diff --git "a/l2-7b-eu/checkpoint-1500/trainer_state.json" "b/l2-7b-eu/checkpoint-1500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/l2-7b-eu/checkpoint-1500/trainer_state.json"
@@ -0,0 +1,5291 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.4058990664321472,
+  "eval_steps": 500,
+  "global_step": 1500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00027059937762143147,
+      "grad_norm": 4.086390018463135,
+      "learning_rate": 0.0,
+      "loss": 3.2754,
+      "step": 1
+    },
+    {
+      "epoch": 0.0005411987552428629,
+      "grad_norm": 3.758815288543701,
+      "learning_rate": 9.017132551848513e-08,
+      "loss": 3.2863,
+      "step": 2
+    },
+    {
+      "epoch": 0.0010823975104857259,
+      "grad_norm": 3.8250608444213867,
+      "learning_rate": 2.705139765554554e-07,
+      "loss": 3.3425,
+      "step": 4
+    },
+    {
+      "epoch": 0.0016235962657285888,
+      "grad_norm": 3.8092095851898193,
+      "learning_rate": 4.5085662759242564e-07,
+      "loss": 3.3165,
+      "step": 6
+    },
+    {
+      "epoch": 0.0021647950209714517,
+      "grad_norm": 3.7621052265167236,
+      "learning_rate": 6.311992786293959e-07,
+      "loss": 3.3295,
+      "step": 8
+    },
+    {
+      "epoch": 0.002705993776214315,
+      "grad_norm": 3.4136276245117188,
+      "learning_rate": 8.115419296663661e-07,
+      "loss": 3.3073,
+      "step": 10
+    },
+    {
+      "epoch": 0.0032471925314571776,
+      "grad_norm": 2.855100393295288,
+      "learning_rate": 9.918845807033363e-07,
+      "loss": 3.3031,
+      "step": 12
+    },
+    {
+      "epoch": 0.0037883912867000408,
+      "grad_norm": 2.491767406463623,
+      "learning_rate": 1.1722272317403068e-06,
+      "loss": 3.2943,
+      "step": 14
+    },
+    {
+      "epoch": 0.0043295900419429035,
+      "grad_norm": 2.359778642654419,
+      "learning_rate": 1.3525698827772768e-06,
+      "loss": 3.2622,
+      "step": 16
+    },
+    {
+      "epoch": 0.004870788797185766,
+      "grad_norm": 2.037504196166992,
+      "learning_rate": 1.5329125338142473e-06,
+      "loss": 3.239,
+      "step": 18
+    },
+    {
+      "epoch": 0.00541198755242863,
+      "grad_norm": 2.8542497158050537,
+      "learning_rate": 1.7132551848512173e-06,
+      "loss": 3.2031,
+      "step": 20
+    },
+    {
+      "epoch": 0.0059531863076714925,
+      "grad_norm": 2.297046661376953,
+      "learning_rate": 1.8935978358881876e-06,
+      "loss": 3.1721,
+      "step": 22
+    },
+    {
+      "epoch": 0.006494385062914355,
+      "grad_norm": 2.2149112224578857,
+      "learning_rate": 2.0739404869251576e-06,
+      "loss": 3.121,
+      "step": 24
+    },
+    {
+      "epoch": 0.007035583818157218,
+      "grad_norm": 1.8048591613769531,
+      "learning_rate": 2.254283137962128e-06,
+      "loss": 3.0857,
+      "step": 26
+    },
+    {
+      "epoch": 0.0075767825734000815,
+      "grad_norm": 1.7466434240341187,
+      "learning_rate": 2.4346257889990986e-06,
+      "loss": 3.0489,
+      "step": 28
+    },
+    {
+      "epoch": 0.008117981328642944,
+      "grad_norm": 2.1722524166107178,
+      "learning_rate": 2.6149684400360686e-06,
+      "loss": 3.0016,
+      "step": 30
+    },
+    {
+      "epoch": 0.008659180083885807,
+      "grad_norm": 1.364578366279602,
+      "learning_rate": 2.7953110910730386e-06,
+      "loss": 2.9587,
+      "step": 32
+    },
+    {
+      "epoch": 0.00920037883912867,
+      "grad_norm": 1.5823427438735962,
+      "learning_rate": 2.9756537421100095e-06,
+      "loss": 2.931,
+      "step": 34
+    },
+    {
+      "epoch": 0.009741577594371532,
+      "grad_norm": 1.2367908954620361,
+      "learning_rate": 3.1559963931469796e-06,
+      "loss": 2.8953,
+      "step": 36
+    },
+    {
+      "epoch": 0.010282776349614395,
+      "grad_norm": 1.0437366962432861,
+      "learning_rate": 3.3363390441839496e-06,
+      "loss": 2.8412,
+      "step": 38
+    },
+    {
+      "epoch": 0.01082397510485726,
+      "grad_norm": 1.081803798675537,
+      "learning_rate": 3.5166816952209197e-06,
+      "loss": 2.7832,
+      "step": 40
+    },
+    {
+      "epoch": 0.011365173860100122,
+      "grad_norm": 0.9715840220451355,
+      "learning_rate": 3.69702434625789e-06,
+      "loss": 2.7729,
+      "step": 42
+    },
+    {
+      "epoch": 0.011906372615342985,
+      "grad_norm": 0.8603936433792114,
+      "learning_rate": 3.877366997294861e-06,
+      "loss": 2.6904,
+      "step": 44
+    },
+    {
+      "epoch": 0.012447571370585848,
+      "grad_norm": 0.8236231803894043,
+      "learning_rate": 4.057709648331831e-06,
+      "loss": 2.6908,
+      "step": 46
+    },
+    {
+      "epoch": 0.01298877012582871,
+      "grad_norm": 0.7681186199188232,
+      "learning_rate": 4.2380522993688015e-06,
+      "loss": 2.6212,
+      "step": 48
+    },
+    {
+      "epoch": 0.013529968881071573,
+      "grad_norm": 0.8002827167510986,
+      "learning_rate": 4.4183949504057716e-06,
+      "loss": 2.6035,
+      "step": 50
+    },
+    {
+      "epoch": 0.014071167636314436,
+      "grad_norm": 0.6757120490074158,
+      "learning_rate": 4.598737601442742e-06,
+      "loss": 2.595,
+      "step": 52
+    },
+    {
+      "epoch": 0.014612366391557299,
+      "grad_norm": 0.6619369387626648,
+      "learning_rate": 4.779080252479712e-06,
+      "loss": 2.5522,
+      "step": 54
+    },
+    {
+      "epoch": 0.015153565146800163,
+      "grad_norm": 0.6247105598449707,
+      "learning_rate": 4.959422903516682e-06,
+      "loss": 2.5079,
+      "step": 56
+    },
+    {
+      "epoch": 0.015694763902043024,
+      "grad_norm": 0.6559263467788696,
+      "learning_rate": 5.139765554553652e-06,
+      "loss": 2.5009,
+      "step": 58
+    },
+    {
+      "epoch": 0.01623596265728589,
+      "grad_norm": 0.6590877175331116,
+      "learning_rate": 5.320108205590623e-06,
+      "loss": 2.4648,
+      "step": 60
+    },
+    {
+      "epoch": 0.01677716141252875,
+      "grad_norm": 0.6045516133308411,
+      "learning_rate": 5.500450856627593e-06,
+      "loss": 2.421,
+      "step": 62
+    },
+    {
+      "epoch": 0.017318360167771614,
+      "grad_norm": 0.6533932089805603,
+      "learning_rate": 5.680793507664563e-06,
+      "loss": 2.3966,
+      "step": 64
+    },
+    {
+      "epoch": 0.01785955892301448,
+      "grad_norm": 0.6478094458580017,
+      "learning_rate": 5.861136158701533e-06,
+      "loss": 2.3903,
+      "step": 66
+    },
+    {
+      "epoch": 0.01840075767825734,
+      "grad_norm": 0.7349300980567932,
+      "learning_rate": 6.041478809738504e-06,
+      "loss": 2.3552,
+      "step": 68
+    },
+    {
+      "epoch": 0.018941956433500204,
+      "grad_norm": 0.6454821825027466,
+      "learning_rate": 6.221821460775474e-06,
+      "loss": 2.3262,
+      "step": 70
+    },
+    {
+      "epoch": 0.019483155188743065,
+      "grad_norm": 0.7321672439575195,
+      "learning_rate": 6.402164111812444e-06,
+      "loss": 2.3197,
+      "step": 72
+    },
+    {
+      "epoch": 0.02002435394398593,
+      "grad_norm": 0.7664237022399902,
+      "learning_rate": 6.582506762849414e-06,
+      "loss": 2.2992,
+      "step": 74
+    },
+    {
+      "epoch": 0.02056555269922879,
+      "grad_norm": 0.6843811869621277,
+      "learning_rate": 6.762849413886384e-06,
+      "loss": 2.2927,
+      "step": 76
+    },
+    {
+      "epoch": 0.021106751454471655,
+      "grad_norm": 0.7199612259864807,
+      "learning_rate": 6.9431920649233556e-06,
+      "loss": 2.2525,
+      "step": 78
+    },
+    {
+      "epoch": 0.02164795020971452,
+      "grad_norm": 0.778446614742279,
+      "learning_rate": 7.123534715960326e-06,
+      "loss": 2.2267,
+      "step": 80
+    },
+    {
+      "epoch": 0.02218914896495738,
+      "grad_norm": 0.9287930727005005,
+      "learning_rate": 7.303877366997296e-06,
+      "loss": 2.2206,
+      "step": 82
+    },
+    {
+      "epoch": 0.022730347720200245,
+      "grad_norm": 1.033782958984375,
+      "learning_rate": 7.484220018034266e-06,
+      "loss": 2.2063,
+      "step": 84
+    },
+    {
+      "epoch": 0.023271546475443106,
+      "grad_norm": 1.0132615566253662,
+      "learning_rate": 7.664562669071236e-06,
+      "loss": 2.1677,
+      "step": 86
+    },
+    {
+      "epoch": 0.02381274523068597,
+      "grad_norm": 0.9043529033660889,
+      "learning_rate": 7.844905320108207e-06,
+      "loss": 2.1696,
+      "step": 88
+    },
+    {
+      "epoch": 0.02435394398592883,
+      "grad_norm": 0.6718290448188782,
+      "learning_rate": 8.025247971145176e-06,
+      "loss": 2.1492,
+      "step": 90
+    },
+    {
+      "epoch": 0.024895142741171696,
+      "grad_norm": 0.9615944027900696,
+      "learning_rate": 8.205590622182147e-06,
+      "loss": 2.1452,
+      "step": 92
+    },
+    {
+      "epoch": 0.02543634149641456,
+      "grad_norm": 0.9435996413230896,
+      "learning_rate": 8.385933273219116e-06,
+      "loss": 2.1098,
+      "step": 94
+    },
+    {
+      "epoch": 0.02597754025165742,
+      "grad_norm": 0.7614261507987976,
+      "learning_rate": 8.566275924256087e-06,
+      "loss": 2.1286,
+      "step": 96
+    },
+    {
+      "epoch": 0.026518739006900285,
+      "grad_norm": 0.9416339993476868,
+      "learning_rate": 8.746618575293058e-06,
+      "loss": 2.1092,
+      "step": 98
+    },
+    {
+      "epoch": 0.027059937762143146,
+      "grad_norm": 0.9229443073272705,
+      "learning_rate": 8.926961226330027e-06,
+      "loss": 2.0932,
+      "step": 100
+    },
+    {
+      "epoch": 0.02760113651738601,
+      "grad_norm": 0.7135593295097351,
+      "learning_rate": 9.107303877366998e-06,
+      "loss": 2.0699,
+      "step": 102
+    },
+    {
+      "epoch": 0.028142335272628872,
+      "grad_norm": 1.0263723134994507,
+      "learning_rate": 9.287646528403967e-06,
+      "loss": 2.0445,
+      "step": 104
+    },
+    {
+      "epoch": 0.028683534027871736,
+      "grad_norm": 1.0300300121307373,
+      "learning_rate": 9.467989179440938e-06,
+      "loss": 2.0463,
+      "step": 106
+    },
+    {
+      "epoch": 0.029224732783114597,
+      "grad_norm": 0.8331286311149597,
+      "learning_rate": 9.648331830477909e-06,
+      "loss": 2.0381,
+      "step": 108
+    },
+    {
+      "epoch": 0.02976593153835746,
+      "grad_norm": 0.7501435875892639,
+      "learning_rate": 9.828674481514878e-06,
+      "loss": 2.0411,
+      "step": 110
+    },
+    {
+      "epoch": 0.030307130293600326,
+      "grad_norm": 0.6895191073417664,
+      "learning_rate": 1.0009017132551849e-05,
