
+      "loss": 3.8347,
+      "step": 2237952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3349642254168899e-05,
+      "loss": 3.852,
+      "step": 2238464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3341256306658379e-05,
+      "loss": 3.869,
+      "step": 2238976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.333288673795159e-05,
+      "loss": 3.8642,
+      "step": 2239488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.332450079044107e-05,
+      "loss": 3.8492,
+      "step": 2240000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.331611484293055e-05,
+      "loss": 3.8501,
+      "step": 2240512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.330772889542003e-05,
+      "loss": 3.8548,
+      "step": 2241024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3299342947909507e-05,
+      "loss": 3.8457,
+      "step": 2241536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3290957000398987e-05,
+      "loss": 3.8774,
+      "step": 2242048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3282571052888467e-05,
+      "loss": 3.8424,
+      "step": 2242560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3274201484181682e-05,
+      "loss": 3.8652,
+      "step": 2243072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3265815536671158e-05,
+      "loss": 3.8642,
+      "step": 2243584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3257429589160638e-05,
+      "loss": 3.8421,
+      "step": 2244096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.324904364165012e-05,
+      "loss": 3.851,
+      "step": 2244608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.32406576941396e-05,
+      "loss": 3.8473,
+      "step": 2245120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.323227174662908e-05,
+      "loss": 3.8467,
+      "step": 2245632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.322388579911856e-05,
+      "loss": 3.8502,
+      "step": 2246144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.321551623041177e-05,
+      "loss": 3.8686,
+      "step": 2246656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3207146661704981e-05,
+      "loss": 3.858,
+      "step": 2247168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3198760714194461e-05,
+      "loss": 3.8557,
+      "step": 2247680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3190374766683941e-05,
+      "loss": 3.8431,
+      "step": 2248192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3181988819173421e-05,
+      "loss": 3.8529,
+      "step": 2248704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3173602871662901e-05,
+      "loss": 3.86,
+      "step": 2249216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3165216924152383e-05,
+      "loss": 3.8644,
+      "step": 2249728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3156830976641863e-05,
+      "loss": 3.8545,
+      "step": 2250240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.314844502913134e-05,
+      "loss": 3.8617,
+      "step": 2250752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.314005908162082e-05,
+      "loss": 3.873,
+      "step": 2251264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3131689512914033e-05,
+      "loss": 3.8543,
+      "step": 2251776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3123303565403512e-05,
+      "loss": 3.856,
+      "step": 2252288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3114917617892992e-05,
+      "loss": 3.8562,
+      "step": 2252800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3106531670382471e-05,
+      "loss": 3.8584,
+      "step": 2253312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3098145722871951e-05,
+      "loss": 3.8601,
+      "step": 2253824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3089759775361431e-05,
+      "loss": 3.8618,
+      "step": 2254336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3081373827850913e-05,
+      "loss": 3.8664,
+      "step": 2254848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3072987880340393e-05,
+      "loss": 3.8729,
+      "step": 2255360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3064618311633604e-05,
+      "loss": 3.8661,
+      "step": 2255872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3056232364123084e-05,
+      "loss": 3.8447,
+      "step": 2256384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3047846416612564e-05,
+      "loss": 3.8509,
+      "step": 2256896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3039460469102044e-05,
+      "loss": 3.8656,
+      "step": 2257408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3031074521591522e-05,
+      "loss": 3.8648,
+      "step": 2257920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3022688574081002e-05,
+      "loss": 3.8587,
+      "step": 2258432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3014319005374214e-05,
+      "loss": 3.8591,
+      "step": 2258944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3005933057863693e-05,
+      "loss": 3.8662,
+      "step": 2259456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2997547110353173e-05,
+      "loss": 3.8522,
+      "step": 2259968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2989161162842653e-05,
+      "loss": 3.8512,
+      "step": 2260480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2980791594135867e-05,
+      "loss": 3.8656,
+      "step": 2260992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972405646625343e-05,
+      "loss": 3.8552,
+      "step": 2261504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2964019699114823e-05,
+      "loss": 3.8721,
+      "step": 2262016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2955633751604305e-05,
+      "loss": 3.8487,
+      "step": 2262528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2947247804093785e-05,
+      "loss": 3.8506,
+      "step": 2263040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2938861856583265e-05,
+      "loss": 3.8602,
+      "step": 2263552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2930475909072745e-05,
+      "loss": 3.8632,
+      "step": 2264064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2922089961562225e-05,
+      "loss": 3.8665,
+      "step": 2264576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2913720392855435e-05,
+      "loss": 3.8519,
+      "step": 2265088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2905334445344915e-05,
+      "loss": 3.8558,
+      "step": 2265600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2896948497834397e-05,
+      "loss": 3.8445,
+      "step": 2266112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2888562550323874e-05,
+      "loss": 3.8624,
+      "step": 2266624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2880192981617086e-05,
+      "loss": 3.8487,
+      "step": 2267136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2871807034106568e-05,
+      "loss": 3.8513,
+      "step": 2267648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2863421086596048e-05,
+      "loss": 3.8622,
+      "step": 2268160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2855035139085524e-05,
+      "loss": 3.8539,
+      "step": 2268672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2846665570378739e-05,
+      "loss": 3.8515,
+      "step": 2269184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2838296001671948e-05,
+      "loss": 3.8599,
+      "step": 2269696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.282991005416143e-05,
+      "loss": 3.8438,
+      "step": 2270208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.282152410665091e-05,
+      "loss": 3.8634,
+      "step": 2270720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.281313815914039e-05,
+      "loss": 3.8517,
+      "step": 2271232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2804752211629869e-05,
+      "loss": 3.8557,
+      "step": 2271744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2796366264119347e-05,
+      "loss": 3.8617,
+      "step": 2272256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2787980316608827e-05,
+      "loss": 3.861,
+      "step": 2272768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.277961074790204e-05,
+      "loss": 3.8566,
+      "step": 2273280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2771224800391521e-05,
+      "loss": 3.8596,
+      "step": 2273792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2762838852880998e-05,
+      "loss": 3.8691,
+      "step": 2274304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2754452905370478e-05,
+      "loss": 3.856,
+      "step": 2274816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.274606695785996e-05,
+      "loss": 3.8562,
+      "step": 2275328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2737697389153169e-05,
+      "loss": 3.8604,
+      "step": 2275840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.272931144164265e-05,
+      "loss": 3.8597,
+      "step": 2276352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.272092549413213e-05,
+      "loss": 3.8632,
+      "step": 2276864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.271253954662161e-05,
+      "loss": 3.8477,
+      "step": 2277376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.270415359911109e-05,
+      "loss": 3.8492,
+      "step": 2277888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2695784030404301e-05,
+      "loss": 3.8501,
+      "step": 2278400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2687398082893781e-05,
+      "loss": 3.8504,
+      "step": 2278912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2679012135383261e-05,
+      "loss": 3.859,
+      "step": 2279424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2670626187872741e-05,
+      "loss": 3.8587,
+      "step": 2279936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2662240240362223e-05,
+      "loss": 3.8634,
+      "step": 2280448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2653854292851703e-05,
+      "loss": 3.8548,
+      "step": 2280960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2645468345341179e-05,
+      "loss": 3.8644,
+      "step": 2281472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2637098776634393e-05,
+      "loss": 3.8595,
+      "step": 2281984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2628712829123873e-05,
+      "loss": 3.8705,
+      "step": 2282496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2620326881613351e-05,
+      "loss": 3.8565,
+      "step": 2283008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2611940934102831e-05,
+      "loss": 3.8615,
+      "step": 2283520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2603554986592311e-05,
+      "loss": 3.8588,
+      "step": 2284032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2595169039081791e-05,
+      "loss": 3.8679,
+      "step": 2284544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2586783091571271e-05,
+      "loss": 3.8539,
+      "step": 2285056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2578397144060753e-05,
+      "loss": 3.8541,
+      "step": 2285568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2570027575353962e-05,
+      "loss": 3.8655,
+      "step": 2286080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2561641627843444e-05,
+      "loss": 3.8574,
+      "step": 2286592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2553272059136653e-05,
+      "loss": 3.8541,
+      "step": 2287104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2544886111626133e-05,
+      "loss": 3.8667,
+      "step": 2287616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2536500164115614e-05,
+      "loss": 3.8473,
+      "step": 2288128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2528114216605094e-05,
+      "loss": 3.8605,
+      "step": 2288640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2519728269094574e-05,
+      "loss": 3.85,
+      "step": 2289152
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9793496131896973,
+      "eval_runtime": 293.9428,
+      "eval_samples_per_second": 1298.181,
+      "eval_steps_per_second": 40.569,
+      "step": 2289600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2511342321584054e-05,
+      "loss": 3.8535,
+      "step": 2289664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2502956374073533e-05,
+      "loss": 3.8474,
+      "step": 2290176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2494570426563014e-05,
+      "loss": 3.8568,
+      "step": 2290688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2486200857856225e-05,
+      "loss": 3.8684,
+      "step": 2291200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2477814910345705e-05,
+      "loss": 3.8647,
+      "step": 2291712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2469428962835185e-05,
+      "loss": 3.8595,
+      "step": 2292224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2461043015324665e-05,
+      "loss": 3.848,
+      "step": 2292736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2452673446617876e-05,
+      "loss": 3.8597,
+      "step": 2293248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2444287499107356e-05,
+      "loss": 3.8561,
+      "step": 2293760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2435901551596836e-05,
+      "loss": 3.8578,
+      "step": 2294272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2427515604086315e-05,
+      "loss": 3.8605,
+      "step": 2294784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2419146035379526e-05,
+      "loss": 3.8635,
+      "step": 2295296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2410776466672739e-05,
+      "loss": 3.8679,
+      "step": 2295808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2402390519162219e-05,
+      "loss": 3.8447,
+      "step": 2296320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2394004571651697e-05,
+      "loss": 3.8427,
+      "step": 2296832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2385618624141179e-05,
+      "loss": 3.8459,
+      "step": 2297344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2377232676630659e-05,
+      "loss": 3.867,
+      "step": 2297856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2368846729120139e-05,
+      "loss": 3.8592,
+      "step": 2298368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2360460781609617e-05,
+      "loss": 3.8538,
+      "step": 2298880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2352074834099098e-05,
+      "loss": 3.8647,
+      "step": 2299392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2343688886588578e-05,
+      "loss": 3.8748,
+      "step": 2299904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2335302939078057e-05,
+      "loss": 3.8581,
+      "step": 2300416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2326916991567537e-05,
+      "loss": 3.8598,
+      "step": 2300928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2318531044057018e-05,
+      "loss": 3.8598,
+      "step": 2301440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2310161475350229e-05,
+      "loss": 3.8696,
+      "step": 2301952
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2301775527839707e-05,
+      "loss": 3.8523,
+      "step": 2302464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.229340595913292e-05,
+      "loss": 3.8505,
+      "step": 2302976
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.22850200116224e-05,
+      "loss": 3.8648,
+      "step": 2303488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.227663406411188e-05,
+      "loss": 3.8473,
+      "step": 2304000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.226824811660136e-05,
+      "loss": 3.8504,
+      "step": 2304512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.225986216909084e-05,
+      "loss": 3.8481,
+      "step": 2305024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2251508979187783e-05,
+      "loss": 3.8558,
+      "step": 2305536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2243123031677261e-05,
+      "loss": 3.8603,
+      "step": 2306048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2234737084166741e-05,
+      "loss": 3.8673,
+      "step": 2306560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2226351136656223e-05,
+      "loss": 3.8569,
+      "step": 2307072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2217981567949434e-05,
+      "loss": 3.8558,
+      "step": 2307584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2209595620438914e-05,
+      "loss": 3.8649,
+      "step": 2308096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2201209672928394e-05,
+      "loss": 3.8426,
+      "step": 2308608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2192823725417873e-05,
+      "loss": 3.8522,
+      "step": 2309120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2184437777907352e-05,
+      "loss": 3.8533,
+      "step": 2309632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2176051830396833e-05,
+      "loss": 3.853,
+      "step": 2310144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2167665882886313e-05,
+      "loss": 3.847,
+      "step": 2310656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2159279935375792e-05,
+      "loss": 3.8662,