+      "loss": 2.0475,
+      "step": 112
+    },
+    {
+      "epoch": 0.030848329048843187,
+      "grad_norm": 0.95854252576828,
+      "learning_rate": 1.018935978358882e-05,
+      "loss": 2.0071,
+      "step": 114
+    },
+    {
+      "epoch": 0.03138952780408605,
+      "grad_norm": 1.1303929090499878,
+      "learning_rate": 1.036970243462579e-05,
+      "loss": 2.0008,
+      "step": 116
+    },
+    {
+      "epoch": 0.031930726559328916,
+      "grad_norm": 0.7708876729011536,
+      "learning_rate": 1.055004508566276e-05,
+      "loss": 2.0061,
+      "step": 118
+    },
+    {
+      "epoch": 0.03247192531457178,
+      "grad_norm": 0.9773860573768616,
+      "learning_rate": 1.073038773669973e-05,
+      "loss": 2.0096,
+      "step": 120
+    },
+    {
+      "epoch": 0.03301312406981464,
+      "grad_norm": 1.118385910987854,
+      "learning_rate": 1.09107303877367e-05,
+      "loss": 1.9939,
+      "step": 122
+    },
+    {
+      "epoch": 0.0335543228250575,
+      "grad_norm": 0.7215014696121216,
+      "learning_rate": 1.109107303877367e-05,
+      "loss": 1.9515,
+      "step": 124
+    },
+    {
+      "epoch": 0.03409552158030037,
+      "grad_norm": 0.9696834683418274,
+      "learning_rate": 1.1271415689810642e-05,
+      "loss": 1.9639,
+      "step": 126
+    },
+    {
+      "epoch": 0.03463672033554323,
+      "grad_norm": 0.945482611656189,
+      "learning_rate": 1.1451758340847611e-05,
+      "loss": 1.9397,
+      "step": 128
+    },
+    {
+      "epoch": 0.03517791909078609,
+      "grad_norm": 0.7454535365104675,
+      "learning_rate": 1.1632100991884582e-05,
+      "loss": 1.9353,
+      "step": 130
+    },
+    {
+      "epoch": 0.03571911784602896,
+      "grad_norm": 0.7824187278747559,
+      "learning_rate": 1.1812443642921551e-05,
+      "loss": 1.9227,
+      "step": 132
+    },
+    {
+      "epoch": 0.03626031660127182,
+      "grad_norm": 0.7939879894256592,
+      "learning_rate": 1.1992786293958522e-05,
+      "loss": 1.9126,
+      "step": 134
+    },
+    {
+      "epoch": 0.03680151535651468,
+      "grad_norm": 0.7776147723197937,
+      "learning_rate": 1.2173128944995491e-05,
+      "loss": 1.9002,
+      "step": 136
+    },
+    {
+      "epoch": 0.03734271411175754,
+      "grad_norm": 0.6580236554145813,
+      "learning_rate": 1.2353471596032462e-05,
+      "loss": 1.9121,
+      "step": 138
+    },
+    {
+      "epoch": 0.03788391286700041,
+      "grad_norm": 0.7200301289558411,
+      "learning_rate": 1.2533814247069433e-05,
+      "loss": 1.8885,
+      "step": 140
+    },
+    {
+      "epoch": 0.03842511162224327,
+      "grad_norm": 0.7958497405052185,
+      "learning_rate": 1.2714156898106402e-05,
+      "loss": 1.9095,
+      "step": 142
+    },
+    {
+      "epoch": 0.03896631037748613,
+      "grad_norm": 0.9120681881904602,
+      "learning_rate": 1.2894499549143375e-05,
+      "loss": 1.884,
+      "step": 144
+    },
+    {
+      "epoch": 0.039507509132729,
+      "grad_norm": 0.8108247518539429,
+      "learning_rate": 1.3074842200180342e-05,
+      "loss": 1.8656,
+      "step": 146
+    },
+    {
+      "epoch": 0.04004870788797186,
+      "grad_norm": 0.7010449171066284,
+      "learning_rate": 1.3255184851217315e-05,
+      "loss": 1.8635,
+      "step": 148
+    },
+    {
+      "epoch": 0.04058990664321472,
+      "grad_norm": 0.8178524374961853,
+      "learning_rate": 1.3435527502254284e-05,
+      "loss": 1.8933,
+      "step": 150
+    },
+    {
+      "epoch": 0.04113110539845758,
+      "grad_norm": 1.0447405576705933,
+      "learning_rate": 1.3615870153291255e-05,
+      "loss": 1.8523,
+      "step": 152
+    },
+    {
+      "epoch": 0.04167230415370045,
+      "grad_norm": 0.8516271710395813,
+      "learning_rate": 1.3796212804328224e-05,
+      "loss": 1.8528,
+      "step": 154
+    },
+    {
+      "epoch": 0.04221350290894331,
+      "grad_norm": 0.8437328934669495,
+      "learning_rate": 1.3976555455365195e-05,
+      "loss": 1.861,
+      "step": 156
+    },
+    {
+      "epoch": 0.04275470166418617,
+      "grad_norm": 0.851265549659729,
+      "learning_rate": 1.4156898106402164e-05,
+      "loss": 1.8315,
+      "step": 158
+    },
+    {
+      "epoch": 0.04329590041942904,
+      "grad_norm": 0.7337156534194946,
+      "learning_rate": 1.4337240757439135e-05,
+      "loss": 1.8354,
+      "step": 160
+    },
+    {
+      "epoch": 0.0438370991746719,
+      "grad_norm": 0.9754143357276917,
+      "learning_rate": 1.4517583408476104e-05,
+      "loss": 1.8252,
+      "step": 162
+    },
+    {
+      "epoch": 0.04437829792991476,
+      "grad_norm": 0.6172115802764893,
+      "learning_rate": 1.4697926059513075e-05,
+      "loss": 1.8094,
+      "step": 164
+    },
+    {
+      "epoch": 0.04491949668515762,
+      "grad_norm": 0.8304158449172974,
+      "learning_rate": 1.4878268710550044e-05,
+      "loss": 1.8078,
+      "step": 166
+    },
+    {
+      "epoch": 0.04546069544040049,
+      "grad_norm": 0.6388853788375854,
+      "learning_rate": 1.5058611361587017e-05,
+      "loss": 1.8106,
+      "step": 168
+    },
+    {
+      "epoch": 0.04600189419564335,
+      "grad_norm": 0.743231475353241,
+      "learning_rate": 1.5238954012623984e-05,
+      "loss": 1.8144,
+      "step": 170
+    },
+    {
+      "epoch": 0.04654309295088621,
+      "grad_norm": 0.6442289352416992,
+      "learning_rate": 1.5419296663660955e-05,
+      "loss": 1.7831,
+      "step": 172
+    },
+    {
+      "epoch": 0.04708429170612908,
+      "grad_norm": 0.6877187490463257,
+      "learning_rate": 1.559963931469793e-05,
+      "loss": 1.8043,
+      "step": 174
+    },
+    {
+      "epoch": 0.04762549046137194,
+      "grad_norm": 0.9389640688896179,
+      "learning_rate": 1.5779981965734897e-05,
+      "loss": 1.7869,
+      "step": 176
+    },
+    {
+      "epoch": 0.0481666892166148,
+      "grad_norm": 1.0456589460372925,
+      "learning_rate": 1.5960324616771868e-05,
+      "loss": 1.7681,
+      "step": 178
+    },
+    {
+      "epoch": 0.04870788797185766,
+      "grad_norm": 0.9617791175842285,
+      "learning_rate": 1.614066726780884e-05,
+      "loss": 1.7668,
+      "step": 180
+    },
+    {
+      "epoch": 0.04924908672710053,
+      "grad_norm": 0.9334360361099243,
+      "learning_rate": 1.632100991884581e-05,
+      "loss": 1.7893,
+      "step": 182
+    },
+    {
+      "epoch": 0.04979028548234339,
+      "grad_norm": 0.8952531814575195,
+      "learning_rate": 1.6501352569882777e-05,
+      "loss": 1.7758,
+      "step": 184
+    },
+    {
+      "epoch": 0.05033148423758625,
+      "grad_norm": 0.8544924855232239,
+      "learning_rate": 1.6681695220919748e-05,
+      "loss": 1.793,
+      "step": 186
+    },
+    {
+      "epoch": 0.05087268299282912,
+      "grad_norm": 0.7782765030860901,
+      "learning_rate": 1.686203787195672e-05,
+      "loss": 1.768,
+      "step": 188
+    },
+    {
+      "epoch": 0.05141388174807198,
+      "grad_norm": 0.7119695544242859,
+      "learning_rate": 1.704238052299369e-05,
+      "loss": 1.7685,
+      "step": 190
+    },
+    {
+      "epoch": 0.05195508050331484,
+      "grad_norm": 0.9119647145271301,
+      "learning_rate": 1.7222723174030657e-05,
+      "loss": 1.7706,
+      "step": 192
+    },
+    {
+      "epoch": 0.0524962792585577,
+      "grad_norm": 0.6414957642555237,
+      "learning_rate": 1.7403065825067628e-05,
+      "loss": 1.7626,
+      "step": 194
+    },
+    {
+      "epoch": 0.05303747801380057,
+      "grad_norm": 0.8069677352905273,
+      "learning_rate": 1.75834084761046e-05,
+      "loss": 1.7423,
+      "step": 196
+    },
+    {
+      "epoch": 0.05357867676904343,
+      "grad_norm": 0.6549937725067139,
+      "learning_rate": 1.776375112714157e-05,
+      "loss": 1.7428,
+      "step": 198
+    },
+    {
+      "epoch": 0.05411987552428629,
+      "grad_norm": 0.8064024448394775,
+      "learning_rate": 1.7944093778178538e-05,
+      "loss": 1.7448,
+      "step": 200
+    },
+    {
+      "epoch": 0.054661074279529154,
+      "grad_norm": 0.7182701826095581,
+      "learning_rate": 1.8124436429215512e-05,
+      "loss": 1.7248,
+      "step": 202
+    },
+    {
+      "epoch": 0.05520227303477202,
+      "grad_norm": 0.6997919678688049,
+      "learning_rate": 1.830477908025248e-05,
+      "loss": 1.7281,
+      "step": 204
+    },
+    {
+      "epoch": 0.05574347179001488,
+      "grad_norm": 0.7071277499198914,
+      "learning_rate": 1.848512173128945e-05,
+      "loss": 1.714,
+      "step": 206
+    },
+    {
+      "epoch": 0.056284670545257744,
+      "grad_norm": 0.6344273090362549,
+      "learning_rate": 1.866546438232642e-05,
+      "loss": 1.7463,
+      "step": 208
+    },
+    {
+      "epoch": 0.05682586930050061,
+      "grad_norm": 0.7192733883857727,
+      "learning_rate": 1.8845807033363392e-05,
+      "loss": 1.737,
+      "step": 210
+    },
+    {
+      "epoch": 0.05736706805574347,
+      "grad_norm": 0.7418521642684937,
+      "learning_rate": 1.9026149684400363e-05,
+      "loss": 1.7197,
+      "step": 212
+    },
+    {
+      "epoch": 0.057908266810986334,
+      "grad_norm": 0.875845730304718,
+      "learning_rate": 1.920649233543733e-05,
+      "loss": 1.6968,
+      "step": 214
+    },
+    {
+      "epoch": 0.058449465566229195,
+      "grad_norm": 0.7394037842750549,
+      "learning_rate": 1.9386834986474305e-05,
+      "loss": 1.7051,
+      "step": 216
+    },
+    {
+      "epoch": 0.05899066432147206,
+      "grad_norm": 0.6689572930335999,
+      "learning_rate": 1.9567177637511272e-05,
+      "loss": 1.7152,
+      "step": 218
+    },
+    {
+      "epoch": 0.05953186307671492,
+      "grad_norm": 0.7955539226531982,
+      "learning_rate": 1.9747520288548243e-05,
+      "loss": 1.7136,
+      "step": 220
+    },
+    {
+      "epoch": 0.060073061831957784,
+      "grad_norm": 0.7005388140678406,
+      "learning_rate": 1.9927862939585214e-05,
+      "loss": 1.7152,
+      "step": 222
+    },
+    {
+      "epoch": 0.06061426058720065,
+      "grad_norm": 0.6205731630325317,
+      "learning_rate": 2.0108205590622185e-05,
+      "loss": 1.6901,
+      "step": 224
+    },
+    {
+      "epoch": 0.06115545934244351,
+      "grad_norm": 0.7079929709434509,
+      "learning_rate": 2.0288548241659152e-05,
+      "loss": 1.6905,
+      "step": 226
+    },
+    {
+      "epoch": 0.061696658097686374,
+      "grad_norm": 0.6871302723884583,
+      "learning_rate": 2.0468890892696123e-05,
+      "loss": 1.6867,
+      "step": 228
+    },