+      "step": 2311168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2150893987865272e-05,
+      "loss": 3.8507,
+      "step": 2311680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2142508040354753e-05,
+      "loss": 3.8582,
+      "step": 2312192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2134122092844233e-05,
+      "loss": 3.8573,
+      "step": 2312704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2125736145333711e-05,
+      "loss": 3.8552,
+      "step": 2313216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2117350197823191e-05,
+      "loss": 3.8588,
+      "step": 2313728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2108964250312673e-05,
+      "loss": 3.8352,
+      "step": 2314240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2100578302802151e-05,
+      "loss": 3.8506,
+      "step": 2314752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2092192355291631e-05,
+      "loss": 3.8648,
+      "step": 2315264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2083822786584844e-05,
+      "loss": 3.8557,
+      "step": 2315776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2075436839074324e-05,
+      "loss": 3.847,
+      "step": 2316288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2067050891563802e-05,
+      "loss": 3.8451,
+      "step": 2316800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2058664944053284e-05,
+      "loss": 3.8553,
+      "step": 2317312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2050278996542764e-05,
+      "loss": 3.8406,
+      "step": 2317824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2041909427835973e-05,
+      "loss": 3.8702,
+      "step": 2318336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2033523480325454e-05,
+      "loss": 3.8452,
+      "step": 2318848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2025137532814934e-05,
+      "loss": 3.8552,
+      "step": 2319360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2016751585304414e-05,
+      "loss": 3.8607,
+      "step": 2319872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.2008382016597625e-05,
+      "loss": 3.8408,
+      "step": 2320384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1999996069087105e-05,
+      "loss": 3.8451,
+      "step": 2320896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1991610121576585e-05,
+      "loss": 3.848,
+      "step": 2321408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1983224174066065e-05,
+      "loss": 3.8381,
+      "step": 2321920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1974838226555545e-05,
+      "loss": 3.8437,
+      "step": 2322432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1966452279045025e-05,
+      "loss": 3.8668,
+      "step": 2322944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1958066331534505e-05,
+      "loss": 3.8548,
+      "step": 2323456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1949680384023985e-05,
+      "loss": 3.8509,
+      "step": 2323968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1941310815317195e-05,
+      "loss": 3.84,
+      "step": 2324480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1932924867806675e-05,
+      "loss": 3.8467,
+      "step": 2324992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1924538920296155e-05,
+      "loss": 3.8556,
+      "step": 2325504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1916169351589366e-05,
+      "loss": 3.8608,
+      "step": 2326016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1907783404078846e-05,
+      "loss": 3.8509,
+      "step": 2326528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1899397456568328e-05,
+      "loss": 3.8588,
+      "step": 2327040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1891011509057806e-05,
+      "loss": 3.8649,
+      "step": 2327552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1882625561547286e-05,
+      "loss": 3.8543,
+      "step": 2328064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1874255992840499e-05,
+      "loss": 3.8481,
+      "step": 2328576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1865870045329978e-05,
+      "loss": 3.8557,
+      "step": 2329088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1857484097819457e-05,
+      "loss": 3.8535,
+      "step": 2329600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1849098150308938e-05,
+      "loss": 3.8563,
+      "step": 2330112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.184072858160215e-05,
+      "loss": 3.8579,
+      "step": 2330624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1832342634091627e-05,
+      "loss": 3.8608,
+      "step": 2331136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1823956686581109e-05,
+      "loss": 3.8705,
+      "step": 2331648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.1815570739070589e-05,
+      "loss": 3.8593,
+      "step": 2332160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.18072011703638e-05,
+      "loss": 3.8425,
+      "step": 2332672
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.179881522285328e-05,
+      "loss": 3.8499,
+      "step": 2333184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.179042927534276e-05,
+      "loss": 3.864,
+      "step": 2333696
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.178204332783224e-05,
+      "loss": 3.8587,
+      "step": 2334208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.177365738032172e-05,
+      "loss": 3.8554,
+      "step": 2334720
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.176528781161493e-05,
+      "loss": 3.8556,
+      "step": 2335232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.175690186410441e-05,
+      "loss": 3.8601,
+      "step": 2335744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1748515916593892e-05,
+      "loss": 3.8506,
+      "step": 2336256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.174012996908337e-05,
+      "loss": 3.8527,
+      "step": 2336768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.173174402157285e-05,
+      "loss": 3.8523,
+      "step": 2337280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1723374452866063e-05,
+      "loss": 3.8552,
+      "step": 2337792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1714988505355543e-05,
+      "loss": 3.8616,
+      "step": 2338304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1706618936648753e-05,
+      "loss": 3.8556,
+      "step": 2338816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1698232989138233e-05,
+      "loss": 3.8407,
+      "step": 2339328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1689847041627713e-05,
+      "loss": 3.8577,
+      "step": 2339840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1681461094117192e-05,
+      "loss": 3.8587,
+      "step": 2340352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1673075146606673e-05,
+      "loss": 3.8598,
+      "step": 2340864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1664689199096153e-05,
+      "loss": 3.8508,
+      "step": 2341376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1656303251585632e-05,
+      "loss": 3.8494,
+      "step": 2341888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1647917304075111e-05,
+      "loss": 3.8439,
+      "step": 2342400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1639547735368324e-05,
+      "loss": 3.8558,
+      "step": 2342912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1631161787857804e-05,
+      "loss": 3.8441,
+      "step": 2343424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1622775840347284e-05,
+      "loss": 3.8447,
+      "step": 2343936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1614389892836764e-05,
+      "loss": 3.8583,
+      "step": 2344448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1606020324129975e-05,
+      "loss": 3.8523,
+      "step": 2344960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1597634376619455e-05,
+      "loss": 3.8477,
+      "step": 2345472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1589248429108935e-05,
+      "loss": 3.8534,
+      "step": 2345984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1580862481598414e-05,
+      "loss": 3.842,
+      "step": 2346496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1572476534087894e-05,
+      "loss": 3.8566,
+      "step": 2347008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1564090586577374e-05,
+      "loss": 3.8487,
+      "step": 2347520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1555704639066854e-05,
+      "loss": 3.8541,
+      "step": 2348032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1547335070360065e-05,
+      "loss": 3.8569,
+      "step": 2348544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1538949122849547e-05,
+      "loss": 3.8584,
+      "step": 2349056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1530579554142756e-05,
+      "loss": 3.8536,
+      "step": 2349568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1522193606632236e-05,
+      "loss": 3.8545,
+      "step": 2350080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1513807659121717e-05,
+      "loss": 3.8614,
+      "step": 2350592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1505421711611196e-05,
+      "loss": 3.8538,
+      "step": 2351104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1497035764100676e-05,
+      "loss": 3.8493,
+      "step": 2351616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1488649816590157e-05,
+      "loss": 3.8614,
+      "step": 2352128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1480263869079637e-05,
+      "loss": 3.854,
+      "step": 2352640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1471877921569116e-05,
+      "loss": 3.8601,
+      "step": 2353152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1463491974058596e-05,
+      "loss": 3.8443,
+      "step": 2353664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1455106026548077e-05,
+      "loss": 3.8445,
+      "step": 2354176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1446720079037555e-05,
+      "loss": 3.8477,
+      "step": 2354688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1438334131527035e-05,
+      "loss": 3.8515,
+      "step": 2355200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1429964562820248e-05,
+      "loss": 3.8475,
+      "step": 2355712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1421578615309728e-05,
+      "loss": 3.858,
+      "step": 2356224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1413209046602939e-05,
+      "loss": 3.8597,
+      "step": 2356736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1404823099092419e-05,
+      "loss": 3.8496,
+      "step": 2357248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1396437151581899e-05,
+      "loss": 3.8583,
+      "step": 2357760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1388051204071377e-05,
+      "loss": 3.8604,
+      "step": 2358272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1379665256560858e-05,
+      "loss": 3.8659,
+      "step": 2358784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1371279309050338e-05,
+      "loss": 3.8497,
+      "step": 2359296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.136290974034355e-05,
+      "loss": 3.8579,
+      "step": 2359808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.135452379283303e-05,
+      "loss": 3.8604,
+      "step": 2360320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1346137845322509e-05,
+      "loss": 3.8598,
+      "step": 2360832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1337751897811989e-05,
+      "loss": 3.8503,
+      "step": 2361344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1329365950301469e-05,
+      "loss": 3.851,
+      "step": 2361856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1320980002790949e-05,
+      "loss": 3.8612,
+      "step": 2362368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.1312594055280429e-05,
+      "loss": 3.8513,
+      "step": 2362880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.130422448657364e-05,
+      "loss": 3.8498,
+      "step": 2363392
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.129583853906312e-05,
+      "loss": 3.8643,
+      "step": 2363904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.12874525915526e-05,
+      "loss": 3.8474,
+      "step": 2364416
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.127906664404208e-05,
+      "loss": 3.8564,
+      "step": 2364928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.127068069653156e-05,
+      "loss": 3.8483,
+      "step": 2365440
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9779000282287598,
+      "eval_runtime": 294.579,
+      "eval_samples_per_second": 1295.377,
+      "eval_steps_per_second": 40.482,
+      "step": 2365920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.126231112782477e-05,
+      "loss": 3.8205,
+      "step": 2365952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.125392518031425e-05,
+      "loss": 3.8442,
+      "step": 2366464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1245539232803732e-05,
+      "loss": 3.8519,
+      "step": 2366976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.123715328529321e-05,
+      "loss": 3.8648,
+      "step": 2367488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.122876733778269e-05,
+      "loss": 3.8576,
+      "step": 2368000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.122038139027217e-05,
+      "loss": 3.8599,
+      "step": 2368512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.121199544276165e-05,
+      "loss": 3.8474,
+      "step": 2369024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.120360949525113e-05,
+      "loss": 3.857,
+      "step": 2369536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.119522354774061e-05,
+      "loss": 3.8491,
+      "step": 2370048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.118683760023009e-05,
+      "loss": 3.8565,
+      "step": 2370560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.117845165271957e-05,
+      "loss": 3.8535,
+      "step": 2371072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.117006570520905e-05,
+      "loss": 3.8614,
+      "step": 2371584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1161696136502262e-05,
+      "loss": 3.8695,
+      "step": 2372096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.115331018899174e-05,
+      "loss": 3.8437,
+      "step": 2372608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.114492424148122e-05,
+      "loss": 3.8379,
+      "step": 2373120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.11365382939707e-05,
+      "loss": 3.8382,
+      "step": 2373632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1128152346460182e-05,
+      "loss": 3.8579,
+      "step": 2374144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111976639894966e-05,
+      "loss": 3.8593,
+      "step": 2374656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.111138045143914e-05,
+      "loss": 3.8524,
+      "step": 2375168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.110299450392862e-05,
+      "loss": 3.8553,
+      "step": 2375680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.10946085564181e-05,
+      "loss": 3.8785,
+      "step": 2376192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.108622260890758e-05,
+      "loss": 3.8518,
+      "step": 2376704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.107783666139706e-05,
+      "loss": 3.8554,
+      "step": 2377216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.106945071388654e-05,
+      "loss": 3.8583,
+      "step": 2377728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.106106476637602e-05,
+      "loss": 3.867,
+      "step": 2378240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1052695197669231e-05,
+      "loss": 3.8451,
+      "step": 2378752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1044309250158713e-05,
+      "loss": 3.8502,
+      "step": 2379264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.103592330264819e-05,
+      "loss": 3.8571,
+      "step": 2379776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.102753735513767e-05,
+      "loss": 3.8475,
+      "step": 2380288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1019184165234614e-05,
+      "loss": 3.8435,
+      "step": 2380800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1010798217724094e-05,
+      "loss": 3.8398,
+      "step": 2381312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1002412270213574e-05,
+      "loss": 3.8513,
+      "step": 2381824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0994026322703054e-05,
+      "loss": 3.8605,
+      "step": 2382336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0985640375192534e-05,
+      "loss": 3.8611,
+      "step": 2382848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0977270806485745e-05,
+      "loss": 3.8567,
+      "step": 2383360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0968901237778955e-05,
+      "loss": 3.8492,
+      "step": 2383872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0960515290268435e-05,
+      "loss": 3.8604,
+      "step": 2384384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0952129342757917e-05,
+      "loss": 3.8401,
+      "step": 2384896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0943743395247395e-05,
+      "loss": 3.8497,
+      "step": 2385408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0935357447736875e-05,
+      "loss": 3.8487,
+      "step": 2385920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0926971500226355e-05,
+      "loss": 3.846,
+      "step": 2386432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0918585552715835e-05,
+      "loss": 3.8434,
+      "step": 2386944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0910199605205315e-05,
+      "loss": 3.8637,
+      "step": 2387456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0901813657694795e-05,
+      "loss": 3.8435,
+      "step": 2387968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0893427710184275e-05,
+      "loss": 3.8546,
+      "step": 2388480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0885058141477486e-05,
+      "loss": 3.8554,
+      "step": 2388992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0876688572770698e-05,
+      "loss": 3.8501,
+      "step": 2389504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0868302625260178e-05,
+      "loss": 3.8542,
+      "step": 2390016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0859916677749658e-05,
+      "loss": 3.8333,
+      "step": 2390528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0851530730239137e-05,
+      "loss": 3.843,
+      "step": 2391040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0843144782728618e-05,
+      "loss": 3.8618,
+      "step": 2391552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0834758835218098e-05,
+      "loss": 3.8542,
+      "step": 2392064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0826372887707576e-05,
+      "loss": 3.842,
+      "step": 2392576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0818003319000789e-05,
+      "loss": 3.8433,
+      "step": 2393088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0809617371490269e-05,
+      "loss": 3.8482,
+      "step": 2393600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0801231423979749e-05,
+      "loss": 3.8379,
+      "step": 2394112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0792845476469229e-05,
+      "loss": 3.8632,
+      "step": 2394624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0784459528958709e-05,
+      "loss": 3.8433,
+      "step": 2395136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0776073581448189e-05,
+      "loss": 3.851,
+      "step": 2395648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0767687633937667e-05,
+      "loss": 3.8615,
+      "step": 2396160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0759301686427149e-05,
+      "loss": 3.8348,
+      "step": 2396672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.075093211772036e-05,
+      "loss": 3.8384,
+      "step": 2397184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.074254617020984e-05,
+      "loss": 3.8462,
+      "step": 2397696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.073416022269932e-05,
+      "loss": 3.8347,
+      "step": 2398208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.07257742751888e-05,
+      "loss": 3.8421,
+      "step": 2398720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.071738832767828e-05,
+      "loss": 3.8581,
+      "step": 2399232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0709002380167759e-05,
+      "loss": 3.8561,
+      "step": 2399744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0700616432657239e-05,
+      "loss": 3.849,
+      "step": 2400256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0692230485146719e-05,
+      "loss": 3.8338,
+      "step": 2400768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.068386091643993e-05,
+      "loss": 3.8453,
+      "step": 2401280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.067547496892941e-05,
+      "loss": 3.8496,
+      "step": 2401792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.066710540022262e-05,
+      "loss": 3.8548,
+      "step": 2402304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0658719452712102e-05,
+      "loss": 3.8512,
+      "step": 2402816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.065033350520158e-05,
+      "loss": 3.8515,
+      "step": 2403328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.064194755769106e-05,
+      "loss": 3.8623,
+      "step": 2403840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.063356161018054e-05,
+      "loss": 3.8524,
+      "step": 2404352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0625192041473753e-05,
+      "loss": 3.8404,
+      "step": 2404864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0616806093963231e-05,
+      "loss": 3.8547,
+      "step": 2405376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0608420146452711e-05,
+      "loss": 3.8508,
+      "step": 2405888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0600034198942193e-05,
+      "loss": 3.8493,
+      "step": 2406400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0591648251431671e-05,
+      "loss": 3.8579,
+      "step": 2406912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0583262303921151e-05,
+      "loss": 3.8564,
+      "step": 2407424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0574876356410633e-05,
+      "loss": 3.8664,
+      "step": 2407936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0566490408900113e-05,
+      "loss": 3.8584,
+      "step": 2408448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0558137218997054e-05,
+      "loss": 3.8405,
+      "step": 2408960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0549751271486534e-05,
+      "loss": 3.8458,
+      "step": 2409472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0541365323976014e-05,
+      "loss": 3.8519,
+      "step": 2409984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0532979376465494e-05,
+      "loss": 3.8585,
+      "step": 2410496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0524593428954974e-05,
+      "loss": 3.8531,
+      "step": 2411008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0516223860248185e-05,
+      "loss": 3.85,
+      "step": 2411520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0507837912737665e-05,
+      "loss": 3.8574,
+      "step": 2412032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0499451965227145e-05,
+      "loss": 3.8467,
+      "step": 2412544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0491066017716625e-05,
+      "loss": 3.8488,
+      "step": 2413056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0482680070206105e-05,
+      "loss": 3.8468,
+      "step": 2413568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0474294122695585e-05,
+      "loss": 3.8557,
+      "step": 2414080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0465924553988795e-05,
+      "loss": 3.8602,
+      "step": 2414592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0457538606478275e-05,
+      "loss": 3.851,
+      "step": 2415104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0449152658967757e-05,
+      "loss": 3.839,
+      "step": 2415616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0440766711457235e-05,
+      "loss": 3.8503,
+      "step": 2416128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0432380763946715e-05,
+      "loss": 3.8549,
+      "step": 2416640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0423994816436195e-05,
+      "loss": 3.8593,
+      "step": 2417152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0415608868925675e-05,
+      "loss": 3.8459,
+      "step": 2417664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0407222921415155e-05,
+      "loss": 3.8474,
+      "step": 2418176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0398853352708368e-05,
+      "loss": 3.8361,
+      "step": 2418688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0390467405197848e-05,
+      "loss": 3.8537,
+      "step": 2419200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0382081457687326e-05,
+      "loss": 3.8396,
+      "step": 2419712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0373695510176806e-05,
+      "loss": 3.8445,
+      "step": 2420224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0365309562666287e-05,
+      "loss": 3.8542,
+      "step": 2420736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0356923615155766e-05,
+      "loss": 3.8469,
+      "step": 2421248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0348537667645246e-05,
+      "loss": 3.8463,
+      "step": 2421760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0340151720134726e-05,
+      "loss": 3.8449,
+      "step": 2422272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0331782151427938e-05,
+      "loss": 3.8424,
+      "step": 2422784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0323396203917416e-05,
+      "loss": 3.8534,
+      "step": 2423296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0315010256406896e-05,
+      "loss": 3.8442,
+      "step": 2423808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0306624308896378e-05,
+      "loss": 3.8483,
+      "step": 2424320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0298238361385856e-05,
+      "loss": 3.8487,
+      "step": 2424832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0289868792679069e-05,
+      "loss": 3.8552,
+      "step": 2425344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.028149922397228e-05,
+      "loss": 3.848,
+      "step": 2425856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027311327646176e-05,
+      "loss": 3.8524,
+      "step": 2426368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.026472732895124e-05,
+      "loss": 3.8573,
+      "step": 2426880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.025634138144072e-05,
+      "loss": 3.8535,
+      "step": 2427392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.02479554339302e-05,
+      "loss": 3.8377,
+      "step": 2427904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.023956948641968e-05,
+      "loss": 3.8588,
+      "step": 2428416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.023119991771289e-05,
+      "loss": 3.8496,
+      "step": 2428928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.022281397020237e-05,
+      "loss": 3.8592,
+      "step": 2429440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.021442802269185e-05,
+      "loss": 3.8398,
+      "step": 2429952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.020604207518133e-05,
+      "loss": 3.8407,
+      "step": 2430464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.019765612767081e-05,
+      "loss": 3.8422,
+      "step": 2430976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.018927018016029e-05,
+      "loss": 3.8482,
+      "step": 2431488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.018088423264977e-05,
+      "loss": 3.8407,
+      "step": 2432000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.017251466394298e-05,
+      "loss": 3.8556,
+      "step": 2432512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.016412871643246e-05,
+      "loss": 3.857,
+      "step": 2433024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0155742768921942e-05,
+      "loss": 3.8483,
+      "step": 2433536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.014735682141142e-05,
+      "loss": 3.853,
+      "step": 2434048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.01389708739009e-05,
+      "loss": 3.8538,
+      "step": 2434560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0130601305194113e-05,
+      "loss": 3.8591,
+      "step": 2435072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0122231736487324e-05,
+      "loss": 3.8539,
+      "step": 2435584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0113845788976804e-05,
+      "loss": 3.8529,
+      "step": 2436096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0105459841466284e-05,
+      "loss": 3.857,
+      "step": 2436608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0097073893955764e-05,
+      "loss": 3.8536,
+      "step": 2437120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0088687946445243e-05,
+      "loss": 3.8465,
+      "step": 2437632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0080301998934723e-05,
+      "loss": 3.8467,
+      "step": 2438144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0071916051424203e-05,
+      "loss": 3.8581,
+      "step": 2438656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0063530103913683e-05,
+      "loss": 3.8486,
+      "step": 2439168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0055144156403162e-05,
+      "loss": 3.8433,
+      "step": 2439680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0046758208892643e-05,
+      "loss": 3.8636,
+      "step": 2440192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0038388640185854e-05,
+      "loss": 3.8456,
+      "step": 2440704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0030002692675334e-05,
+      "loss": 3.849,
+      "step": 2441216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0021616745164814e-05,
+      "loss": 3.8454,
+      "step": 2441728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.0013230797654294e-05,
+      "loss": 3.8467,
+      "step": 2442240
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.976560592651367,
+      "eval_runtime": 294.5708,
+      "eval_samples_per_second": 1295.413,
+      "eval_steps_per_second": 40.483,
+      "step": 2442240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0004844850143774e-05,
+      "loss": 3.8377,
+      "step": 2442752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.996458902633254e-06,
+      "loss": 3.8481,
+      "step": 2443264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.988072955122734e-06,
+      "loss": 3.8607,
+      "step": 2443776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.979687007612214e-06,
+      "loss": 3.8551,
+      "step": 2444288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.971301060101692e-06,
+      "loss": 3.8569,
+      "step": 2444800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.962915112591174e-06,
+      "loss": 3.845,
+      "step": 2445312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.954529165080654e-06,
+      "loss": 3.8496,
+      "step": 2445824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.946143217570134e-06,
+      "loss": 3.8416,
+      "step": 2446336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.937757270059612e-06,
+      "loss": 3.8572,
+      "step": 2446848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.929371322549093e-06,
+      "loss": 3.85,
+      "step": 2447360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.920985375038573e-06,
+      "loss": 3.8573,
+      "step": 2447872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.912599427528052e-06,
+      "loss": 3.8573,
+      "step": 2448384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.904229858821264e-06,
+      "loss": 3.8464,
+      "step": 2448896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.895843911310744e-06,
+      "loss": 3.837,
+      "step": 2449408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.887457963800222e-06,
+      "loss": 3.839,
+      "step": 2449920
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.879072016289704e-06,
+      "loss": 3.8464,
+      "step": 2450432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.870686068779184e-06,
+      "loss": 3.8554,
+      "step": 2450944
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.862300121268664e-06,
+      "loss": 3.8457,
+      "step": 2451456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.853914173758142e-06,
+      "loss": 3.8541,
+      "step": 2451968
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.845528226247624e-06,
+      "loss": 3.8765,
+      "step": 2452480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.837142278737104e-06,
+      "loss": 3.8488,
+      "step": 2452992
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.828756331226582e-06,
+      "loss": 3.8497,
+      "step": 2453504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.820370383716062e-06,
+      "loss": 3.8529,
+      "step": 2454016
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.812000815009275e-06,
+      "loss": 3.8622,
+      "step": 2454528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.803614867498754e-06,
+      "loss": 3.8426,
+      "step": 2455040
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.795228919988233e-06,
+      "loss": 3.8428,
+      "step": 2455552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.786842972477714e-06,
+      "loss": 3.8576,
+      "step": 2456064
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.778473403770925e-06,
+      "loss": 3.8449,
+      "step": 2456576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.770103835064138e-06,
+      "loss": 3.837,
+      "step": 2457088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.761717887553616e-06,
+      "loss": 3.8388,
+      "step": 2457600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.753348318846828e-06,
+      "loss": 3.85,