+    {
+      "epoch": 0.062237856852929235,
+      "grad_norm": 0.7172162532806396,
+      "learning_rate": 2.0649233543733094e-05,
+      "loss": 1.685,
+      "step": 230
+    },
+    {
+      "epoch": 0.0627790556081721,
+      "grad_norm": 0.6729004979133606,
+      "learning_rate": 2.0829576194770065e-05,
+      "loss": 1.6961,
+      "step": 232
+    },
+    {
+      "epoch": 0.06332025436341496,
+      "grad_norm": 0.7335099577903748,
+      "learning_rate": 2.1009918845807033e-05,
+      "loss": 1.6797,
+      "step": 234
+    },
+    {
+      "epoch": 0.06386145311865783,
+      "grad_norm": 0.6398060321807861,
+      "learning_rate": 2.1190261496844003e-05,
+      "loss": 1.7037,
+      "step": 236
+    },
+    {
+      "epoch": 0.0644026518739007,
+      "grad_norm": 0.7026365399360657,
+      "learning_rate": 2.1370604147880974e-05,
+      "loss": 1.6698,
+      "step": 238
+    },
+    {
+      "epoch": 0.06494385062914355,
+      "grad_norm": 0.7972332239151001,
+      "learning_rate": 2.1550946798917945e-05,
+      "loss": 1.6866,
+      "step": 240
+    },
+    {
+      "epoch": 0.06548504938438642,
+      "grad_norm": 0.7363021969795227,
+      "learning_rate": 2.1731289449954913e-05,
+      "loss": 1.6879,
+      "step": 242
+    },
+    {
+      "epoch": 0.06602624813962928,
+      "grad_norm": 0.7071017026901245,
+      "learning_rate": 2.1911632100991887e-05,
+      "loss": 1.6898,
+      "step": 244
+    },
+    {
+      "epoch": 0.06656744689487214,
+      "grad_norm": 0.8030880093574524,
+      "learning_rate": 2.2091974752028858e-05,
+      "loss": 1.6734,
+      "step": 246
+    },
+    {
+      "epoch": 0.067108645650115,
+      "grad_norm": 0.7429569363594055,
+      "learning_rate": 2.2272317403065825e-05,
+      "loss": 1.6722,
+      "step": 248
+    },
+    {
+      "epoch": 0.06764984440535787,
+      "grad_norm": 0.6807804107666016,
+      "learning_rate": 2.2452660054102796e-05,
+      "loss": 1.6697,
+      "step": 250
+    },
+    {
+      "epoch": 0.06819104316060073,
+      "grad_norm": 0.6632562875747681,
+      "learning_rate": 2.2633002705139767e-05,
+      "loss": 1.6453,
+      "step": 252
+    },
+    {
+      "epoch": 0.0687322419158436,
+      "grad_norm": 0.6661680340766907,
+      "learning_rate": 2.2813345356176738e-05,
+      "loss": 1.6701,
+      "step": 254
+    },
+    {
+      "epoch": 0.06927344067108646,
+      "grad_norm": 0.6747105121612549,
+      "learning_rate": 2.2993688007213706e-05,
+      "loss": 1.6729,
+      "step": 256
+    },
+    {
+      "epoch": 0.06981463942632932,
+      "grad_norm": 0.7698473334312439,
+      "learning_rate": 2.317403065825068e-05,
+      "loss": 1.6528,
+      "step": 258
+    },
+    {
+      "epoch": 0.07035583818157218,
+      "grad_norm": 0.6111325621604919,
+      "learning_rate": 2.3354373309287647e-05,
+      "loss": 1.6412,
+      "step": 260
+    },
+    {
+      "epoch": 0.07089703693681504,
+      "grad_norm": 0.7405019998550415,
+      "learning_rate": 2.3534715960324618e-05,
+      "loss": 1.6564,
+      "step": 262
+    },
+    {
+      "epoch": 0.07143823569205791,
+      "grad_norm": 0.6702501773834229,
+      "learning_rate": 2.371505861136159e-05,
+      "loss": 1.654,
+      "step": 264
+    },
+    {
+      "epoch": 0.07197943444730077,
+      "grad_norm": 0.7076373100280762,
+      "learning_rate": 2.389540126239856e-05,
+      "loss": 1.6301,
+      "step": 266
+    },
+    {
+      "epoch": 0.07252063320254364,
+      "grad_norm": 0.7239627242088318,
+      "learning_rate": 2.4075743913435528e-05,
+      "loss": 1.6575,
+      "step": 268
+    },
+    {
+      "epoch": 0.0730618319577865,
+      "grad_norm": 0.753480076789856,
+      "learning_rate": 2.42560865644725e-05,
+      "loss": 1.6603,
+      "step": 270
+    },
+    {
+      "epoch": 0.07360303071302936,
+      "grad_norm": 0.7261641025543213,
+      "learning_rate": 2.443642921550947e-05,
+      "loss": 1.6449,
+      "step": 272
+    },
+    {
+      "epoch": 0.07414422946827222,
+      "grad_norm": 0.6315119862556458,
+      "learning_rate": 2.461677186654644e-05,
+      "loss": 1.6538,
+      "step": 274
+    },
+    {
+      "epoch": 0.07468542822351508,
+      "grad_norm": 0.5698412656784058,
+      "learning_rate": 2.4797114517583408e-05,
+      "loss": 1.6663,
+      "step": 276
+    },
+    {
+      "epoch": 0.07522662697875795,
+      "grad_norm": 0.5968983173370361,
+      "learning_rate": 2.497745716862038e-05,
+      "loss": 1.643,
+      "step": 278
+    },
+    {
+      "epoch": 0.07576782573400082,
+      "grad_norm": 0.561126172542572,
+      "learning_rate": 2.5157799819657353e-05,
+      "loss": 1.6301,
+      "step": 280
+    },
+    {
+      "epoch": 0.07630902448924368,
+      "grad_norm": 0.7290865778923035,
+      "learning_rate": 2.533814247069432e-05,
+      "loss": 1.6412,
+      "step": 282
+    },
+    {
+      "epoch": 0.07685022324448654,
+      "grad_norm": 0.7629122138023376,
+      "learning_rate": 2.5518485121731288e-05,
+      "loss": 1.6335,
+      "step": 284
+    },
+    {
+      "epoch": 0.0773914219997294,
+      "grad_norm": 0.5383496284484863,
+      "learning_rate": 2.5698827772768262e-05,
+      "loss": 1.6226,
+      "step": 286
+    },
+    {
+      "epoch": 0.07793262075497226,
+      "grad_norm": 0.7778373956680298,
+      "learning_rate": 2.5879170423805233e-05,
+      "loss": 1.6333,
+      "step": 288
+    },
+    {
+      "epoch": 0.07847381951021512,
+      "grad_norm": 0.6851366758346558,
+      "learning_rate": 2.60595130748422e-05,
+      "loss": 1.6251,
+      "step": 290
+    },
+    {
+      "epoch": 0.079015018265458,
+      "grad_norm": 0.5947225689888,
+      "learning_rate": 2.623985572587917e-05,
+      "loss": 1.6298,
+      "step": 292
+    },
+    {
+      "epoch": 0.07955621702070086,
+      "grad_norm": 0.9742544889450073,
+      "learning_rate": 2.6420198376916146e-05,
+      "loss": 1.6252,
+      "step": 294
+    },
+    {
+      "epoch": 0.08009741577594372,
+      "grad_norm": 1.2064323425292969,
+      "learning_rate": 2.6600541027953113e-05,
+      "loss": 1.6152,
+      "step": 296
+    },
+    {
+      "epoch": 0.08063861453118658,
+      "grad_norm": 1.0506716966629028,
+      "learning_rate": 2.678088367899008e-05,
+      "loss": 1.6351,
+      "step": 298
+    },
+    {
+      "epoch": 0.08117981328642944,
+      "grad_norm": 1.2992738485336304,
+      "learning_rate": 2.696122633002705e-05,
+      "loss": 1.6193,
+      "step": 300
+    },
+    {
+      "epoch": 0.0817210120416723,
+      "grad_norm": 1.0616599321365356,
+      "learning_rate": 2.7141568981064026e-05,
+      "loss": 1.6135,
+      "step": 302
+    },
+    {
+      "epoch": 0.08226221079691516,
+      "grad_norm": 1.037997841835022,
+      "learning_rate": 2.7321911632100993e-05,
+      "loss": 1.6344,
+      "step": 304
+    },
+    {
+      "epoch": 0.08280340955215804,
+      "grad_norm": 0.8937569856643677,
+      "learning_rate": 2.7502254283137964e-05,
+      "loss": 1.6077,
+      "step": 306
+    },
+    {
+      "epoch": 0.0833446083074009,
+      "grad_norm": 1.1334234476089478,
+      "learning_rate": 2.7682596934174932e-05,
+      "loss": 1.6193,
+      "step": 308
+    },
+    {
+      "epoch": 0.08388580706264376,
+      "grad_norm": 0.8336219191551208,
+      "learning_rate": 2.7862939585211906e-05,
+      "loss": 1.5948,
+      "step": 310
+    },
+    {
+      "epoch": 0.08442700581788662,
+      "grad_norm": 1.1825398206710815,
+      "learning_rate": 2.8043282236248874e-05,
+      "loss": 1.6239,
+      "step": 312
+    },
+    {
+      "epoch": 0.08496820457312948,
+      "grad_norm": 0.7945433259010315,
+      "learning_rate": 2.8223624887285844e-05,
+      "loss": 1.6119,
+      "step": 314
+    },
+    {
+      "epoch": 0.08550940332837234,
+      "grad_norm": 0.6971009969711304,
+      "learning_rate": 2.8403967538322812e-05,
+      "loss": 1.5822,
+      "step": 316
+    },
+    {
+      "epoch": 0.0860506020836152,
+      "grad_norm": 0.6050766706466675,
+      "learning_rate": 2.8584310189359786e-05,
+      "loss": 1.6161,
+      "step": 318
+    },
+    {
+      "epoch": 0.08659180083885808,
+      "grad_norm": 0.6123189330101013,
+      "learning_rate": 2.8764652840396754e-05,
+      "loss": 1.5941,
+      "step": 320
+    },
+    {
+      "epoch": 0.08713299959410094,
+      "grad_norm": 0.5471253395080566,
+      "learning_rate": 2.8944995491433725e-05,
+      "loss": 1.603,
+      "step": 322
+    },
+    {
+      "epoch": 0.0876741983493438,
+      "grad_norm": 0.5793882608413696,
+      "learning_rate": 2.91253381424707e-05,
+      "loss": 1.6076,
+      "step": 324
+    },
+    {
+      "epoch": 0.08821539710458666,
+      "grad_norm": 0.5409413576126099,
+      "learning_rate": 2.9305680793507666e-05,
+      "loss": 1.5825,
+      "step": 326
+    },
+    {
+      "epoch": 0.08875659585982952,
+      "grad_norm": 6.757148265838623,
+      "learning_rate": 2.9486023444544637e-05,
+      "loss": 1.5942,
+      "step": 328
+    },
+    {
+      "epoch": 0.08929779461507238,
+      "grad_norm": 1.3357856273651123,
+      "learning_rate": 2.9666366095581605e-05,
+      "loss": 1.642,
+      "step": 330
+    },
+    {
+      "epoch": 0.08983899337031524,
+      "grad_norm": 0.8245829939842224,
+      "learning_rate": 2.984670874661858e-05,
+      "loss": 1.6062,
+      "step": 332
+    },
+    {
+      "epoch": 0.09038019212555812,
+      "grad_norm": 0.8888993263244629,
+      "learning_rate": 3.0027051397655547e-05,
+      "loss": 1.5952,
+      "step": 334
+    },
+    {
+      "epoch": 0.09092139088080098,
+      "grad_norm": 0.8923915028572083,
+      "learning_rate": 3.0207394048692517e-05,
+      "loss": 1.5977,
+      "step": 336
+    },
+    {
+      "epoch": 0.09146258963604384,
+      "grad_norm": 0.7443459033966064,
+      "learning_rate": 3.0387736699729485e-05,
+      "loss": 1.5738,
+      "step": 338
+    },
+    {
+      "epoch": 0.0920037883912867,
+      "grad_norm": 0.7297430038452148,
+      "learning_rate": 3.056807935076646e-05,
+      "loss": 1.5907,
+      "step": 340
+    },
+    {
+      "epoch": 0.09254498714652956,
+      "grad_norm": 0.6882812976837158,
+      "learning_rate": 3.074842200180343e-05,
+      "loss": 1.5767,
+      "step": 342
+    },
+    {
+      "epoch": 0.09308618590177242,
+      "grad_norm": 0.6150392889976501,
+      "learning_rate": 3.0928764652840394e-05,
+      "loss": 1.5747,
+      "step": 344
+    },
+    {
+      "epoch": 0.09362738465701528,
+      "grad_norm": 0.6230599284172058,