+      "step": 2458112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.744962371336308e-06,
+      "loss": 3.8546,
+      "step": 2458624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.736576423825787e-06,
+      "loss": 3.8627,
+      "step": 2459136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.728190476315267e-06,
+      "loss": 3.8477,
+      "step": 2459648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.719804528804748e-06,
+      "loss": 3.8452,
+      "step": 2460160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.711434960097959e-06,
+      "loss": 3.8555,
+      "step": 2460672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.703049012587439e-06,
+      "loss": 3.8365,
+      "step": 2461184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.694663065076919e-06,
+      "loss": 3.8482,
+      "step": 2461696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.686277117566399e-06,
+      "loss": 3.8435,
+      "step": 2462208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.677891170055877e-06,
+      "loss": 3.8465,
+      "step": 2462720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.669505222545359e-06,
+      "loss": 3.8379,
+      "step": 2463232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.661119275034839e-06,
+      "loss": 3.8614,
+      "step": 2463744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.652733327524319e-06,
+      "loss": 3.8428,
+      "step": 2464256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.644347380013797e-06,
+      "loss": 3.8479,
+      "step": 2464768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.63597781130701e-06,
+      "loss": 3.8512,
+      "step": 2465280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.62759186379649e-06,
+      "loss": 3.8486,
+      "step": 2465792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.61920591628597e-06,
+      "loss": 3.852,
+      "step": 2466304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.61081996877545e-06,
+      "loss": 3.8346,
+      "step": 2466816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.60243402126493e-06,
+      "loss": 3.8357,
+      "step": 2467328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.59406445255814e-06,
+      "loss": 3.8568,
+      "step": 2467840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.58567850504762e-06,
+      "loss": 3.8513,
+      "step": 2468352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.5772925575371e-06,
+      "loss": 3.8389,
+      "step": 2468864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.56890661002658e-06,
+      "loss": 3.8421,
+      "step": 2469376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.56052066251606e-06,
+      "loss": 3.8468,
+      "step": 2469888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.55215109380927e-06,
+      "loss": 3.8376,
+      "step": 2470400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.54376514629875e-06,
+      "loss": 3.8525,
+      "step": 2470912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.535395577591963e-06,
+      "loss": 3.8412,
+      "step": 2471424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.527009630081441e-06,
+      "loss": 3.8463,
+      "step": 2471936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.518623682570921e-06,
+      "loss": 3.8526,
+      "step": 2472448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.510237735060403e-06,
+      "loss": 3.8411,
+      "step": 2472960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.501851787549883e-06,
+      "loss": 3.8317,
+      "step": 2473472
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.493465840039361e-06,
+      "loss": 3.8435,
+      "step": 2473984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.485079892528841e-06,
+      "loss": 3.8322,
+      "step": 2474496
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.476693945018323e-06,
+      "loss": 3.838,
+      "step": 2475008
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.468307997507801e-06,
+      "loss": 3.8523,
+      "step": 2475520
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.459922049997281e-06,
+      "loss": 3.8544,
+      "step": 2476032
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.451536102486763e-06,
+      "loss": 3.8436,
+      "step": 2476544
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.443166533779973e-06,
+      "loss": 3.8348,
+      "step": 2477056
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.434780586269452e-06,
+      "loss": 3.8355,
+      "step": 2477568
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.426394638758933e-06,
+      "loss": 3.8446,
+      "step": 2478080
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.418008691248413e-06,
+      "loss": 3.8499,
+      "step": 2478592
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.409639122541624e-06,
+      "loss": 3.8499,
+      "step": 2479104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.401253175031104e-06,
+      "loss": 3.849,
+      "step": 2479616
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.392867227520584e-06,
+      "loss": 3.8566,
+      "step": 2480128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.384481280010064e-06,
+      "loss": 3.8542,
+      "step": 2480640
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.376095332499544e-06,
+      "loss": 3.8343,
+      "step": 2481152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.367709384989024e-06,
+      "loss": 3.8484,
+      "step": 2481664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.359339816282235e-06,
+      "loss": 3.8514,
+      "step": 2482176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.350953868771715e-06,
+      "loss": 3.8406,
+      "step": 2482688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.342567921261195e-06,
+      "loss": 3.8559,
+      "step": 2483200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.334181973750675e-06,
+      "loss": 3.8544,
+      "step": 2483712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.325796026240153e-06,
+      "loss": 3.8613,
+      "step": 2484224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.317410078729634e-06,
+      "loss": 3.8555,
+      "step": 2484736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.309024131219114e-06,
+      "loss": 3.8363,
+      "step": 2485248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.300654562512325e-06,
+      "loss": 3.844,
+      "step": 2485760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.292268615001805e-06,
+      "loss": 3.8511,
+      "step": 2486272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.283882667491285e-06,
+      "loss": 3.8536,
+      "step": 2486784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.275496719980765e-06,
+      "loss": 3.8498,
+      "step": 2487296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.267127151273978e-06,
+      "loss": 3.8468,
+      "step": 2487808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.25877396137092e-06,
+      "loss": 3.8523,
+      "step": 2488320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.250388013860399e-06,
+      "loss": 3.8463,
+      "step": 2488832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.242002066349879e-06,
+      "loss": 3.8405,
+      "step": 2489344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.233616118839359e-06,
+      "loss": 3.8457,
+      "step": 2489856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.225230171328839e-06,
+      "loss": 3.8506,
+      "step": 2490368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.216844223818319e-06,
+      "loss": 3.8597,
+      "step": 2490880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.208458276307799e-06,
+      "loss": 3.842,
+      "step": 2491392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.200072328797279e-06,
+      "loss": 3.8393,
+      "step": 2491904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.191686381286759e-06,
+      "loss": 3.8439,
+      "step": 2492416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.183300433776239e-06,
+      "loss": 3.8508,
+      "step": 2492928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.174914486265717e-06,
+      "loss": 3.8569,
+      "step": 2493440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.166528538755199e-06,
+      "loss": 3.8419,
+      "step": 2493952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.158142591244679e-06,
+      "loss": 3.8504,
+      "step": 2494464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.149756643734159e-06,
+      "loss": 3.8306,
+      "step": 2494976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.141370696223637e-06,
+      "loss": 3.8497,
+      "step": 2495488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.132984748713119e-06,
+      "loss": 3.8396,
+      "step": 2496000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.12461518000633e-06,
+      "loss": 3.8391,
+      "step": 2496512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.116245611299542e-06,
+      "loss": 3.848,
+      "step": 2497024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.10785966378902e-06,
+      "loss": 3.8473,
+      "step": 2497536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.0994737162785e-06,
+      "loss": 3.8411,
+      "step": 2498048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.09108776876798e-06,
+      "loss": 3.8454,
+      "step": 2498560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.08270182125746e-06,
+      "loss": 3.8405,
+      "step": 2499072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.07433225255067e-06,
+      "loss": 3.8486,
+      "step": 2499584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.065946305040152e-06,
+      "loss": 3.8369,
+      "step": 2500096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.057560357529632e-06,
+      "loss": 3.8465,
+      "step": 2500608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.04917441001911e-06,
+      "loss": 3.8494,
+      "step": 2501120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.040804841312323e-06,
+      "loss": 3.8518,
+      "step": 2501632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.032435272605534e-06,
+      "loss": 3.8466,
+      "step": 2502144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.024049325095014e-06,
+      "loss": 3.848,
+      "step": 2502656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.015663377584494e-06,
+      "loss": 3.853,
+      "step": 2503168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.007277430073974e-06,
+      "loss": 3.8474,
+      "step": 2503680
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.998891482563454e-06,
+      "loss": 3.8357,
+      "step": 2504192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.990505535052934e-06,
+      "loss": 3.8573,
+      "step": 2504704
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.982119587542414e-06,
+      "loss": 3.8462,
+      "step": 2505216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.973733640031894e-06,
+      "loss": 3.8549,
+      "step": 2505728
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.965347692521372e-06,
+      "loss": 3.8427,
+      "step": 2506240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.956961745010853e-06,
+      "loss": 3.8378,
+      "step": 2506752
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.948592176304064e-06,
+      "loss": 3.8374,
+      "step": 2507264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.940206228793544e-06,
+      "loss": 3.8429,
+      "step": 2507776
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.931820281283024e-06,
+      "loss": 3.837,
+      "step": 2508288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.923434333772504e-06,
+      "loss": 3.854,
+      "step": 2508800
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.915048386261984e-06,
+      "loss": 3.8556,
+      "step": 2509312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.906678817555197e-06,
+      "loss": 3.8432,
+      "step": 2509824
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.898292870044675e-06,
+      "loss": 3.846,
+      "step": 2510336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.889906922534155e-06,
+      "loss": 3.8511,
+      "step": 2510848
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.881520975023635e-06,
+      "loss": 3.8551,
+      "step": 2511360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.873135027513115e-06,
+      "loss": 3.8541,
+      "step": 2511872
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.864765458806325e-06,
+      "loss": 3.854,
+      "step": 2512384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.856379511295807e-06,
+      "loss": 3.8483,
+      "step": 2512896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.847993563785287e-06,
+      "loss": 3.8479,
+      "step": 2513408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.839607616274765e-06,
+      "loss": 3.8456,
+      "step": 2513920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.831221668764245e-06,
+      "loss": 3.8442,
+      "step": 2514432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.822835721253727e-06,
+      "loss": 3.8573,
+      "step": 2514944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.814449773743205e-06,
+      "loss": 3.8474,
+      "step": 2515456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.806063826232685e-06,
+      "loss": 3.8369,
+      "step": 2515968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.797694257525898e-06,
+      "loss": 3.8626,
+      "step": 2516480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.789308310015378e-06,
+      "loss": 3.8441,
+      "step": 2516992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.780922362504856e-06,
+      "loss": 3.8449,
+      "step": 2517504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.772552793798068e-06,
+      "loss": 3.8422,
+      "step": 2518016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.764166846287548e-06,
+      "loss": 3.8417,
+      "step": 2518528
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.975619316101074,
+      "eval_runtime": 287.4351,
+      "eval_samples_per_second": 1327.573,
+      "eval_steps_per_second": 41.488,
+      "step": 2518560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.755780898777027e-06,
+      "loss": 3.8342,
+      "step": 2519040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.747394951266508e-06,
+      "loss": 3.8504,
+      "step": 2519552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.739009003755988e-06,
+      "loss": 3.8617,
+      "step": 2520064
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.730639435049199e-06,
+      "loss": 3.8446,
+      "step": 2520576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.722253487538679e-06,
+      "loss": 3.862,
+      "step": 2521088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.713867540028159e-06,
+      "loss": 3.8384,
+      "step": 2521600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.705481592517639e-06,
+      "loss": 3.8487,
+      "step": 2522112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.697095645007119e-06,
+      "loss": 3.8384,
+      "step": 2522624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.688709697496599e-06,
+      "loss": 3.8512,
+      "step": 2523136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.68034012878981e-06,
+      "loss": 3.8494,
+      "step": 2523648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.671954181279291e-06,
+      "loss": 3.8531,
+      "step": 2524160
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.66356823376877e-06,
+      "loss": 3.8543,
+      "step": 2524672
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.65519866506198e-06,
+      "loss": 3.8399,
+      "step": 2525184
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.646812717551462e-06,
+      "loss": 3.8351,
+      "step": 2525696
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.63842677004094e-06,
+      "loss": 3.8381,
+      "step": 2526208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.63004082253042e-06,
+      "loss": 3.8437,
+      "step": 2526720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.6216548750199e-06,
+      "loss": 3.8522,
+      "step": 2527232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.613268927509382e-06,
+      "loss": 3.8443,
+      "step": 2527744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.60488297999886e-06,
+      "loss": 3.8511,
+      "step": 2528256
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.59649703248834e-06,
+      "loss": 3.8731,
+      "step": 2528768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.58811108497782e-06,
+      "loss": 3.8474,
+      "step": 2529280
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57974151627103e-06,
+      "loss": 3.844,
+      "step": 2529792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.57135556876051e-06,
+      "loss": 3.8515,
+      "step": 2530304
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.562969621249992e-06,
+      "loss": 3.8587,
+      "step": 2530816
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.554583673739472e-06,
+      "loss": 3.839,
+      "step": 2531328
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.54619772622895e-06,
+      "loss": 3.845,