+      "learning_rate": 3.110910730387737e-05,
+      "loss": 1.583,
+      "step": 346
+    },
+    {
+      "epoch": 0.09416858341225816,
+      "grad_norm": 0.6081874966621399,
+      "learning_rate": 3.128944995491434e-05,
+      "loss": 1.5875,
+      "step": 348
+    },
+    {
+      "epoch": 0.09470978216750102,
+      "grad_norm": 0.5467821955680847,
+      "learning_rate": 3.146979260595131e-05,
+      "loss": 1.575,
+      "step": 350
+    },
+    {
+      "epoch": 0.09525098092274388,
+      "grad_norm": 0.5629361271858215,
+      "learning_rate": 3.165013525698828e-05,
+      "loss": 1.5828,
+      "step": 352
+    },
+    {
+      "epoch": 0.09579217967798674,
+      "grad_norm": 0.5995283126831055,
+      "learning_rate": 3.1830477908025245e-05,
+      "loss": 1.5872,
+      "step": 354
+    },
+    {
+      "epoch": 0.0963333784332296,
+      "grad_norm": 0.556450366973877,
+      "learning_rate": 3.201082055906222e-05,
+      "loss": 1.553,
+      "step": 356
+    },
+    {
+      "epoch": 0.09687457718847246,
+      "grad_norm": 0.6498537063598633,
+      "learning_rate": 3.219116321009919e-05,
+      "loss": 1.5667,
+      "step": 358
+    },
+    {
+      "epoch": 0.09741577594371532,
+      "grad_norm": 0.5891172885894775,
+      "learning_rate": 3.237150586113616e-05,
+      "loss": 1.5818,
+      "step": 360
+    },
+    {
+      "epoch": 0.0979569746989582,
+      "grad_norm": 0.6487797498703003,
+      "learning_rate": 3.2551848512173136e-05,
+      "loss": 1.5582,
+      "step": 362
+    },
+    {
+      "epoch": 0.09849817345420106,
+      "grad_norm": 0.5860658884048462,
+      "learning_rate": 3.27321911632101e-05,
+      "loss": 1.5725,
+      "step": 364
+    },
+    {
+      "epoch": 0.09903937220944392,
+      "grad_norm": 0.5619581937789917,
+      "learning_rate": 3.291253381424707e-05,
+      "loss": 1.5779,
+      "step": 366
+    },
+    {
+      "epoch": 0.09958057096468678,
+      "grad_norm": 0.7147429585456848,
+      "learning_rate": 3.309287646528404e-05,
+      "loss": 1.5766,
+      "step": 368
+    },
+    {
+      "epoch": 0.10012176971992964,
+      "grad_norm": 0.5840562582015991,
+      "learning_rate": 3.327321911632101e-05,
+      "loss": 1.5609,
+      "step": 370
+    },
+    {
+      "epoch": 0.1006629684751725,
+      "grad_norm": 0.6277860403060913,
+      "learning_rate": 3.345356176735798e-05,
+      "loss": 1.5645,
+      "step": 372
+    },
+    {
+      "epoch": 0.10120416723041536,
+      "grad_norm": 0.6395567655563354,
+      "learning_rate": 3.3633904418394954e-05,
+      "loss": 1.545,
+      "step": 374
+    },
+    {
+      "epoch": 0.10174536598565824,
+      "grad_norm": 0.6651553511619568,
+      "learning_rate": 3.381424706943192e-05,
+      "loss": 1.5643,
+      "step": 376
+    },
+    {
+      "epoch": 0.1022865647409011,
+      "grad_norm": 0.6691033244132996,
+      "learning_rate": 3.3994589720468896e-05,
+      "loss": 1.5705,
+      "step": 378
+    },
+    {
+      "epoch": 0.10282776349614396,
+      "grad_norm": 0.5426511764526367,
+      "learning_rate": 3.4174932371505863e-05,
+      "loss": 1.536,
+      "step": 380
+    },
+    {
+      "epoch": 0.10336896225138682,
+      "grad_norm": 0.6677694916725159,
+      "learning_rate": 3.435527502254283e-05,
+      "loss": 1.5664,
+      "step": 382
+    },
+    {
+      "epoch": 0.10391016100662968,
+      "grad_norm": 0.5283762216567993,
+      "learning_rate": 3.45356176735798e-05,
+      "loss": 1.5474,
+      "step": 384
+    },
+    {
+      "epoch": 0.10445135976187254,
+      "grad_norm": 0.652812659740448,
+      "learning_rate": 3.471596032461677e-05,
+      "loss": 1.5509,
+      "step": 386
+    },
+    {
+      "epoch": 0.1049925585171154,
+      "grad_norm": 0.8639987111091614,
+      "learning_rate": 3.489630297565375e-05,
+      "loss": 1.5563,
+      "step": 388
+    },
+    {
+      "epoch": 0.10553375727235827,
+      "grad_norm": 0.7726946473121643,
+      "learning_rate": 3.5076645626690715e-05,
+      "loss": 1.5682,
+      "step": 390
+    },
+    {
+      "epoch": 0.10607495602760114,
+      "grad_norm": 0.6511155962944031,
+      "learning_rate": 3.525698827772768e-05,
+      "loss": 1.5571,
+      "step": 392
+    },
+    {
+      "epoch": 0.106616154782844,
+      "grad_norm": 0.6578395962715149,
+      "learning_rate": 3.5437330928764656e-05,
+      "loss": 1.5452,
+      "step": 394
+    },
+    {
+      "epoch": 0.10715735353808686,
+      "grad_norm": 0.642919659614563,
+      "learning_rate": 3.5617673579801624e-05,
+      "loss": 1.5508,
+      "step": 396
+    },
+    {
+      "epoch": 0.10769855229332972,
+      "grad_norm": 0.5190348029136658,
+      "learning_rate": 3.579801623083859e-05,
+      "loss": 1.5432,
+      "step": 398
+    },
+    {
+      "epoch": 0.10823975104857259,
+      "grad_norm": 0.48932549357414246,
+      "learning_rate": 3.5978358881875566e-05,
+      "loss": 1.5544,
+      "step": 400
+    },
+    {
+      "epoch": 0.10878094980381545,
+      "grad_norm": 0.5018340945243835,
+      "learning_rate": 3.615870153291254e-05,
+      "loss": 1.5322,
+      "step": 402
+    },
+    {
+      "epoch": 0.10932214855905831,
+      "grad_norm": 0.5701499581336975,
+      "learning_rate": 3.633904418394951e-05,
+      "loss": 1.5288,
+      "step": 404
+    },
+    {
+      "epoch": 0.10986334731430118,
+      "grad_norm": 0.6049205660820007,
+      "learning_rate": 3.6519386834986475e-05,
+      "loss": 1.5627,
+      "step": 406
+    },
+    {
+      "epoch": 0.11040454606954404,
+      "grad_norm": 0.5781517028808594,
+      "learning_rate": 3.669972948602345e-05,
+      "loss": 1.542,
+      "step": 408
+    },
+    {
+      "epoch": 0.1109457448247869,
+      "grad_norm": 0.5594660043716431,
+      "learning_rate": 3.688007213706042e-05,
+      "loss": 1.5461,
+      "step": 410
+    },
+    {
+      "epoch": 0.11148694358002977,
+      "grad_norm": 0.5319619178771973,
+      "learning_rate": 3.7060414788097384e-05,
+      "loss": 1.5668,
+      "step": 412
+    },
+    {
+      "epoch": 0.11202814233527263,
+      "grad_norm": 0.5311123728752136,
+      "learning_rate": 3.724075743913435e-05,
+      "loss": 1.528,
+      "step": 414
+    },
+    {
+      "epoch": 0.11256934109051549,
+      "grad_norm": 0.5555101633071899,
+      "learning_rate": 3.7421100090171326e-05,
+      "loss": 1.5392,
+      "step": 416
+    },
+    {
+      "epoch": 0.11311053984575835,
+      "grad_norm": 0.5486223101615906,
+      "learning_rate": 3.76014427412083e-05,
+      "loss": 1.5337,
+      "step": 418
+    },
+    {
+      "epoch": 0.11365173860100122,
+      "grad_norm": 0.5156669020652771,
+      "learning_rate": 3.778178539224527e-05,
+      "loss": 1.5105,
+      "step": 420
+    },
+    {
+      "epoch": 0.11419293735624408,
+      "grad_norm": 0.49596554040908813,
+      "learning_rate": 3.7962128043282235e-05,
+      "loss": 1.515,
+      "step": 422
+    },
+    {
+      "epoch": 0.11473413611148695,
+      "grad_norm": 0.641333281993866,
+      "learning_rate": 3.814247069431921e-05,
+      "loss": 1.5328,
+      "step": 424
+    },
+    {
+      "epoch": 0.1152753348667298,
+      "grad_norm": 0.6106113195419312,
+      "learning_rate": 3.832281334535618e-05,
+      "loss": 1.5189,
+      "step": 426
+    },
+    {
+      "epoch": 0.11581653362197267,
+      "grad_norm": 0.5619134306907654,
+      "learning_rate": 3.8503155996393145e-05,
+      "loss": 1.5295,
+      "step": 428
+    },
+    {
+      "epoch": 0.11635773237721553,
+      "grad_norm": 0.5396978259086609,
+      "learning_rate": 3.868349864743012e-05,
+      "loss": 1.5173,
+      "step": 430
+    },
+    {
+      "epoch": 0.11689893113245839,
+      "grad_norm": 0.5466894507408142,
+      "learning_rate": 3.886384129846709e-05,
+      "loss": 1.5191,
+      "step": 432
+    },
+    {
+      "epoch": 0.11744012988770126,
+      "grad_norm": 0.5601218342781067,
+      "learning_rate": 3.904418394950406e-05,
+      "loss": 1.5285,
+      "step": 434
+    },
+    {
+      "epoch": 0.11798132864294412,
+      "grad_norm": 0.6620492935180664,
+      "learning_rate": 3.922452660054103e-05,
+      "loss": 1.4946,
+      "step": 436
+    },
+    {
+      "epoch": 0.11852252739818699,
+      "grad_norm": 0.49140048027038574,
+      "learning_rate": 3.9404869251578e-05,
+      "loss": 1.512,
+      "step": 438
+    },
+    {
+      "epoch": 0.11906372615342985,
+      "grad_norm": 0.5824118256568909,
+      "learning_rate": 3.958521190261497e-05,
+      "loss": 1.5244,
+      "step": 440
+    },
+    {
+      "epoch": 0.11960492490867271,
+      "grad_norm": 0.4967150092124939,
+      "learning_rate": 3.976555455365194e-05,
+      "loss": 1.5273,
+      "step": 442
+    },
+    {
+      "epoch": 0.12014612366391557,
+      "grad_norm": 0.5089767575263977,
+      "learning_rate": 3.994589720468891e-05,
+      "loss": 1.5119,
+      "step": 444
+    },
+    {
+      "epoch": 0.12068732241915843,
+      "grad_norm": 0.5404312014579773,
+      "learning_rate": 4.0126239855725886e-05,
+      "loss": 1.5072,
+      "step": 446
+    },
+    {
+      "epoch": 0.1212285211744013,
+      "grad_norm": 0.5239550471305847,
+      "learning_rate": 4.0306582506762853e-05,
+      "loss": 1.5336,
+      "step": 448
+    },
+    {
+      "epoch": 0.12176971992964417,
+      "grad_norm": 0.4974781274795532,
+      "learning_rate": 4.048692515779982e-05,
+      "loss": 1.5225,
+      "step": 450
+    },
+    {
+      "epoch": 0.12231091868488703,
+      "grad_norm": 0.5363791584968567,
+      "learning_rate": 4.066726780883679e-05,
+      "loss": 1.5176,
+      "step": 452
+    },
+    {
+      "epoch": 0.12285211744012989,
+      "grad_norm": 0.5095157027244568,
+      "learning_rate": 4.084761045987376e-05,
+      "loss": 1.4936,
+      "step": 454
+    },
+    {
+      "epoch": 0.12339331619537275,
+      "grad_norm": 0.4920356869697571,
+      "learning_rate": 4.102795311091073e-05,
+      "loss": 1.5269,
+      "step": 456
+    },
+    {
+      "epoch": 0.12393451495061561,
+      "grad_norm": 0.4940793514251709,
+      "learning_rate": 4.1208295761947705e-05,
+      "loss": 1.5072,
+      "step": 458
+    },
+    {
+      "epoch": 0.12447571370585847,
+      "grad_norm": 0.4805227220058441,
+      "learning_rate": 4.138863841298467e-05,
+      "loss": 1.4987,
+      "step": 460
+    },
+    {
+      "epoch": 0.12501691246110133,
+      "grad_norm": 0.49683934450149536,
+      "learning_rate": 4.1568981064021646e-05,
+      "loss": 1.5008,
+      "step": 462
+    },
+    {
+      "epoch": 0.1255581112163442,
+      "grad_norm": 0.5283801555633545,
+      "learning_rate": 4.1749323715058614e-05,
+      "loss": 1.5177,
+      "step": 464
+    },
+    {
+      "epoch": 0.12609930997158705,
+      "grad_norm": 0.5395119190216064,
+      "learning_rate": 4.192966636609558e-05,
+      "loss": 1.5106,
+      "step": 466
+    },
+    {
+      "epoch": 0.12664050872682991,
+      "grad_norm": 0.5403693914413452,
+      "learning_rate": 4.211000901713255e-05,
+      "loss": 1.4854,
+      "step": 468
+    },
+    {
+      "epoch": 0.1271817074820728,
+      "grad_norm": 0.4690951406955719,
+      "learning_rate": 4.229035166816952e-05,
+      "loss": 1.5079,
+      "step": 470
+    },
+    {
+      "epoch": 0.12772290623731566,
+      "grad_norm": 0.5077293515205383,
+      "learning_rate": 4.24706943192065e-05,
+      "loss": 1.4953,
+      "step": 472
+    },
+    {
+      "epoch": 0.12826410499255853,
+      "grad_norm": 0.440019816160202,
+      "learning_rate": 4.2651036970243465e-05,
+      "loss": 1.4864,
+      "step": 474
+    },
+    {
+      "epoch": 0.1288053037478014,
+      "grad_norm": 0.48672759532928467,
+      "learning_rate": 4.283137962128044e-05,
+      "loss": 1.5205,
+      "step": 476
+    },
+    {
+      "epoch": 0.12934650250304425,
+      "grad_norm": 0.4732811450958252,
+      "learning_rate": 4.301172227231741e-05,
+      "loss": 1.4998,
+      "step": 478
+    },
+    {
+      "epoch": 0.1298877012582871,
+      "grad_norm": 0.46713048219680786,
+      "learning_rate": 4.3192064923354374e-05,
+      "loss": 1.4893,
+      "step": 480
+    },
+    {
+      "epoch": 0.13042890001352997,
+      "grad_norm": 0.502356231212616,
+      "learning_rate": 4.337240757439134e-05,
+      "loss": 1.5125,
+      "step": 482
+    },
+    {
+      "epoch": 0.13097009876877283,
+      "grad_norm": 0.45067864656448364,
+      "learning_rate": 4.3552750225428316e-05,
+      "loss": 1.4978,
+      "step": 484
+    },
+    {
+      "epoch": 0.1315112975240157,
+      "grad_norm": 0.46964120864868164,
+      "learning_rate": 4.373309287646529e-05,
+      "loss": 1.5006,
+      "step": 486
+    },
+    {
+      "epoch": 0.13205249627925855,
+      "grad_norm": 0.47723180055618286,
+      "learning_rate": 4.391343552750226e-05,
+      "loss": 1.513,
+      "step": 488
+    },
+    {
+      "epoch": 0.1325936950345014,
+      "grad_norm": 0.5100542306900024,
+      "learning_rate": 4.4093778178539225e-05,
+      "loss": 1.5279,
+      "step": 490
+    },
+    {
+      "epoch": 0.13313489378974427,
+      "grad_norm": 0.5344257354736328,
+      "learning_rate": 4.42741208295762e-05,
+      "loss": 1.5193,
+      "step": 492
+    },
+    {
+      "epoch": 0.13367609254498714,
+      "grad_norm": 0.5867893695831299,
+      "learning_rate": 4.445446348061317e-05,
+      "loss": 1.512,
+      "step": 494
+    },
+    {
+      "epoch": 0.13421729130023,
+      "grad_norm": 0.7811394929885864,
+      "learning_rate": 4.4634806131650134e-05,
+      "loss": 1.5038,
+      "step": 496
+    },
+    {
+      "epoch": 0.13475849005547288,
+      "grad_norm": 0.8505339622497559,
+      "learning_rate": 4.48151487826871e-05,
+      "loss": 1.5169,
+      "step": 498
+    },
+    {
+      "epoch": 0.13529968881071575,
+      "grad_norm": 0.6337641477584839,
+      "learning_rate": 4.4995491433724076e-05,
+      "loss": 1.4951,
+      "step": 500
+    },
+    {
+      "epoch": 0.1358408875659586,
+      "grad_norm": 0.7979961633682251,
+      "learning_rate": 4.517583408476105e-05,
+      "loss": 1.5031,
+      "step": 502
+    },
+    {
+      "epoch": 0.13638208632120147,
+      "grad_norm": 0.6946894526481628,
+      "learning_rate": 4.535617673579802e-05,
+      "loss": 1.501,
+      "step": 504
+    },
+    {
+      "epoch": 0.13692328507644433,
+      "grad_norm": 0.6830259561538696,
+      "learning_rate": 4.5536519386834986e-05,
+      "loss": 1.4896,
+      "step": 506
+    },
+    {
+      "epoch": 0.1374644838316872,
+      "grad_norm": 0.5908662676811218,
+      "learning_rate": 4.571686203787196e-05,
+      "loss": 1.4992,
+      "step": 508
+    },
+    {
+      "epoch": 0.13800568258693005,
+      "grad_norm": 0.7655865550041199,
+      "learning_rate": 4.589720468890893e-05,
+      "loss": 1.4911,
+      "step": 510
+    },
+    {
+      "epoch": 0.1385468813421729,
+      "grad_norm": 0.5924785733222961,
+      "learning_rate": 4.6077547339945895e-05,
+      "loss": 1.4719,
+      "step": 512
+    },
+    {
+      "epoch": 0.13908808009741577,
+      "grad_norm": 0.6654263138771057,
+      "learning_rate": 4.625788999098287e-05,
+      "loss": 1.5109,
+      "step": 514
+    },
+    {
+      "epoch": 0.13962927885265863,
+      "grad_norm": 0.5296297073364258,
+      "learning_rate": 4.6438232642019843e-05,
+      "loss": 1.4934,
+      "step": 516
+    },
+    {
+      "epoch": 0.1401704776079015,
+      "grad_norm": 0.5698690414428711,
+      "learning_rate": 4.661857529305681e-05,
+      "loss": 1.4954,
+      "step": 518
+    },
+    {
+      "epoch": 0.14071167636314436,
+      "grad_norm": 0.5790325403213501,
+      "learning_rate": 4.679891794409378e-05,
+      "loss": 1.4673,
+      "step": 520
+    },
+    {
+      "epoch": 0.14125287511838722,
+      "grad_norm": 0.551480770111084,
+      "learning_rate": 4.697926059513075e-05,
+      "loss": 1.476,
+      "step": 522
+    },
+    {
+      "epoch": 0.14179407387363008,
+      "grad_norm": 0.5201780796051025,
+      "learning_rate": 4.715960324616772e-05,
+      "loss": 1.4701,
+      "step": 524
+    },
+    {
+      "epoch": 0.14233527262887297,
+      "grad_norm": 0.46442562341690063,
+      "learning_rate": 4.733994589720469e-05,
+      "loss": 1.4831,
+      "step": 526
+    },
+    {
+      "epoch": 0.14287647138411583,
+      "grad_norm": 0.5558522939682007,
+      "learning_rate": 4.752028854824166e-05,
+      "loss": 1.4729,
+      "step": 528
+    },
+    {
+      "epoch": 0.1434176701393587,
+      "grad_norm": 0.48511791229248047,
+      "learning_rate": 4.7700631199278636e-05,
+      "loss": 1.4742,
+      "step": 530
+    },
+    {
+      "epoch": 0.14395886889460155,
+      "grad_norm": 0.5244829058647156,
+      "learning_rate": 4.7880973850315604e-05,
+      "loss": 1.4928,
+      "step": 532
+    },
+    {
+      "epoch": 0.1445000676498444,
+      "grad_norm": 0.48878946900367737,
+      "learning_rate": 4.806131650135257e-05,
+      "loss": 1.4921,
+      "step": 534
+    },
+    {
+      "epoch": 0.14504126640508727,
+      "grad_norm": 0.5348760485649109,
+      "learning_rate": 4.824165915238954e-05,
+      "loss": 1.4917,
+      "step": 536
+    },
+    {
+      "epoch": 0.14558246516033013,
+      "grad_norm": 0.5444923639297485,
+      "learning_rate": 4.842200180342651e-05,
+      "loss": 1.4546,
+      "step": 538
+    },
+    {
+      "epoch": 0.146123663915573,
+      "grad_norm": 0.494761198759079,
+      "learning_rate": 4.860234445446348e-05,
+      "loss": 1.4751,
+      "step": 540
+    },
+    {
+      "epoch": 0.14666486267081585,
+      "grad_norm": 0.4921441674232483,
+      "learning_rate": 4.8782687105500455e-05,
+      "loss": 1.4767,
+      "step": 542
+    },
+    {
+      "epoch": 0.14720606142605872,
+      "grad_norm": 0.48382577300071716,
+      "learning_rate": 4.896302975653742e-05,
+      "loss": 1.485,
+      "step": 544
+    },
+    {
+      "epoch": 0.14774726018130158,
+      "grad_norm": 0.4616708755493164,
+      "learning_rate": 4.9143372407574397e-05,
+      "loss": 1.4732,
+      "step": 546
+    },
+    {
+      "epoch": 0.14828845893654444,
+      "grad_norm": 0.5030043125152588,
+      "learning_rate": 4.9323715058611364e-05,
+      "loss": 1.4799,
+      "step": 548
+    },
+    {
+      "epoch": 0.1488296576917873,
+      "grad_norm": 0.467230886220932,
+      "learning_rate": 4.950405770964833e-05,
+      "loss": 1.4594,
+      "step": 550
+    },
+    {
+      "epoch": 0.14937085644703016,
+      "grad_norm": 0.42864304780960083,
+      "learning_rate": 4.9684400360685306e-05,
+      "loss": 1.4748,
+      "step": 552
+    },
+    {
+      "epoch": 0.14991205520227305,
+      "grad_norm": 0.43733683228492737,
+      "learning_rate": 4.986474301172227e-05,
+      "loss": 1.462,
+      "step": 554
+    },
+    {
+      "epoch": 0.1504532539575159,
+      "grad_norm": 0.45550286769866943,
+      "learning_rate": 5.004508566275925e-05,
+      "loss": 1.475,
+      "step": 556
+    },
+    {
+      "epoch": 0.15099445271275877,
+      "grad_norm": 0.44999995827674866,
+      "learning_rate": 5.022542831379622e-05,
+      "loss": 1.4794,
+      "step": 558
+    },
+    {
+      "epoch": 0.15153565146800163,
+      "grad_norm": 0.5035279989242554,
+      "learning_rate": 5.040577096483319e-05,
+      "loss": 1.471,
+      "step": 560
+    },
+    {
+      "epoch": 0.1520768502232445,
+      "grad_norm": 0.44605591893196106,
+      "learning_rate": 5.058611361587016e-05,
+      "loss": 1.4461,
+      "step": 562
+    },
+    {
+      "epoch": 0.15261804897848735,
+      "grad_norm": 0.5482723712921143,
+      "learning_rate": 5.0766456266907124e-05,
+      "loss": 1.4597,
+      "step": 564
+    },
+    {
+      "epoch": 0.1531592477337302,
+      "grad_norm": 0.5323627591133118,
+      "learning_rate": 5.094679891794409e-05,
+      "loss": 1.4743,
+      "step": 566
+    },
+    {
+      "epoch": 0.15370044648897307,
+      "grad_norm": 0.5289944410324097,
+      "learning_rate": 5.1127141568981066e-05,
+      "loss": 1.5,
+      "step": 568
+    },
+    {
+      "epoch": 0.15424164524421594,
+      "grad_norm": 0.5446243286132812,
+      "learning_rate": 5.1307484220018034e-05,
+      "loss": 1.4751,
+      "step": 570
+    },
+    {
+      "epoch": 0.1547828439994588,
+      "grad_norm": 0.525830090045929,
+      "learning_rate": 5.1487826871055015e-05,
+      "loss": 1.4639,
+      "step": 572
+    },
+    {
+      "epoch": 0.15532404275470166,
+      "grad_norm": 0.48129191994667053,
+      "learning_rate": 5.166816952209198e-05,
+      "loss": 1.4652,
+      "step": 574
+    },
+    {
+      "epoch": 0.15586524150994452,
+      "grad_norm": 0.47915297746658325,
+      "learning_rate": 5.184851217312895e-05,
+      "loss": 1.4627,
+      "step": 576
+    },
+    {
+      "epoch": 0.15640644026518738,
+      "grad_norm": 0.5229325294494629,
+      "learning_rate": 5.202885482416592e-05,
+      "loss": 1.4525,
+      "step": 578
+    },
+    {
+      "epoch": 0.15694763902043024,
+      "grad_norm": 0.5452600121498108,