+      "step": 2531840
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.53781177871843e-06,
+      "loss": 3.849,
+      "step": 2532352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.529425831207912e-06,
+      "loss": 3.8475,
+      "step": 2532864
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.52103988369739e-06,
+      "loss": 3.8315,
+      "step": 2533376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.512670314990603e-06,
+      "loss": 3.8395,
+      "step": 2533888
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.504284367480083e-06,
+      "loss": 3.8411,
+      "step": 2534400
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.495898419969563e-06,
+      "loss": 3.8534,
+      "step": 2534912
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.487512472459041e-06,
+      "loss": 3.8628,
+      "step": 2535424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.479142903752253e-06,
+      "loss": 3.8435,
+      "step": 2535936
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.470789713849197e-06,
+      "loss": 3.8426,
+      "step": 2536448
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.462403766338677e-06,
+      "loss": 3.8551,
+      "step": 2536960
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.454017818828155e-06,
+      "loss": 3.8343,
+      "step": 2537472
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.445631871317635e-06,
+      "loss": 3.8442,
+      "step": 2537984
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.437245923807117e-06,
+      "loss": 3.838,
+      "step": 2538496
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.428859976296595e-06,
+      "loss": 3.8423,
+      "step": 2539008
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.420474028786075e-06,
+      "loss": 3.8393,
+      "step": 2539520
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.412088081275555e-06,
+      "loss": 3.8576,
+      "step": 2540032
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.403702133765036e-06,
+      "loss": 3.842,
+      "step": 2540544
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.395316186254515e-06,
+      "loss": 3.8451,
+      "step": 2541056
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.386946617547727e-06,
+      "loss": 3.8444,
+      "step": 2541568
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.378560670037207e-06,
+      "loss": 3.8498,
+      "step": 2542080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.370174722526685e-06,
+      "loss": 3.8448,
+      "step": 2542592
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.361788775016165e-06,
+      "loss": 3.8346,
+      "step": 2543104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.353402827505647e-06,
+      "loss": 3.8321,
+      "step": 2543616
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.345033258798858e-06,
+      "loss": 3.8543,
+      "step": 2544128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.336647311288338e-06,
+      "loss": 3.8453,
+      "step": 2544640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.328261363777818e-06,
+      "loss": 3.8357,
+      "step": 2545152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.319875416267298e-06,
+      "loss": 3.8391,
+      "step": 2545664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.311489468756776e-06,
+      "loss": 3.8441,
+      "step": 2546176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.303103521246258e-06,
+      "loss": 3.8339,
+      "step": 2546688
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.294717573735738e-06,
+      "loss": 3.8497,
+      "step": 2547200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.286331626225218e-06,
+      "loss": 3.8406,
+      "step": 2547712
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.277962057518428e-06,
+      "loss": 3.8425,
+      "step": 2548224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.269592488811639e-06,
+      "loss": 3.8519,
+      "step": 2548736
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.261206541301119e-06,
+      "loss": 3.8316,
+      "step": 2549248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.2528205937906e-06,
+      "loss": 3.8327,
+      "step": 2549760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.244434646280079e-06,
+      "loss": 3.8451,
+      "step": 2550272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.236048698769559e-06,
+      "loss": 3.824,
+      "step": 2550784
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.227662751259039e-06,
+      "loss": 3.8338,
+      "step": 2551296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.219276803748519e-06,
+      "loss": 3.8463,
+      "step": 2551808
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.210890856237999e-06,
+      "loss": 3.8522,
+      "step": 2552320
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.202521287531211e-06,
+      "loss": 3.8403,
+      "step": 2552832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.194151718824422e-06,
+      "loss": 3.8349,
+      "step": 2553344
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.1857657713139e-06,
+      "loss": 3.8301,
+      "step": 2553856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.177379823803382e-06,
+      "loss": 3.8447,
+      "step": 2554368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.168993876292862e-06,
+      "loss": 3.8451,
+      "step": 2554880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.16060792878234e-06,
+      "loss": 3.8478,
+      "step": 2555392
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.15222198127182e-06,
+      "loss": 3.8458,
+      "step": 2555904
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.143836033761302e-06,
+      "loss": 3.853,
+      "step": 2556416
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.135450086250782e-06,
+      "loss": 3.8526,
+      "step": 2556928
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.127096896347723e-06,
+      "loss": 3.8312,
+      "step": 2557440
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.118710948837203e-06,
+      "loss": 3.8461,
+      "step": 2557952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.110325001326683e-06,
+      "loss": 3.8491,
+      "step": 2558464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.101939053816163e-06,
+      "loss": 3.8319,
+      "step": 2558976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.093553106305643e-06,
+      "loss": 3.8578,
+      "step": 2559488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.085167158795123e-06,
+      "loss": 3.853,
+      "step": 2560000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.076781211284603e-06,
+      "loss": 3.8558,
+      "step": 2560512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.068411642577814e-06,
+      "loss": 3.8544,
+      "step": 2561024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.060025695067294e-06,
+      "loss": 3.8311,
+      "step": 2561536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.051639747556774e-06,
+      "loss": 3.846,
+      "step": 2562048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.043253800046254e-06,
+      "loss": 3.8444,
+      "step": 2562560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.034884231339465e-06,
+      "loss": 3.8509,
+      "step": 2563072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.026498283828946e-06,
+      "loss": 3.8487,
+      "step": 2563584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.018128715122157e-06,
+      "loss": 3.8422,
+      "step": 2564096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.009742767611635e-06,
+      "loss": 3.8518,
+      "step": 2564608
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.001356820101117e-06,
+      "loss": 3.848,
+      "step": 2565120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.992970872590597e-06,
+      "loss": 3.8333,
+      "step": 2565632
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.984601303883808e-06,
+      "loss": 3.8402,
+      "step": 2566144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.976215356373288e-06,
+      "loss": 3.8516,
+      "step": 2566656
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.967829408862768e-06,
+      "loss": 3.8571,
+      "step": 2567168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.959443461352248e-06,
+      "loss": 3.8436,
+      "step": 2567680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.951057513841727e-06,
+      "loss": 3.8304,
+      "step": 2568192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.942671566331207e-06,
+      "loss": 3.8424,
+      "step": 2568704
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.934285618820687e-06,
+      "loss": 3.8507,
+      "step": 2569216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.925899671310167e-06,
+      "loss": 3.8501,
+      "step": 2569728
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.917530102603378e-06,
+      "loss": 3.8434,
+      "step": 2570240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.909144155092858e-06,
+      "loss": 3.8443,
+      "step": 2570752
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.900758207582338e-06,
+      "loss": 3.8333,
+      "step": 2571264
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.892372260071818e-06,
+      "loss": 3.844,
+      "step": 2571776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.883986312561298e-06,
+      "loss": 3.8362,
+      "step": 2572288
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.875600365050778e-06,
+      "loss": 3.8383,
+      "step": 2572800
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.867214417540258e-06,
+      "loss": 3.8444,
+      "step": 2573312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.858828470029738e-06,
+      "loss": 3.8439,
+      "step": 2573824
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.850442522519218e-06,
+      "loss": 3.8374,
+      "step": 2574336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.842072953812429e-06,
+      "loss": 3.8452,
+      "step": 2574848
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.833687006301909e-06,
+      "loss": 3.8375,
+      "step": 2575360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.825301058791388e-06,
+      "loss": 3.8444,
+      "step": 2575872
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.816915111280868e-06,
+      "loss": 3.8369,
+      "step": 2576384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.808529163770348e-06,
+      "loss": 3.8388,
+      "step": 2576896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.80015959506356e-06,
+      "loss": 3.8512,
+      "step": 2577408
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.791773647553039e-06,
+      "loss": 3.8484,
+      "step": 2577920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.783404078846252e-06,
+      "loss": 3.8469,
+      "step": 2578432
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.775018131335732e-06,
+      "loss": 3.8425,
+      "step": 2578944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.76663218382521e-06,
+      "loss": 3.8491,
+      "step": 2579456
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.758246236314691e-06,
+      "loss": 3.8454,
+      "step": 2579968
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.749860288804171e-06,
+      "loss": 3.8336,
+      "step": 2580480
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.74147434129365e-06,
+      "loss": 3.8556,
+      "step": 2580992
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.733088393783131e-06,
+      "loss": 3.8428,
+      "step": 2581504
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.724702446272611e-06,
+      "loss": 3.851,
+      "step": 2582016
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.71631649876209e-06,
+      "loss": 3.8372,
+      "step": 2582528
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.707946930055302e-06,
+      "loss": 3.8356,
+      "step": 2583040
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.699560982544782e-06,
+      "loss": 3.8331,
+      "step": 2583552
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.691191413837993e-06,
+      "loss": 3.837,
+      "step": 2584064
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.682805466327473e-06,
+      "loss": 3.8342,
+      "step": 2584576
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.674419518816953e-06,
+      "loss": 3.8509,
+      "step": 2585088
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.666049950110164e-06,
+      "loss": 3.856,
+      "step": 2585600
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.657664002599645e-06,
+      "loss": 3.8468,
+      "step": 2586112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.649278055089123e-06,
+      "loss": 3.837,
+      "step": 2586624
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.640892107578603e-06,
+      "loss": 3.848,
+      "step": 2587136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.632506160068083e-06,
+      "loss": 3.8513,
+      "step": 2587648
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.624120212557563e-06,
+      "loss": 3.8546,
+      "step": 2588160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.615734265047043e-06,
+      "loss": 3.848,
+      "step": 2588672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.607348317536524e-06,
+      "loss": 3.8483,
+      "step": 2589184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.598978748829736e-06,
+      "loss": 3.839,
+      "step": 2589696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.590592801319214e-06,
+      "loss": 3.8472,
+      "step": 2590208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.582206853808695e-06,
+      "loss": 3.8414,
+      "step": 2590720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.573820906298175e-06,
+      "loss": 3.8494,
+      "step": 2591232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.565434958787654e-06,
+      "loss": 3.8476,
+      "step": 2591744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.557049011277134e-06,
+      "loss": 3.8396,
+      "step": 2592256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.5486630637666146e-06,
+      "loss": 3.8586,
+      "step": 2592768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.5402771162560945e-06,
+      "loss": 3.8426,
+      "step": 2593280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.531891168745574e-06,
+      "loss": 3.8397,
+      "step": 2593792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.5235052212350544e-06,
+      "loss": 3.8413,
+      "step": 2594304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 7.515119273724534e-06,
+      "loss": 3.8381,
+      "step": 2594816
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.9748899936676025,
+      "eval_runtime": 279.3772,
+      "eval_samples_per_second": 1365.863,
+      "eval_steps_per_second": 42.684,
+      "step": 2594880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.5067333262140135e-06,
+      "loss": 3.8292,
+      "step": 2595328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4983473787034935e-06,
+      "loss": 3.8473,
+      "step": 2595840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.489961431192974e-06,
+      "loss": 3.8563,
+      "step": 2596352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4815754836824525e-06,
+      "loss": 3.8427,
+      "step": 2596864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.473189536171933e-06,
+      "loss": 3.8582,
+      "step": 2597376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.464803588661413e-06,
+      "loss": 3.8375,
+      "step": 2597888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.456417641150892e-06,
+      "loss": 3.8472,
+      "step": 2598400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.448031693640372e-06,
+      "loss": 3.8318,
+      "step": 2598912
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.439645746129853e-06,
+      "loss": 3.851,
+      "step": 2599424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.431259798619333e-06,
+      "loss": 3.8451,
+      "step": 2599936
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.422873851108812e-06,
+      "loss": 3.8477,
+      "step": 2600448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.414487903598292e-06,
+      "loss": 3.8557,
+      "step": 2600960
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.406118334891504e-06,
+      "loss": 3.8371,
+      "step": 2601472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.397732387380983e-06,
+      "loss": 3.8337,
+      "step": 2601984
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.389346439870464e-06,
+      "loss": 3.8327,
+      "step": 2602496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.380960492359944e-06,
+      "loss": 3.8426,
+      "step": 2603008
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.3725745448494245e-06,
+      "loss": 3.8476,
+      "step": 2603520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.364204976142634e-06,
+      "loss": 3.8412,
+      "step": 2604032
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.355819028632114e-06,
+      "loss": 3.8462,
+      "step": 2604544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.347433081121595e-06,
+      "loss": 3.8698,