+      "learning_rate": 5.2209197475202885e-05,
+      "loss": 1.458,
+      "step": 580
+    },
+    {
+      "epoch": 0.15748883777567313,
+      "grad_norm": 0.427432656288147,
+      "learning_rate": 5.238954012623985e-05,
+      "loss": 1.4773,
+      "step": 582
+    },
+    {
+      "epoch": 0.158030036530916,
+      "grad_norm": 0.450712114572525,
+      "learning_rate": 5.2569882777276827e-05,
+      "loss": 1.469,
+      "step": 584
+    },
+    {
+      "epoch": 0.15857123528615885,
+      "grad_norm": 0.5500516891479492,
+      "learning_rate": 5.27502254283138e-05,
+      "loss": 1.4603,
+      "step": 586
+    },
+    {
+      "epoch": 0.1591124340414017,
+      "grad_norm": 0.457157164812088,
+      "learning_rate": 5.2930568079350775e-05,
+      "loss": 1.4785,
+      "step": 588
+    },
+    {
+      "epoch": 0.15965363279664457,
+      "grad_norm": 0.49750396609306335,
+      "learning_rate": 5.311091073038774e-05,
+      "loss": 1.4603,
+      "step": 590
+    },
+    {
+      "epoch": 0.16019483155188743,
+      "grad_norm": 0.5720525979995728,
+      "learning_rate": 5.329125338142471e-05,
+      "loss": 1.4753,
+      "step": 592
+    },
+    {
+      "epoch": 0.1607360303071303,
+      "grad_norm": 0.4425548315048218,
+      "learning_rate": 5.347159603246168e-05,
+      "loss": 1.462,
+      "step": 594
+    },
+    {
+      "epoch": 0.16127722906237316,
+      "grad_norm": 0.5064132809638977,
+      "learning_rate": 5.3651938683498645e-05,
+      "loss": 1.4596,
+      "step": 596
+    },
+    {
+      "epoch": 0.16181842781761602,
+      "grad_norm": 0.518460750579834,
+      "learning_rate": 5.383228133453562e-05,
+      "loss": 1.4763,
+      "step": 598
+    },
+    {
+      "epoch": 0.16235962657285888,
+      "grad_norm": 0.4613576829433441,
+      "learning_rate": 5.401262398557259e-05,
+      "loss": 1.4487,
+      "step": 600
+    },
+    {
+      "epoch": 0.16290082532810174,
+      "grad_norm": 0.7046213746070862,
+      "learning_rate": 5.419296663660957e-05,
+      "loss": 1.472,
+      "step": 602
+    },
+    {
+      "epoch": 0.1634420240833446,
+      "grad_norm": 0.6164196133613586,
+      "learning_rate": 5.4373309287646535e-05,
+      "loss": 1.4424,
+      "step": 604
+    },
+    {
+      "epoch": 0.16398322283858746,
+      "grad_norm": 0.5106020569801331,
+      "learning_rate": 5.45536519386835e-05,
+      "loss": 1.4567,
+      "step": 606
+    },
+    {
+      "epoch": 0.16452442159383032,
+      "grad_norm": 0.4291236400604248,
+      "learning_rate": 5.473399458972047e-05,
+      "loss": 1.4514,
+      "step": 608
+    },
+    {
+      "epoch": 0.16506562034907318,
+      "grad_norm": 0.46577414870262146,
+      "learning_rate": 5.491433724075744e-05,
+      "loss": 1.4408,
+      "step": 610
+    },
+    {
+      "epoch": 0.16560681910431607,
+      "grad_norm": 0.4729917049407959,
+      "learning_rate": 5.509467989179441e-05,
+      "loss": 1.4493,
+      "step": 612
+    },
+    {
+      "epoch": 0.16614801785955893,
+      "grad_norm": 0.4651925563812256,
+      "learning_rate": 5.527502254283138e-05,
+      "loss": 1.465,
+      "step": 614
+    },
+    {
+      "epoch": 0.1666892166148018,
+      "grad_norm": 0.4756859540939331,
+      "learning_rate": 5.545536519386835e-05,
+      "loss": 1.4641,
+      "step": 616
+    },
+    {
+      "epoch": 0.16723041537004465,
+      "grad_norm": 0.42555975914001465,
+      "learning_rate": 5.563570784490533e-05,
+      "loss": 1.4569,
+      "step": 618
+    },
+    {
+      "epoch": 0.16777161412528752,
+      "grad_norm": 0.5162522196769714,
+      "learning_rate": 5.5816050495942296e-05,
+      "loss": 1.4344,
+      "step": 620
+    },
+    {
+      "epoch": 0.16831281288053038,
+      "grad_norm": 0.5867063999176025,
+      "learning_rate": 5.599639314697926e-05,
+      "loss": 1.4647,
+      "step": 622
+    },
+    {
+      "epoch": 0.16885401163577324,
+      "grad_norm": 0.6629165410995483,
+      "learning_rate": 5.617673579801623e-05,
+      "loss": 1.473,
+      "step": 624
+    },
+    {
+      "epoch": 0.1693952103910161,
+      "grad_norm": 0.5905330777168274,
+      "learning_rate": 5.6357078449053205e-05,
+      "loss": 1.4459,
+      "step": 626
+    },
+    {
+      "epoch": 0.16993640914625896,
+      "grad_norm": 0.7457858324050903,
+      "learning_rate": 5.653742110009017e-05,
+      "loss": 1.4603,
+      "step": 628
+    },
+    {
+      "epoch": 0.17047760790150182,
+      "grad_norm": 0.5977684855461121,
+      "learning_rate": 5.671776375112714e-05,
+      "loss": 1.4621,
+      "step": 630
+    },
+    {
+      "epoch": 0.17101880665674468,
+      "grad_norm": 0.7097992897033691,
+      "learning_rate": 5.689810640216412e-05,
+      "loss": 1.4646,
+      "step": 632
+    },
+    {
+      "epoch": 0.17156000541198754,
+      "grad_norm": 0.5895450711250305,
+      "learning_rate": 5.707844905320109e-05,
+      "loss": 1.4338,
+      "step": 634
+    },
+    {
+      "epoch": 0.1721012041672304,
+      "grad_norm": 0.576877772808075,
+      "learning_rate": 5.7258791704238056e-05,
+      "loss": 1.4666,
+      "step": 636
+    },
+    {
+      "epoch": 0.17264240292247326,
+      "grad_norm": 0.541110098361969,
+      "learning_rate": 5.7439134355275024e-05,
+      "loss": 1.4624,
+      "step": 638
+    },
+    {
+      "epoch": 0.17318360167771615,
+      "grad_norm": 0.5172320604324341,
+      "learning_rate": 5.7619477006312e-05,
+      "loss": 1.473,
+      "step": 640
+    },
+    {
+      "epoch": 0.17372480043295901,
+      "grad_norm": 0.47511357069015503,
+      "learning_rate": 5.7799819657348965e-05,
+      "loss": 1.446,
+      "step": 642
+    },
+    {
+      "epoch": 0.17426599918820188,
+      "grad_norm": 0.48614808917045593,
+      "learning_rate": 5.798016230838593e-05,
+      "loss": 1.4394,
+      "step": 644
+    },
+    {
+      "epoch": 0.17480719794344474,
+      "grad_norm": 0.4435577094554901,
+      "learning_rate": 5.81605049594229e-05,
+      "loss": 1.43,
+      "step": 646
+    },
+    {
+      "epoch": 0.1753483966986876,
+      "grad_norm": 0.4458653926849365,
+      "learning_rate": 5.834084761045988e-05,
+      "loss": 1.46,
+      "step": 648
+    },
+    {
+      "epoch": 0.17588959545393046,
+      "grad_norm": 0.40675726532936096,
+      "learning_rate": 5.852119026149685e-05,
+      "loss": 1.4565,
+      "step": 650
+    },
+    {
+      "epoch": 0.17643079420917332,
+      "grad_norm": 0.4132504165172577,
+      "learning_rate": 5.8701532912533817e-05,
+      "loss": 1.4522,
+      "step": 652
+    },
+    {
+      "epoch": 0.17697199296441618,
+      "grad_norm": 0.40881386399269104,
+      "learning_rate": 5.888187556357079e-05,
+      "loss": 1.4232,
+      "step": 654
+    },
+    {
+      "epoch": 0.17751319171965904,
+      "grad_norm": 0.40527868270874023,
+      "learning_rate": 5.906221821460776e-05,
+      "loss": 1.441,
+      "step": 656
+    },
+    {
+      "epoch": 0.1780543904749019,
+      "grad_norm": 0.40227004885673523,
+      "learning_rate": 5.9242560865644726e-05,
+      "loss": 1.4259,
+      "step": 658
+    },
+    {
+      "epoch": 0.17859558923014476,
+      "grad_norm": 0.4043656289577484,
+      "learning_rate": 5.942290351668169e-05,
+      "loss": 1.4298,
+      "step": 660
+    },
+    {
+      "epoch": 0.17913678798538762,
+      "grad_norm": 0.4288482666015625,
+      "learning_rate": 5.9603246167718674e-05,
+      "loss": 1.4439,
+      "step": 662
+    },
+    {
+      "epoch": 0.17967798674063049,
+      "grad_norm": 0.4385060966014862,
+      "learning_rate": 5.978358881875564e-05,
+      "loss": 1.4237,
+      "step": 664
+    },
+    {
+      "epoch": 0.18021918549587335,
+      "grad_norm": 0.396980345249176,
+      "learning_rate": 5.996393146979261e-05,
+      "loss": 1.4174,
+      "step": 666
+    },
+    {
+      "epoch": 0.18076038425111624,
+      "grad_norm": 0.4060603678226471,
+      "learning_rate": 6.014427412082958e-05,
+      "loss": 1.4479,
+      "step": 668
+    },
+    {
+      "epoch": 0.1813015830063591,
+      "grad_norm": 0.4485025703907013,
+      "learning_rate": 6.032461677186655e-05,
+      "loss": 1.4493,
+      "step": 670
+    },
+    {
+      "epoch": 0.18184278176160196,
+      "grad_norm": 0.44034305214881897,
+      "learning_rate": 6.050495942290352e-05,
+      "loss": 1.4461,
+      "step": 672
+    },
+    {
+      "epoch": 0.18238398051684482,
+      "grad_norm": 0.418074369430542,
+      "learning_rate": 6.0685302073940486e-05,
+      "loss": 1.4287,
+      "step": 674
+    },
+    {
+      "epoch": 0.18292517927208768,
+      "grad_norm": 0.41937318444252014,
+      "learning_rate": 6.0865644724977454e-05,
+      "loss": 1.4338,
+      "step": 676
+    },
+    {
+      "epoch": 0.18346637802733054,
+      "grad_norm": 0.4103530943393707,
+      "learning_rate": 6.104598737601444e-05,
+      "loss": 1.4391,
+      "step": 678
+    },
+    {
+      "epoch": 0.1840075767825734,
+      "grad_norm": 0.4066039025783539,
+      "learning_rate": 6.122633002705141e-05,
+      "loss": 1.4357,
+      "step": 680
+    },
+    {
+      "epoch": 0.18454877553781626,
+      "grad_norm": 0.36903437972068787,
+      "learning_rate": 6.140667267808838e-05,
+      "loss": 1.4111,
+      "step": 682
+    },
+    {
+      "epoch": 0.18508997429305912,
+      "grad_norm": 0.37125757336616516,
+      "learning_rate": 6.158701532912534e-05,
+      "loss": 1.4233,
+      "step": 684
+    },
+    {
+      "epoch": 0.18563117304830198,
+      "grad_norm": 0.44102513790130615,
+      "learning_rate": 6.176735798016231e-05,
+      "loss": 1.4437,
+      "step": 686
+    },
+    {
+      "epoch": 0.18617237180354484,
+      "grad_norm": 0.4337277114391327,
+      "learning_rate": 6.194770063119928e-05,
+      "loss": 1.4425,
+      "step": 688
+    },
+    {
+      "epoch": 0.1867135705587877,
+      "grad_norm": 0.37394315004348755,
+      "learning_rate": 6.212804328223625e-05,
+      "loss": 1.4452,
+      "step": 690
+    },
+    {
+      "epoch": 0.18725476931403057,
+      "grad_norm": 0.41764944791793823,
+      "learning_rate": 6.230838593327321e-05,
+      "loss": 1.4535,
+      "step": 692
+    },
+    {
+      "epoch": 0.18779596806927343,
+      "grad_norm": 0.4214741289615631,
+      "learning_rate": 6.24887285843102e-05,
+      "loss": 1.4391,
+      "step": 694
+    },
+    {
+      "epoch": 0.18833716682451632,
+      "grad_norm": 0.4159027338027954,
+      "learning_rate": 6.266907123534716e-05,
+      "loss": 1.4197,
+      "step": 696
+    },
+    {
+      "epoch": 0.18887836557975918,
+      "grad_norm": 0.38865673542022705,
+      "learning_rate": 6.284941388638413e-05,
+      "loss": 1.4329,
+      "step": 698
+    },
+    {
+      "epoch": 0.18941956433500204,
+      "grad_norm": 0.43646490573883057,
+      "learning_rate": 6.30297565374211e-05,
+      "loss": 1.4147,
+      "step": 700
+    },
+    {
+      "epoch": 0.1899607630902449,
+      "grad_norm": 0.41997334361076355,
+      "learning_rate": 6.321009918845807e-05,
+      "loss": 1.4275,
+      "step": 702
+    },
+    {
+      "epoch": 0.19050196184548776,
+      "grad_norm": 0.38556602597236633,
+      "learning_rate": 6.339044183949505e-05,
+      "loss": 1.4258,
+      "step": 704
+    },
+    {
+      "epoch": 0.19104316060073062,
+      "grad_norm": 0.42955082654953003,
+      "learning_rate": 6.357078449053201e-05,
+      "loss": 1.4201,
+      "step": 706
+    },
+    {
+      "epoch": 0.19158435935597348,
+      "grad_norm": 0.3844427764415741,
+      "learning_rate": 6.3751127141569e-05,
+      "loss": 1.4448,
+      "step": 708
+    },
+    {
+      "epoch": 0.19212555811121634,
+      "grad_norm": 0.4312956929206848,
+      "learning_rate": 6.393146979260596e-05,
+      "loss": 1.4051,
+      "step": 710
+    },
+    {
+      "epoch": 0.1926667568664592,
+      "grad_norm": 0.4556865394115448,
+      "learning_rate": 6.411181244364293e-05,
+      "loss": 1.4305,
+      "step": 712
+    },
+    {
+      "epoch": 0.19320795562170207,
+      "grad_norm": 0.37053731083869934,
+      "learning_rate": 6.42921550946799e-05,
+      "loss": 1.4301,
+      "step": 714
+    },
+    {
+      "epoch": 0.19374915437694493,
+      "grad_norm": 0.3996010720729828,
+      "learning_rate": 6.447249774571686e-05,
+      "loss": 1.4282,
+      "step": 716
+    },
+    {
+      "epoch": 0.1942903531321878,
+      "grad_norm": 0.37610816955566406,
+      "learning_rate": 6.465284039675383e-05,
+      "loss": 1.4277,
+      "step": 718
+    },
+    {
+      "epoch": 0.19483155188743065,
+      "grad_norm": 0.3677166998386383,
+      "learning_rate": 6.48331830477908e-05,
+      "loss": 1.4029,
+      "step": 720
+    },
+    {
+      "epoch": 0.1953727506426735,
+      "grad_norm": 0.3841564357280731,
+      "learning_rate": 6.501352569882777e-05,
+      "loss": 1.4144,
+      "step": 722
+    },
+    {
+      "epoch": 0.1959139493979164,
+      "grad_norm": 0.3687719404697418,
+      "learning_rate": 6.519386834986475e-05,
+      "loss": 1.4079,
+      "step": 724
+    },
+    {
+      "epoch": 0.19645514815315926,
+      "grad_norm": 0.38350847363471985,
+      "learning_rate": 6.537421100090172e-05,
+      "loss": 1.4269,
+      "step": 726
+    },
+    {
+      "epoch": 0.19699634690840212,
+      "grad_norm": 0.39060813188552856,
+      "learning_rate": 6.555455365193868e-05,
+      "loss": 1.4265,
+      "step": 728
+    },
+    {
+      "epoch": 0.19753754566364498,
+      "grad_norm": 0.36068469285964966,
+      "learning_rate": 6.573489630297565e-05,
+      "loss": 1.4325,
+      "step": 730
+    },
+    {
+      "epoch": 0.19807874441888784,
+      "grad_norm": 0.41185086965560913,
+      "learning_rate": 6.591523895401263e-05,
+      "loss": 1.4348,
+      "step": 732
+    },
+    {
+      "epoch": 0.1986199431741307,
+      "grad_norm": 0.4441224932670593,
+      "learning_rate": 6.60955816050496e-05,
+      "loss": 1.4103,
+      "step": 734
+    },
+    {
+      "epoch": 0.19916114192937356,
+      "grad_norm": 0.3727317452430725,
+      "learning_rate": 6.627592425608657e-05,
+      "loss": 1.4188,
+      "step": 736
+    },
+    {
+      "epoch": 0.19970234068461643,
+      "grad_norm": 0.394972562789917,
+      "learning_rate": 6.645626690712355e-05,
+      "loss": 1.4095,
+      "step": 738
+    },
+    {
+      "epoch": 0.20024353943985929,
+      "grad_norm": 0.40716880559921265,
+      "learning_rate": 6.663660955816052e-05,
+      "loss": 1.4127,
+      "step": 740
+    },
+    {
+      "epoch": 0.20078473819510215,
+      "grad_norm": 0.4156644344329834,
+      "learning_rate": 6.681695220919748e-05,
+      "loss": 1.4189,
+      "step": 742
+    },
+    {
+      "epoch": 0.201325936950345,
+      "grad_norm": 0.3787958323955536,
+      "learning_rate": 6.699729486023445e-05,
+      "loss": 1.4221,
+      "step": 744
+    },
+    {
+      "epoch": 0.20186713570558787,
+      "grad_norm": 0.42427608370780945,
+      "learning_rate": 6.717763751127142e-05,
+      "loss": 1.4192,
+      "step": 746
+    },
+    {
+      "epoch": 0.20240833446083073,
+      "grad_norm": 0.4778277277946472,
+      "learning_rate": 6.735798016230839e-05,
+      "loss": 1.4024,
+      "step": 748
+    },
+    {
+      "epoch": 0.2029495332160736,
+      "grad_norm": 0.44801151752471924,
+      "learning_rate": 6.753832281334535e-05,
+      "loss": 1.4222,
+      "step": 750
+    },
+    {
+      "epoch": 0.20349073197131648,
+      "grad_norm": 0.46737611293792725,
+      "learning_rate": 6.771866546438232e-05,
+      "loss": 1.4117,
+      "step": 752
+    },
+    {
+      "epoch": 0.20403193072655934,
+      "grad_norm": 0.4184872806072235,
+      "learning_rate": 6.78990081154193e-05,
+      "loss": 1.4066,
+      "step": 754
+    },
+    {
+      "epoch": 0.2045731294818022,
+      "grad_norm": 0.40458211302757263,
+      "learning_rate": 6.807935076645627e-05,
+      "loss": 1.4274,
+      "step": 756
+    },
+    {
+      "epoch": 0.20511432823704506,
+      "grad_norm": 0.43926185369491577,
+      "learning_rate": 6.825969341749324e-05,
+      "loss": 1.4231,
+      "step": 758
+    },
+    {
+      "epoch": 0.20565552699228792,
+      "grad_norm": 0.4434867203235626,
+      "learning_rate": 6.844003606853022e-05,
+      "loss": 1.4121,
+      "step": 760
+    },
+    {
+      "epoch": 0.20619672574753078,
+      "grad_norm": 0.4500143826007843,
+      "learning_rate": 6.862037871956719e-05,
+      "loss": 1.4179,
+      "step": 762
+    },
+    {
+      "epoch": 0.20673792450277365,
+      "grad_norm": 0.45456650853157043,
+      "learning_rate": 6.880072137060415e-05,
+      "loss": 1.3912,
+      "step": 764
+    },
+    {
+      "epoch": 0.2072791232580165,
+      "grad_norm": 0.4214187264442444,
+      "learning_rate": 6.898106402164112e-05,
+      "loss": 1.3962,
+      "step": 766
+    },
+    {
+      "epoch": 0.20782032201325937,
+      "grad_norm": 0.427682101726532,
+      "learning_rate": 6.916140667267809e-05,
+      "loss": 1.4316,
+      "step": 768
+    },
+    {
+      "epoch": 0.20836152076850223,
+      "grad_norm": 0.44491469860076904,
+      "learning_rate": 6.934174932371507e-05,
+      "loss": 1.4218,
+      "step": 770
+    },
+    {
+      "epoch": 0.2089027195237451,
+      "grad_norm": 0.42736080288887024,
+      "learning_rate": 6.952209197475204e-05,
+      "loss": 1.3931,
+      "step": 772
+    },
+    {
+      "epoch": 0.20944391827898795,
+      "grad_norm": 0.4041571021080017,
+      "learning_rate": 6.9702434625789e-05,
+      "loss": 1.4201,
+      "step": 774
+    },
+    {
+      "epoch": 0.2099851170342308,
+      "grad_norm": 0.4250961244106293,
+      "learning_rate": 6.988277727682597e-05,
+      "loss": 1.4299,
+      "step": 776
+    },
+    {
+      "epoch": 0.21052631578947367,
+      "grad_norm": 0.4335261881351471,
+      "learning_rate": 7.006311992786294e-05,
+      "loss": 1.4125,
+      "step": 778
+    },
+    {
+      "epoch": 0.21106751454471653,
+      "grad_norm": 0.42000851035118103,
+      "learning_rate": 7.02434625788999e-05,
+      "loss": 1.3969,
+      "step": 780
+    },
+    {
+      "epoch": 0.21160871329995942,
+      "grad_norm": 0.38111838698387146,
+      "learning_rate": 7.042380522993687e-05,
+      "loss": 1.3795,
+      "step": 782
+    },
+    {
+      "epoch": 0.21214991205520228,
+      "grad_norm": 0.38366812467575073,
+      "learning_rate": 7.060414788097385e-05,
+      "loss": 1.4041,
+      "step": 784
+    },
+    {
+      "epoch": 0.21269111081044514,
+      "grad_norm": 0.4334602355957031,
+      "learning_rate": 7.078449053201082e-05,
+      "loss": 1.415,
+      "step": 786
+    },
+    {
+      "epoch": 0.213232309565688,
+      "grad_norm": 0.40296411514282227,
+      "learning_rate": 7.096483318304779e-05,
+      "loss": 1.4052,
+      "step": 788
+    },
+    {
+      "epoch": 0.21377350832093087,
+      "grad_norm": 0.4197232723236084,
+      "learning_rate": 7.114517583408477e-05,
+      "loss": 1.4205,
+      "step": 790
+    },
+    {
+      "epoch": 0.21431470707617373,
+      "grad_norm": 0.40287715196609497,
+      "learning_rate": 7.132551848512174e-05,
+      "loss": 1.4047,
+      "step": 792
+    },
+    {
+      "epoch": 0.2148559058314166,
+      "grad_norm": 0.37324196100234985,
+      "learning_rate": 7.15058611361587e-05,
+      "loss": 1.4398,
+      "step": 794
+    },
+    {
+      "epoch": 0.21539710458665945,
+      "grad_norm": 0.4409985840320587,
+      "learning_rate": 7.168620378719567e-05,
+      "loss": 1.3873,
+      "step": 796
+    },
+    {
+      "epoch": 0.2159383033419023,
+      "grad_norm": 0.41441893577575684,
+      "learning_rate": 7.186654643823264e-05,
+      "loss": 1.4174,
+      "step": 798
+    },
+    {
+      "epoch": 0.21647950209714517,
+      "grad_norm": 0.4271719455718994,
+      "learning_rate": 7.204688908926962e-05,
+      "loss": 1.3987,
+      "step": 800
+    },
+    {
+      "epoch": 0.21702070085238803,
+      "grad_norm": 0.4969992935657501,
+      "learning_rate": 7.222723174030659e-05,
+      "loss": 1.4049,
+      "step": 802
+    },
+    {
+      "epoch": 0.2175618996076309,
+      "grad_norm": 0.45711180567741394,
+      "learning_rate": 7.240757439134356e-05,
+      "loss": 1.4061,
+      "step": 804
+    },
+    {
+      "epoch": 0.21810309836287375,
+      "grad_norm": 0.4479979872703552,
+      "learning_rate": 7.258791704238052e-05,
+      "loss": 1.4049,
+      "step": 806
+    },
+    {
+      "epoch": 0.21864429711811662,
+      "grad_norm": 0.4708006978034973,
+      "learning_rate": 7.276825969341749e-05,
+      "loss": 1.3971,
+      "step": 808
+    },
+    {
+      "epoch": 0.2191854958733595,
+      "grad_norm": 0.4387456774711609,
+      "learning_rate": 7.294860234445446e-05,
+      "loss": 1.4272,
+      "step": 810
+    },
+    {
+      "epoch": 0.21972669462860236,
+      "grad_norm": 0.5285756587982178,
+      "learning_rate": 7.312894499549143e-05,
+      "loss": 1.3902,
+      "step": 812
+    },
+    {
+      "epoch": 0.22026789338384523,
+      "grad_norm": 0.5111876726150513,