+      "step": 2605056
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.339047133611074e-06,
+      "loss": 3.8434,
+      "step": 2605568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.330661186100554e-06,
+      "loss": 3.8415,
+      "step": 2606080
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.322291617393766e-06,
+      "loss": 3.8494,
+      "step": 2606592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.313905669883246e-06,
+      "loss": 3.8542,
+      "step": 2607104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.305519722372725e-06,
+      "loss": 3.8398,
+      "step": 2607616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.297133774862206e-06,
+      "loss": 3.8424,
+      "step": 2608128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.288747827351686e-06,
+      "loss": 3.845,
+      "step": 2608640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.280361879841165e-06,
+      "loss": 3.8439,
+      "step": 2609152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.271975932330645e-06,
+      "loss": 3.8306,
+      "step": 2609664
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2635899848201256e-06,
+      "loss": 3.8393,
+      "step": 2610176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2552040373096055e-06,
+      "loss": 3.8409,
+      "step": 2610688
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2468344686028155e-06,
+      "loss": 3.8468,
+      "step": 2611200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.238448521092296e-06,
+      "loss": 3.8605,
+      "step": 2611712
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.230062573581776e-06,
+      "loss": 3.837,
+      "step": 2612224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.221693004874988e-06,
+      "loss": 3.842,
+      "step": 2612736
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.213307057364467e-06,
+      "loss": 3.8498,
+      "step": 2613248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.204921109853948e-06,
+      "loss": 3.8366,
+      "step": 2613760
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.196535162343428e-06,
+      "loss": 3.8383,
+      "step": 2614272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.188165593636638e-06,
+      "loss": 3.8349,
+      "step": 2614784
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1797796461261185e-06,
+      "loss": 3.8403,
+      "step": 2615296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.171393698615598e-06,
+      "loss": 3.8321,
+      "step": 2615808
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.163007751105079e-06,
+      "loss": 3.8575,
+      "step": 2616320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1546218035945575e-06,
+      "loss": 3.8353,
+      "step": 2616832
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.146235856084038e-06,
+      "loss": 3.8477,
+      "step": 2617344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.137849908573518e-06,
+      "loss": 3.8441,
+      "step": 2617856
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.129463961062997e-06,
+      "loss": 3.8408,
+      "step": 2618368
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.121078013552477e-06,
+      "loss": 3.8471,
+      "step": 2618880
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.112692066041958e-06,
+      "loss": 3.8295,
+      "step": 2619392
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.104306118531437e-06,
+      "loss": 3.8274,
+      "step": 2619904
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.095920171020917e-06,
+      "loss": 3.852,
+      "step": 2620416
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.087534223510397e-06,
+      "loss": 3.8406,
+      "step": 2620928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.079148275999878e-06,
+      "loss": 3.8345,
+      "step": 2621440
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.070762328489357e-06,
+      "loss": 3.8391,
+      "step": 2621952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.062376380978837e-06,
+      "loss": 3.8434,
+      "step": 2622464
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.054006812272049e-06,
+      "loss": 3.8252,
+      "step": 2622976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.04563724356526e-06,
+      "loss": 3.8463,
+      "step": 2623488
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.037251296054739e-06,
+      "loss": 3.8393,
+      "step": 2624000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.028865348544219e-06,
+      "loss": 3.8375,
+      "step": 2624512
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0204794010337e-06,
+      "loss": 3.8518,
+      "step": 2625024
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.01210983232691e-06,
+      "loss": 3.8306,
+      "step": 2625536
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.003723884816391e-06,
+      "loss": 3.8283,
+      "step": 2626048
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.995337937305871e-06,
+      "loss": 3.8415,
+      "step": 2626560
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.986951989795351e-06,
+      "loss": 3.8261,
+      "step": 2627072
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.97856604228483e-06,
+      "loss": 3.8318,
+      "step": 2627584
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.970180094774311e-06,
+      "loss": 3.8451,
+      "step": 2628096
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.961794147263791e-06,
+      "loss": 3.8479,
+      "step": 2628608
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.95340819975327e-06,
+      "loss": 3.8345,
+      "step": 2629120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.945038631046481e-06,
+      "loss": 3.8356,
+      "step": 2629632
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.936669062339693e-06,
+      "loss": 3.8245,
+      "step": 2630144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.928283114829173e-06,
+      "loss": 3.8443,
+      "step": 2630656
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.919913546122384e-06,
+      "loss": 3.8408,
+      "step": 2631168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9115439774155954e-06,
+      "loss": 3.8463,
+      "step": 2631680
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.903158029905075e-06,
+      "loss": 3.8406,
+      "step": 2632192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.894772082394556e-06,
+      "loss": 3.8485,
+      "step": 2632704
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.8863861348840345e-06,
+      "loss": 3.8547,
+      "step": 2633216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.878000187373515e-06,
+      "loss": 3.829,
+      "step": 2633728
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.869614239862995e-06,
+      "loss": 3.843,
+      "step": 2634240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.861228292352474e-06,
+      "loss": 3.8478,
+      "step": 2634752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.852842344841954e-06,
+      "loss": 3.8312,
+      "step": 2635264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.844456397331435e-06,
+      "loss": 3.8531,
+      "step": 2635776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.836070449820914e-06,
+      "loss": 3.8484,
+      "step": 2636288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.827684502310394e-06,
+      "loss": 3.8525,
+      "step": 2636800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.819298554799874e-06,
+      "loss": 3.8528,
+      "step": 2637312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.810928986093086e-06,
+      "loss": 3.8282,
+      "step": 2637824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.802543038582565e-06,
+      "loss": 3.8419,
+      "step": 2638336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.794157091072046e-06,
+      "loss": 3.8395,
+      "step": 2638848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.785771143561526e-06,
+      "loss": 3.8507,
+      "step": 2639360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.777385196051005e-06,
+      "loss": 3.8431,
+      "step": 2639872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.769015627344216e-06,
+      "loss": 3.8417,
+      "step": 2640384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.760629679833696e-06,
+      "loss": 3.8473,
+      "step": 2640896
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.752243732323177e-06,
+      "loss": 3.8453,
+      "step": 2641408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.743874163616387e-06,
+      "loss": 3.8312,
+      "step": 2641920
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.735488216105868e-06,
+      "loss": 3.8381,
+      "step": 2642432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.727102268595348e-06,
+      "loss": 3.8495,
+      "step": 2642944
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.718716321084828e-06,
+      "loss": 3.8517,
+      "step": 2643456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.710330373574307e-06,
+      "loss": 3.8436,
+      "step": 2643968
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.701944426063788e-06,
+      "loss": 3.8284,
+      "step": 2644480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.693558478553268e-06,
+      "loss": 3.8395,
+      "step": 2644992
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.685172531042747e-06,
+      "loss": 3.8483,
+      "step": 2645504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.676786583532227e-06,
+      "loss": 3.8462,
+      "step": 2646016
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.6684006360217075e-06,
+      "loss": 3.8405,
+      "step": 2646528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.660014688511186e-06,
+      "loss": 3.842,
+      "step": 2647040
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.651645119804397e-06,
+      "loss": 3.8293,
+      "step": 2647552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.643259172293878e-06,
+      "loss": 3.8466,
+      "step": 2648064
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.63488960358709e-06,
+      "loss": 3.8284,
+      "step": 2648576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.626503656076569e-06,
+      "loss": 3.8387,
+      "step": 2649088
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.618117708566049e-06,
+      "loss": 3.8347,
+      "step": 2649600
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.609731761055529e-06,
+      "loss": 3.8469,
+      "step": 2650112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.601362192348741e-06,
+      "loss": 3.8319,
+      "step": 2650624
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.59297624483822e-06,
+      "loss": 3.8445,
+      "step": 2651136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5845902973277e-06,
+      "loss": 3.8355,
+      "step": 2651648
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.57620434981718e-06,
+      "loss": 3.8373,
+      "step": 2652160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5678184023066595e-06,
+      "loss": 3.8393,
+      "step": 2652672
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5594324547961394e-06,
+      "loss": 3.8344,
+      "step": 2653184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.55104650728562e-06,
+      "loss": 3.8489,
+      "step": 2653696
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5426605597751e-06,
+      "loss": 3.8457,
+      "step": 2654208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.534307369872042e-06,
+      "loss": 3.8432,
+      "step": 2654720
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.525921422361523e-06,
+      "loss": 3.8382,
+      "step": 2655232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5175354748510026e-06,
+      "loss": 3.8449,
+      "step": 2655744
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5091495273404825e-06,
+      "loss": 3.8473,
+      "step": 2656256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.500763579829962e-06,
+      "loss": 3.8284,
+      "step": 2656768
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4923776323194424e-06,
+      "loss": 3.853,
+      "step": 2657280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.484008063612654e-06,
+      "loss": 3.8437,
+      "step": 2657792
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.475622116102132e-06,
+      "loss": 3.8476,
+      "step": 2658304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.467236168591613e-06,
+      "loss": 3.8364,
+      "step": 2658816
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.458850221081093e-06,
+      "loss": 3.8372,
+      "step": 2659328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.450464273570574e-06,
+      "loss": 3.8268,
+      "step": 2659840
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.442094704863784e-06,
+      "loss": 3.8355,
+      "step": 2660352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.433708757353265e-06,
+      "loss": 3.8329,
+      "step": 2660864
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.425322809842745e-06,
+      "loss": 3.8469,
+      "step": 2661376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.416936862332224e-06,
+      "loss": 3.8536,
+      "step": 2661888
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.408550914821704e-06,
+      "loss": 3.8453,
+      "step": 2662400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.4001649673111845e-06,
+      "loss": 3.8349,
+      "step": 2662912
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.391795398604396e-06,
+      "loss": 3.8432,
+      "step": 2663424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.383425829897606e-06,
+      "loss": 3.8469,
+      "step": 2663936
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.375039882387086e-06,
+      "loss": 3.8513,
+      "step": 2664448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.366653934876567e-06,
+      "loss": 3.8483,
+      "step": 2664960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.358267987366047e-06,
+      "loss": 3.8407,
+      "step": 2665472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.349882039855526e-06,
+      "loss": 3.8429,
+      "step": 2665984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.341496092345006e-06,
+      "loss": 3.8456,
+      "step": 2666496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3331265236382175e-06,
+      "loss": 3.8377,
+      "step": 2667008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.3247405761276966e-06,
+      "loss": 3.8473,
+      "step": 2667520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.316354628617177e-06,
+      "loss": 3.8423,
+      "step": 2668032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.307968681106657e-06,
+      "loss": 3.8365,
+      "step": 2668544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.2995827335961364e-06,
+      "loss": 3.8545,
+      "step": 2669056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.291196786085616e-06,
+      "loss": 3.8425,
+      "step": 2669568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.282827217378828e-06,
+      "loss": 3.8359,
+      "step": 2670080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.274441269868309e-06,
+      "loss": 3.8392,
+      "step": 2670592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.266055322357788e-06,
+      "loss": 3.8329,
+      "step": 2671104
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.9739737510681152,
+      "eval_runtime": 291.4227,
+      "eval_samples_per_second": 1309.407,
+      "eval_steps_per_second": 40.92,
+      "step": 2671200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.257669374847268e-06,
+      "loss": 3.8349,
+      "step": 2671616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.249283427336748e-06,
+      "loss": 3.8405,
+      "step": 2672128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.240897479826228e-06,
+      "loss": 3.855,
+      "step": 2672640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.232511532315708e-06,
+      "loss": 3.8384,
+      "step": 2673152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.224125584805188e-06,
+      "loss": 3.8586,
+      "step": 2673664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.215739637294668e-06,
+      "loss": 3.8328,
+      "step": 2674176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.207353689784148e-06,
+      "loss": 3.845,
+      "step": 2674688
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.198967742273628e-06,
+      "loss": 3.8293,
+      "step": 2675200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.190581794763107e-06,
+      "loss": 3.8475,
+      "step": 2675712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.182212226056318e-06,
+      "loss": 3.8465,
+      "step": 2676224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.173826278545798e-06,
+      "loss": 3.8422,
+      "step": 2676736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.16545670983901e-06,
+      "loss": 3.8509,
+      "step": 2677248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1570871411322216e-06,
+      "loss": 3.8348,
+      "step": 2677760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.148701193621701e-06,
+      "loss": 3.8333,
+      "step": 2678272
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.140315246111181e-06,
+      "loss": 3.8332,
+      "step": 2678784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1319292986006614e-06,
+      "loss": 3.8377,
+      "step": 2679296
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1235433510901405e-06,
+      "loss": 3.8454,
+      "step": 2679808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.115157403579621e-06,
+      "loss": 3.8406,
+      "step": 2680320
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.1067714560691004e-06,
+      "loss": 3.8431,
+      "step": 2680832
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.09838550855858e-06,
+      "loss": 3.8643,
+      "step": 2681344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.08999956104806e-06,
+      "loss": 3.846,
+      "step": 2681856
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.08161361353754e-06,
+      "loss": 3.8424,
+      "step": 2682368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.073227666027019e-06,
+      "loss": 3.8469,
+      "step": 2682880
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.0648417185165e-06,
+      "loss": 3.8488,
+      "step": 2683392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.056472149809711e-06,
+      "loss": 3.8362,
+      "step": 2683904
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.048086202299191e-06,
+      "loss": 3.8432,
+      "step": 2684416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.039700254788671e-06,
+      "loss": 3.8437,
+      "step": 2684928
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.0313306860818826e-06,
+      "loss": 3.8353,
+      "step": 2685440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.0229447385713625e-06,
+      "loss": 3.8336,
+      "step": 2685952
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.0145587910608425e-06,
+      "loss": 3.8331,
+      "step": 2686464
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.0061728435503224e-06,
+      "loss": 3.8417,
+      "step": 2686976
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.997819653647265e-06,
+      "loss": 3.8423,
+      "step": 2687488
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.989433706136745e-06,
+      "loss": 3.8566,
+      "step": 2688000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.981047758626225e-06,
+      "loss": 3.842,
+      "step": 2688512
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.9726781899194365e-06,
+      "loss": 3.8338,
+      "step": 2689024
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.964292242408916e-06,
+      "loss": 3.8452,
+      "step": 2689536
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.955906294898396e-06,
+      "loss": 3.8384,
+      "step": 2690048
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.9475203473878755e-06,
+      "loss": 3.8317,
+      "step": 2690560
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.939134399877356e-06,
+      "loss": 3.8346,
+      "step": 2691072
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.930748452366835e-06,
+      "loss": 3.8353,
+      "step": 2691584
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.922362504856316e-06,
+      "loss": 3.833,
+      "step": 2692096
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.913976557345795e-06,
+      "loss": 3.8511,
+      "step": 2692608
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.905590609835275e-06,
+      "loss": 3.8351,
+      "step": 2693120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.897204662324755e-06,
+      "loss": 3.8435,
+      "step": 2693632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.888818714814235e-06,
+      "loss": 3.8446,
+      "step": 2694144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.880432767303715e-06,
+      "loss": 3.8362,
+      "step": 2694656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.872046819793195e-06,
+      "loss": 3.8458,
+      "step": 2695168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.863660872282674e-06,
+      "loss": 3.8269,
+      "step": 2695680
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.855291303575886e-06,
+      "loss": 3.8187,
+      "step": 2696192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.846905356065366e-06,
+      "loss": 3.8528,
+      "step": 2696704
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.838519408554846e-06,
+      "loss": 3.8391,
+      "step": 2697216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.830133461044326e-06,
+      "loss": 3.8364,
+      "step": 2697728
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.821763892337537e-06,
+      "loss": 3.8341,
+      "step": 2698240
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.813377944827017e-06,
+      "loss": 3.8387,
+      "step": 2698752
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.804991997316497e-06,
+      "loss": 3.8255,
+      "step": 2699264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.796606049805977e-06,
+      "loss": 3.8411,
+      "step": 2699776
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.788220102295456e-06,
+      "loss": 3.8404,
+      "step": 2700288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.779834154784937e-06,
+      "loss": 3.8359,
+      "step": 2700800
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.771464586078148e-06,
+      "loss": 3.8471,
+      "step": 2701312
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.763078638567628e-06,
+      "loss": 3.8289,
+      "step": 2701824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.754692691057108e-06,
+      "loss": 3.8278,
+      "step": 2702336
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.746306743546588e-06,
+      "loss": 3.8437,
+      "step": 2702848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.737920796036068e-06,
+      "loss": 3.82,
+      "step": 2703360
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.729551227329279e-06,
+      "loss": 3.8248,
+      "step": 2703872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.721165279818759e-06,
+      "loss": 3.8442,
+      "step": 2704384
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.712779332308238e-06,
+      "loss": 3.8475,
+      "step": 2704896
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.704393384797719e-06,
+      "loss": 3.8296,
+      "step": 2705408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.69602381609093e-06,
+      "loss": 3.8372,
+      "step": 2705920
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.68763786858041e-06,
+      "loss": 3.8198,
+      "step": 2706432
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.67925192106989e-06,
+      "loss": 3.8427,
+      "step": 2706944
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.67086597355937e-06,
+      "loss": 3.8367,
+      "step": 2707456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.66248002604885e-06,
+      "loss": 3.8442,
+      "step": 2707968
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.65409407853833e-06,
+      "loss": 3.8373,
+      "step": 2708480
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.64570813102781e-06,
+      "loss": 3.8462,
+      "step": 2708992
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.63732218351729e-06,
+      "loss": 3.8535,
+      "step": 2709504
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.628952614810501e-06,
+      "loss": 3.8248,
+      "step": 2710016
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.620583046103712e-06,
+      "loss": 3.844,
+      "step": 2710528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.612197098593192e-06,
+      "loss": 3.8456,
+      "step": 2711040
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.603827529886403e-06,
+      "loss": 3.8226,
+      "step": 2711552
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.595441582375884e-06,
+      "loss": 3.8531,
+      "step": 2712064
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.587055634865363e-06,
+      "loss": 3.8472,
+      "step": 2712576
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.578669687354844e-06,
+      "loss": 3.8476,
+      "step": 2713088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.570283739844323e-06,
+      "loss": 3.8545,
+      "step": 2713600
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.561897792333803e-06,
+      "loss": 3.8315,
+      "step": 2714112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.553511844823283e-06,
+      "loss": 3.8352,
+      "step": 2714624
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.5451258973127626e-06,
+      "loss": 3.8376,
+      "step": 2715136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.5367399498022425e-06,
+      "loss": 3.8472,
+      "step": 2715648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.5283540022917225e-06,
+      "loss": 3.8401,
+      "step": 2716160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.519984433584934e-06,
+      "loss": 3.841,
+      "step": 2716672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.511598486074414e-06,
+      "loss": 3.8389,
+      "step": 2717184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.503228917367626e-06,
+      "loss": 3.8491,
+      "step": 2717696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.494842969857105e-06,
+      "loss": 3.8307,
+      "step": 2718208
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.486457022346585e-06,
+      "loss": 3.835,
+      "step": 2718720
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.478071074836065e-06,
+      "loss": 3.8474,
+      "step": 2719232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.469685127325545e-06,
+      "loss": 3.8488,
+      "step": 2719744
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.461315558618756e-06,
+      "loss": 3.8442,
+      "step": 2720256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.452929611108236e-06,
+      "loss": 3.8262,
+      "step": 2720768
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.444543663597715e-06,
+      "loss": 3.8355,
+      "step": 2721280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.436157716087196e-06,
+      "loss": 3.8441,
+      "step": 2721792
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.427771768576675e-06,
+      "loss": 3.8447,
+      "step": 2722304
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.419385821066156e-06,
+      "loss": 3.8413,
+      "step": 2722816
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.410999873555635e-06,
+      "loss": 3.8373,
+      "step": 2723328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.402613926045115e-06,
+      "loss": 3.8292,
+      "step": 2723840
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.394227978534595e-06,
+      "loss": 3.8422,
+      "step": 2724352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.385858409827806e-06,
+      "loss": 3.8254,
+      "step": 2724864
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.377472462317287e-06,
+      "loss": 3.8341,
+      "step": 2725376
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.369086514806766e-06,
+      "loss": 3.8291,
+      "step": 2725888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.360700567296247e-06,
+      "loss": 3.8439,
+      "step": 2726400
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.352314619785726e-06,
+      "loss": 3.8314,
+      "step": 2726912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.343928672275206e-06,
+      "loss": 3.8463,
+      "step": 2727424
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.3355427247646865e-06,
+      "loss": 3.8309,
+      "step": 2727936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.327173156057897e-06,
+      "loss": 3.8361,
+      "step": 2728448
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.318787208547377e-06,
+      "loss": 3.8347,
+      "step": 2728960
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.310401261036857e-06,
+      "loss": 3.8341,
+      "step": 2729472
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.302015313526337e-06,
+      "loss": 3.8414,
+      "step": 2729984
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.293629366015817e-06,
+      "loss": 3.8504,
+      "step": 2730496
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.285259797309029e-06,
+      "loss": 3.8414,
+      "step": 2731008
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.276873849798508e-06,
+      "loss": 3.8366,
+      "step": 2731520
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.268487902287988e-06,
+      "loss": 3.8433,
+      "step": 2732032
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.260101954777468e-06,
+      "loss": 3.8435,
+      "step": 2732544
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.251716007266948e-06,
+      "loss": 3.8272,
+      "step": 2733056
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.243330059756428e-06,
+      "loss": 3.8491,
+      "step": 2733568
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.234960491049639e-06,
+      "loss": 3.8376,
+      "step": 2734080
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.226574543539119e-06,
+      "loss": 3.8475,
+      "step": 2734592
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.218188596028599e-06,
+      "loss": 3.8348,
+      "step": 2735104
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.209802648518078e-06,
+      "loss": 3.8366,
+      "step": 2735616
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.201416701007559e-06,
+      "loss": 3.8223,
+      "step": 2736128
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.193030753497038e-06,
+      "loss": 3.8379,
+      "step": 2736640
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.184644805986519e-06,
+      "loss": 3.8308,
+      "step": 2737152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.17627523727973e-06,
+      "loss": 3.8403,
+      "step": 2737664
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.16788928976921e-06,
+      "loss": 3.8456,
+      "step": 2738176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.1595197210624214e-06,
+      "loss": 3.8461,
+      "step": 2738688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.151133773551901e-06,
+      "loss": 3.8313,
+      "step": 2739200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.142747826041381e-06,
+      "loss": 3.8426,
+      "step": 2739712
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.1343618785308604e-06,
+      "loss": 3.8451,
+      "step": 2740224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.125992309824072e-06,
+      "loss": 3.8455,
+      "step": 2740736
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.117606362313552e-06,
+      "loss": 3.8458,
+      "step": 2741248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.109220414803032e-06,
+      "loss": 3.8403,
+      "step": 2741760
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.100834467292512e-06,
+      "loss": 3.8396,
+      "step": 2742272
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.092448519781992e-06,
+      "loss": 3.8466,
+      "step": 2742784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.084078951075203e-06,
+      "loss": 3.8336,
+      "step": 2743296
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.0756930035646835e-06,
+      "loss": 3.8477,
+      "step": 2743808
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.067307056054163e-06,
+      "loss": 3.8362,
+      "step": 2744320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.0589211085436426e-06,
+      "loss": 3.839,
+      "step": 2744832
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.0505351610331225e-06,
+      "loss": 3.8488,
+      "step": 2745344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.0421492135226025e-06,
+      "loss": 3.8418,
+      "step": 2745856
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.033763266012083e-06,
+      "loss": 3.8324,
+      "step": 2746368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.025393697305294e-06,
+      "loss": 3.8337,
+      "step": 2746880
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.017007749794774e-06,
+      "loss": 3.8346,
+      "step": 2747392
+    },
+    {
+      "epoch": 1.03,
+      "eval_loss": 3.973461151123047,
+      "eval_runtime": 289.5777,
+      "eval_samples_per_second": 1317.75,
+      "eval_steps_per_second": 41.181,
+      "step": 2747520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.008621802284254e-06,
+      "loss": 3.8244,
+      "step": 2747904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000235854773733e-06,
+      "loss": 3.8399,
+      "step": 2748416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.991849907263214e-06,
+      "loss": 3.8494,
+      "step": 2748928
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.983463959752693e-06,
+      "loss": 3.8398,
+      "step": 2749440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.975078012242174e-06,
+      "loss": 3.8537,
+      "step": 2749952
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.966692064731653e-06,
+      "loss": 3.8369,
+      "step": 2750464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.958306117221133e-06,
+      "loss": 3.839,
+      "step": 2750976
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.949920169710613e-06,
+      "loss": 3.8316,
+      "step": 2751488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.941534222200093e-06,
+      "loss": 3.839,
+      "step": 2752000
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.933148274689573e-06,
+      "loss": 3.8436,
+      "step": 2752512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.924762327179053e-06,
+      "loss": 3.8415,
+      "step": 2753024
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.916376379668533e-06,
+      "loss": 3.8482,
+      "step": 2753536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.908006810961744e-06,
+      "loss": 3.8318,
+      "step": 2754048
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.899620863451223e-06,
+      "loss": 3.8308,
+      "step": 2754560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.891234915940704e-06,
+      "loss": 3.8316,
+      "step": 2755072
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.882848968430183e-06,
+      "loss": 3.8327,
+      "step": 2755584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.874463020919663e-06,
+      "loss": 3.8412,
+      "step": 2756096
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.866077073409143e-06,
+      "loss": 3.8349,
+      "step": 2756608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.857691125898623e-06,
+      "loss": 3.8437,
+      "step": 2757120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849305178388103e-06,
+      "loss": 3.8598,
+      "step": 2757632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.840919230877583e-06,
+      "loss": 3.8452,
+      "step": 2758144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.832549662170795e-06,
+      "loss": 3.839,
+      "step": 2758656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.824163714660275e-06,
+      "loss": 3.8407,
+      "step": 2759168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.815777767149754e-06,
+      "loss": 3.8452,
+      "step": 2759680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.807391819639235e-06,
+      "loss": 3.8351,
+      "step": 2760192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.799005872128714e-06,
+      "loss": 3.8412,
+      "step": 2760704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7906199246181945e-06,
+      "loss": 3.8388,
+      "step": 2761216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.782233977107674e-06,
+      "loss": 3.8348,
+      "step": 2761728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7738480295971536e-06,
+      "loss": 3.8305,
+      "step": 2762240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.765478460890365e-06,
+      "loss": 3.8343,
+      "step": 2762752
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.757092513379844e-06,
+      "loss": 3.8361,
+      "step": 2763264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.748706565869325e-06,
+      "loss": 3.8405,
+      "step": 2763776
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.740320618358804e-06,
+      "loss": 3.8543,
+      "step": 2764288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.731951049652016e-06,
+      "loss": 3.8397,
+      "step": 2764800
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7235814809452275e-06,
+      "loss": 3.8309,
+      "step": 2765312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7151955334347074e-06,
+      "loss": 3.843,
+      "step": 2765824
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.706809585924187e-06,
+      "loss": 3.8427,
+      "step": 2766336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.698423638413667e-06,
+      "loss": 3.8252,
+      "step": 2766848
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.690037690903147e-06,
+      "loss": 3.8302,
+      "step": 2767360
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.681651743392626e-06,
+      "loss": 3.8339,
+      "step": 2767872
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.673265795882107e-06,
+      "loss": 3.8287,
+      "step": 2768384
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.664879848371586e-06,
+      "loss": 3.852,
+      "step": 2768896
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.656493900861067e-06,
+      "loss": 3.8349,
+      "step": 2769408
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.648107953350546e-06,
+      "loss": 3.8367,
+      "step": 2769920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.639722005840026e-06,
+      "loss": 3.8409,
+      "step": 2770432
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.631336058329506e-06,
+      "loss": 3.8373,
+      "step": 2770944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.622966489622718e-06,
+      "loss": 3.8409,
+      "step": 2771456
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6145969209159294e-06,
+      "loss": 3.8272,
+      "step": 2771968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6062109734054085e-06,
+      "loss": 3.812,
+      "step": 2772480
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.597825025894889e-06,
+      "loss": 3.856,
+      "step": 2772992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5894390783843684e-06,
+      "loss": 3.832,
+      "step": 2773504
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.581053130873849e-06,
+      "loss": 3.8361,
+      "step": 2774016
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.572667183363328e-06,
+      "loss": 3.8342,
+      "step": 2774528
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.56429761465654e-06,
+      "loss": 3.8287,
+      "step": 2775040
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.55591166714602e-06,
+      "loss": 3.8261,
+      "step": 2775552
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5475257196355e-06,
+      "loss": 3.836,
+      "step": 2776064
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.53913977212498e-06,
+      "loss": 3.8389,
+      "step": 2776576
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.53075382461446e-06,
+      "loss": 3.8342,
+      "step": 2777088
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5223842559076715e-06,
+      "loss": 3.8507,
+      "step": 2777600
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5139983083971506e-06,
+      "loss": 3.8228,
+      "step": 2778112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.505612360886631e-06,
+      "loss": 3.8215,
+      "step": 2778624
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4972264133761105e-06,
+      "loss": 3.8447,
+      "step": 2779136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4888404658655904e-06,
+      "loss": 3.8159,
+      "step": 2779648
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.48045451835507e-06,
+      "loss": 3.8194,
+      "step": 2780160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.47206857084455e-06,
+      "loss": 3.843,
+      "step": 2780672
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.46368262333403e-06,
+      "loss": 3.8453,
+      "step": 2781184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455313054627241e-06,
+      "loss": 3.829,
+      "step": 2781696
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.446927107116722e-06,
+      "loss": 3.8368,
+      "step": 2782208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.438541159606201e-06,
+      "loss": 3.8191,
+      "step": 2782720
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.430155212095681e-06,
+      "loss": 3.8375,
+      "step": 2783232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.421769264585161e-06,
+      "loss": 3.8339,
+      "step": 2783744
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4133996958783725e-06,
+      "loss": 3.8404,
+      "step": 2784256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4050137483678525e-06,
+      "loss": 3.8339,
+      "step": 2784768
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3966278008573325e-06,
+      "loss": 3.8449,
+      "step": 2785280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.388241853346812e-06,
+      "loss": 3.8514,
+      "step": 2785792
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.379855905836292e-06,
+      "loss": 3.8253,
+      "step": 2786304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.371486337129504e-06,
+      "loss": 3.8356,
+      "step": 2786816
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.363100389618983e-06,
+      "loss": 3.8443,
+      "step": 2787328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.354714442108463e-06,
+      "loss": 3.8222,
+      "step": 2787840
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.346328494597943e-06,
+      "loss": 3.8479,
+      "step": 2788352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.337942547087423e-06,
+      "loss": 3.8485,
+      "step": 2788864
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.329556599576903e-06,
+      "loss": 3.8444,
+      "step": 2789376
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.321187030870115e-06,
+      "loss": 3.851,
+      "step": 2789888
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3128010833595945e-06,
+      "loss": 3.8306,
+      "step": 2790400
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3044151358490745e-06,
+      "loss": 3.8298,
+      "step": 2790912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.296029188338554e-06,
+      "loss": 3.8364,
+      "step": 2791424
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.287643240828034e-06,
+      "loss": 3.8477,
+      "step": 2791936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2792572933175135e-06,
+      "loss": 3.8408,
+      "step": 2792448
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.270887724610725e-06,
+      "loss": 3.8364,
+      "step": 2792960
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.262501777100205e-06,
+      "loss": 3.8344,
+      "step": 2793472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.254115829589684e-06,
+      "loss": 3.8476,
+      "step": 2793984
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.245729882079165e-06,
+      "loss": 3.8314,
+      "step": 2794496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.237343934568645e-06,
+      "loss": 3.8313,
+      "step": 2795008
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.228974365861857e-06,
+      "loss": 3.8434,
+      "step": 2795520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.220588418351336e-06,
+      "loss": 3.8499,
+      "step": 2796032
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.212218849644547e-06,
+      "loss": 3.841,
+      "step": 2796544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.203832902134027e-06,
+      "loss": 3.8266,
+      "step": 2797056
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.195446954623507e-06,
+      "loss": 3.8322,
+      "step": 2797568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.187061007112987e-06,
+      "loss": 3.837,
+      "step": 2798080
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.178675059602466e-06,
+      "loss": 3.8433,
+      "step": 2798592
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.170289112091947e-06,
+      "loss": 3.8384,
+      "step": 2799104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.161903164581426e-06,
+      "loss": 3.8395,
+      "step": 2799616
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.153517217070907e-06,
+      "loss": 3.828,
+      "step": 2800128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.145131269560386e-06,
+      "loss": 3.8339,
+      "step": 2800640
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.136745322049866e-06,
+      "loss": 3.8262,
+      "step": 2801152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.128375753343078e-06,
+      "loss": 3.8364,
+      "step": 2801664
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.119989805832558e-06,
+      "loss": 3.8195,
+      "step": 2802176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.111603858322038e-06,
+      "loss": 3.8494,
+      "step": 2802688
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.103217910811518e-06,
+      "loss": 3.8293,
+      "step": 2803200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.094848342104729e-06,
+      "loss": 3.8377,
+      "step": 2803712
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.086462394594208e-06,
+      "loss": 3.8306,
+      "step": 2804224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.078076447083689e-06,
+      "loss": 3.8328,
+      "step": 2804736
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.069690499573168e-06,
+      "loss": 3.8318,
+      "step": 2805248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.061304552062648e-06,
+      "loss": 3.8352,
+      "step": 2805760
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.05293498335586e-06,
+      "loss": 3.8391,
+      "step": 2806272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04454903584534e-06,
+      "loss": 3.844,
+      "step": 2806784
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0361794671385515e-06,
+      "loss": 3.8386,
+      "step": 2807296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0277935196280306e-06,
+      "loss": 3.8349,
+      "step": 2807808
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.019407572117511e-06,
+      "loss": 3.8381,
+      "step": 2808320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0110216246069905e-06,
+      "loss": 3.8468,
+      "step": 2808832
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.002635677096471e-06,
+      "loss": 3.8213,
+      "step": 2809344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.99424972958595e-06,
+      "loss": 3.8487,
+      "step": 2809856
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.98586378207543e-06,
+      "loss": 3.834,
+      "step": 2810368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.97747783456491e-06,
+      "loss": 3.8429,
+      "step": 2810880
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.96909188705439e-06,
+      "loss": 3.8324,
+      "step": 2811392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.960722318347602e-06,
+      "loss": 3.8358,
+      "step": 2811904
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.952336370837081e-06,
+      "loss": 3.8233,
+      "step": 2812416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.943950423326562e-06,
+      "loss": 3.8324,
+      "step": 2812928
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.935564475816042e-06,
+      "loss": 3.832,
+      "step": 2813440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.927178528305521e-06,
+      "loss": 3.8364,
+      "step": 2813952
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.918825338402464e-06,
+      "loss": 3.8407,
+      "step": 2814464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.910439390891944e-06,
+      "loss": 3.8476,
+      "step": 2814976
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902053443381424e-06,
+      "loss": 3.8274,
+      "step": 2815488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.893667495870903e-06,
+      "loss": 3.84,
+      "step": 2816000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.885281548360384e-06,
+      "loss": 3.8474,
+      "step": 2816512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.876895600849863e-06,
+      "loss": 3.8432,
+      "step": 2817024
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.868509653339344e-06,
+      "loss": 3.8421,
+      "step": 2817536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.860123705828823e-06,
+      "loss": 3.8414,
+      "step": 2818048
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.851737758318303e-06,
+      "loss": 3.8344,
+      "step": 2818560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.843351810807783e-06,
+      "loss": 3.8449,
+      "step": 2819072
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.834965863297263e-06,
+      "loss": 3.8301,
+      "step": 2819584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.826579915786743e-06,
+      "loss": 3.8441,
+      "step": 2820096
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8182103470799545e-06,
+      "loss": 3.8347,
+      "step": 2820608
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8098243995694344e-06,
+      "loss": 3.8369,
+      "step": 2821120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.801438452058914e-06,
+      "loss": 3.8467,
+      "step": 2821632
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7930688833521256e-06,
+      "loss": 3.841,
+      "step": 2822144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7846829358416056e-06,
+      "loss": 3.8313,
+      "step": 2822656
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.776296988331085e-06,
+      "loss": 3.8264,
+      "step": 2823168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7679110408205655e-06,
+      "loss": 3.8324,
+      "step": 2823680
+    },
+    {
+      "epoch": 0.03,
+      "eval_loss": 3.972881317138672,
+      "eval_runtime": 289.8053,
+      "eval_samples_per_second": 1316.715,
+      "eval_steps_per_second": 41.148,
+      "step": 2823840
+    }
+  ],
+  "logging_steps": 512,
+  "max_steps": 3052726,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 10,
+  "total_flos": 1.1618274350653952e+18,
+  "trial_name": null,
+  "trial_params": null
+}