+      "learning_rate": 7.330928764652841e-05,
+      "loss": 1.4176,
+      "step": 814
+    },
+    {
+      "epoch": 0.2208090921390881,
+      "grad_norm": 0.4643821716308594,
+      "learning_rate": 7.348963029756538e-05,
+      "loss": 1.4216,
+      "step": 816
+    },
+    {
+      "epoch": 0.22135029089433095,
+      "grad_norm": 0.5162214040756226,
+      "learning_rate": 7.366997294860236e-05,
+      "loss": 1.4025,
+      "step": 818
+    },
+    {
+      "epoch": 0.2218914896495738,
+      "grad_norm": 0.4296860992908478,
+      "learning_rate": 7.385031559963932e-05,
+      "loss": 1.3919,
+      "step": 820
+    },
+    {
+      "epoch": 0.22243268840481667,
+      "grad_norm": 0.4449775815010071,
+      "learning_rate": 7.403065825067629e-05,
+      "loss": 1.4002,
+      "step": 822
+    },
+    {
+      "epoch": 0.22297388716005953,
+      "grad_norm": 0.39713212847709656,
+      "learning_rate": 7.421100090171326e-05,
+      "loss": 1.4012,
+      "step": 824
+    },
+    {
+      "epoch": 0.2235150859153024,
+      "grad_norm": 0.41655346751213074,
+      "learning_rate": 7.439134355275023e-05,
+      "loss": 1.4155,
+      "step": 826
+    },
+    {
+      "epoch": 0.22405628467054525,
+      "grad_norm": 0.3751365542411804,
+      "learning_rate": 7.45716862037872e-05,
+      "loss": 1.4021,
+      "step": 828
+    },
+    {
+      "epoch": 0.2245974834257881,
+      "grad_norm": 0.41483408212661743,
+      "learning_rate": 7.475202885482417e-05,
+      "loss": 1.4207,
+      "step": 830
+    },
+    {
+      "epoch": 0.22513868218103097,
+      "grad_norm": 0.397360235452652,
+      "learning_rate": 7.493237150586114e-05,
+      "loss": 1.392,
+      "step": 832
+    },
+    {
+      "epoch": 0.22567988093627384,
+      "grad_norm": 0.3874877691268921,
+      "learning_rate": 7.511271415689811e-05,
+      "loss": 1.4143,
+      "step": 834
+    },
+    {
+      "epoch": 0.2262210796915167,
+      "grad_norm": 0.4382254481315613,
+      "learning_rate": 7.529305680793508e-05,
+      "loss": 1.4109,
+      "step": 836
+    },
+    {
+      "epoch": 0.22676227844675959,
+      "grad_norm": 0.3728530704975128,
+      "learning_rate": 7.547339945897204e-05,
+      "loss": 1.4215,
+      "step": 838
+    },
+    {
+      "epoch": 0.22730347720200245,
+      "grad_norm": 0.41155338287353516,
+      "learning_rate": 7.565374211000901e-05,
+      "loss": 1.3963,
+      "step": 840
+    },
+    {
+      "epoch": 0.2278446759572453,
+      "grad_norm": 0.3550320267677307,
+      "learning_rate": 7.5834084761046e-05,
+      "loss": 1.3998,
+      "step": 842
+    },
+    {
+      "epoch": 0.22838587471248817,
+      "grad_norm": 0.3858035206794739,
+      "learning_rate": 7.601442741208296e-05,
+      "loss": 1.387,
+      "step": 844
+    },
+    {
+      "epoch": 0.22892707346773103,
+      "grad_norm": 0.38636457920074463,
+      "learning_rate": 7.619477006311994e-05,
+      "loss": 1.387,
+      "step": 846
+    },
+    {
+      "epoch": 0.2294682722229739,
+      "grad_norm": 0.41915518045425415,
+      "learning_rate": 7.637511271415691e-05,
+      "loss": 1.3917,
+      "step": 848
+    },
+    {
+      "epoch": 0.23000947097821675,
+      "grad_norm": 0.35796865820884705,
+      "learning_rate": 7.655545536519388e-05,
+      "loss": 1.406,
+      "step": 850
+    },
+    {
+      "epoch": 0.2305506697334596,
+      "grad_norm": 0.35221853852272034,
+      "learning_rate": 7.673579801623084e-05,
+      "loss": 1.3892,
+      "step": 852
+    },
+    {
+      "epoch": 0.23109186848870247,
+      "grad_norm": 0.3815077245235443,
+      "learning_rate": 7.691614066726781e-05,
+      "loss": 1.3845,
+      "step": 854
+    },
+    {
+      "epoch": 0.23163306724394533,
+      "grad_norm": 0.3554491400718689,
+      "learning_rate": 7.709648331830478e-05,
+      "loss": 1.3644,
+      "step": 856
+    },
+    {
+      "epoch": 0.2321742659991882,
+      "grad_norm": 0.3762814998626709,
+      "learning_rate": 7.727682596934175e-05,
+      "loss": 1.3976,
+      "step": 858
+    },
+    {
+      "epoch": 0.23271546475443106,
+      "grad_norm": 0.34575173258781433,
+      "learning_rate": 7.745716862037873e-05,
+      "loss": 1.3925,
+      "step": 860
+    },
+    {
+      "epoch": 0.23325666350967392,
+      "grad_norm": 0.37864556908607483,
+      "learning_rate": 7.76375112714157e-05,
+      "loss": 1.3993,
+      "step": 862
+    },
+    {
+      "epoch": 0.23379786226491678,
+      "grad_norm": 0.34448474645614624,
+      "learning_rate": 7.781785392245266e-05,
+      "loss": 1.3855,
+      "step": 864
+    },
+    {
+      "epoch": 0.23433906102015967,
+      "grad_norm": 0.40932390093803406,
+      "learning_rate": 7.799819657348963e-05,
+      "loss": 1.395,
+      "step": 866
+    },
+    {
+      "epoch": 0.23488025977540253,
+      "grad_norm": 0.3737650513648987,
+      "learning_rate": 7.81785392245266e-05,
+      "loss": 1.3918,
+      "step": 868
+    },
+    {
+      "epoch": 0.2354214585306454,
+      "grad_norm": 0.42988118529319763,
+      "learning_rate": 7.835888187556357e-05,
+      "loss": 1.3837,
+      "step": 870
+    },
+    {
+      "epoch": 0.23596265728588825,
+      "grad_norm": 0.3865496814250946,
+      "learning_rate": 7.853922452660055e-05,
+      "loss": 1.3976,
+      "step": 872
+    },
+    {
+      "epoch": 0.2365038560411311,
+      "grad_norm": 0.3682670295238495,
+      "learning_rate": 7.871956717763751e-05,
+      "loss": 1.3792,
+      "step": 874
+    },
+    {
+      "epoch": 0.23704505479637397,
+      "grad_norm": 0.4236462712287903,
+      "learning_rate": 7.88999098286745e-05,
+      "loss": 1.4032,
+      "step": 876
+    },
+    {
+      "epoch": 0.23758625355161683,
+      "grad_norm": 0.3742213249206543,
+      "learning_rate": 7.908025247971146e-05,
+      "loss": 1.3709,
+      "step": 878
+    },
+    {
+      "epoch": 0.2381274523068597,
+      "grad_norm": 0.38234424591064453,
+      "learning_rate": 7.926059513074843e-05,
+      "loss": 1.3862,
+      "step": 880
+    },
+    {
+      "epoch": 0.23866865106210255,
+      "grad_norm": 0.37414151430130005,
+      "learning_rate": 7.94409377817854e-05,
+      "loss": 1.3751,
+      "step": 882
+    },
+    {
+      "epoch": 0.23920984981734542,
+      "grad_norm": 0.3838132619857788,
+      "learning_rate": 7.962128043282237e-05,
+      "loss": 1.3805,
+      "step": 884
+    },
+    {
+      "epoch": 0.23975104857258828,
+      "grad_norm": 0.3818622827529907,
+      "learning_rate": 7.980162308385933e-05,
+      "loss": 1.3735,
+      "step": 886
+    },
+    {
+      "epoch": 0.24029224732783114,
+      "grad_norm": 0.38791927695274353,
+      "learning_rate": 7.99819657348963e-05,
+      "loss": 1.3958,
+      "step": 888
+    },
+    {
+      "epoch": 0.240833446083074,
+      "grad_norm": 0.4164978861808777,
+      "learning_rate": 8.016230838593328e-05,
+      "loss": 1.421,
+      "step": 890
+    },
+    {
+      "epoch": 0.24137464483831686,
+      "grad_norm": 0.3721414804458618,
+      "learning_rate": 8.034265103697025e-05,
+      "loss": 1.3977,
+      "step": 892
+    },
+    {
+      "epoch": 0.24191584359355975,
+      "grad_norm": 0.37698984146118164,
+      "learning_rate": 8.052299368800722e-05,
+      "loss": 1.3854,
+      "step": 894
+    },
+    {
+      "epoch": 0.2424570423488026,
+      "grad_norm": 0.3553116023540497,
+      "learning_rate": 8.070333633904418e-05,
+      "loss": 1.3925,
+      "step": 896
+    },
+    {
+      "epoch": 0.24299824110404547,
+      "grad_norm": 0.37809059023857117,
+      "learning_rate": 8.088367899008115e-05,
+      "loss": 1.368,
+      "step": 898
+    },
+    {
+      "epoch": 0.24353943985928833,
+      "grad_norm": 0.3835943043231964,
+      "learning_rate": 8.106402164111813e-05,
+      "loss": 1.3992,
+      "step": 900
+    },
+    {
+      "epoch": 0.2440806386145312,
+      "grad_norm": 0.4013379216194153,
+      "learning_rate": 8.12443642921551e-05,
+      "loss": 1.3912,
+      "step": 902
+    },
+    {
+      "epoch": 0.24462183736977405,
+      "grad_norm": 0.37845560908317566,
+      "learning_rate": 8.142470694319207e-05,
+      "loss": 1.3934,
+      "step": 904
+    },
+    {
+      "epoch": 0.24516303612501691,
+      "grad_norm": 0.39762255549430847,
+      "learning_rate": 8.160504959422905e-05,
+      "loss": 1.3782,
+      "step": 906
+    },
+    {
+      "epoch": 0.24570423488025978,
+      "grad_norm": 0.36652496457099915,
+      "learning_rate": 8.178539224526602e-05,
+      "loss": 1.3787,
+      "step": 908
+    },
+    {
+      "epoch": 0.24624543363550264,
+      "grad_norm": 0.39953047037124634,
+      "learning_rate": 8.196573489630298e-05,
+      "loss": 1.3752,
+      "step": 910
+    },
+    {
+      "epoch": 0.2467866323907455,
+      "grad_norm": 0.35875022411346436,
+      "learning_rate": 8.214607754733995e-05,
+      "loss": 1.3768,
+      "step": 912
+    },
+    {
+      "epoch": 0.24732783114598836,
+      "grad_norm": 0.3617067337036133,
+      "learning_rate": 8.232642019837692e-05,
+      "loss": 1.3859,
+      "step": 914
+    },
+    {
+      "epoch": 0.24786902990123122,
+      "grad_norm": 0.38250839710235596,
+      "learning_rate": 8.250676284941389e-05,
+      "loss": 1.3897,
+      "step": 916
+    },
+    {
+      "epoch": 0.24841022865647408,
+      "grad_norm": 0.3404116928577423,
+      "learning_rate": 8.268710550045085e-05,
+      "loss": 1.3933,
+      "step": 918
+    },
+    {
+      "epoch": 0.24895142741171694,
+      "grad_norm": 0.3547706604003906,
+      "learning_rate": 8.286744815148782e-05,
+      "loss": 1.3787,
+      "step": 920
+    },
+    {
+      "epoch": 0.2494926261669598,
+      "grad_norm": 0.32752275466918945,
+      "learning_rate": 8.30477908025248e-05,
+      "loss": 1.3905,
+      "step": 922
+    },
+    {
+      "epoch": 0.25003382492220266,
+      "grad_norm": 0.3413980007171631,
+      "learning_rate": 8.322813345356177e-05,
+      "loss": 1.385,
+      "step": 924
+    },
+    {
+      "epoch": 0.25057502367744555,
+      "grad_norm": 0.5574982762336731,
+      "learning_rate": 8.340847610459874e-05,
+      "loss": 1.3869,
+      "step": 926
+    },
+    {
+      "epoch": 0.2511162224326884,
+      "grad_norm": 0.41128844022750854,
+      "learning_rate": 8.358881875563572e-05,
+      "loss": 1.3583,
+      "step": 928
+    },
+    {
+      "epoch": 0.2516574211879313,
+      "grad_norm": 0.3476073145866394,
+      "learning_rate": 8.376916140667269e-05,
+      "loss": 1.3832,
+      "step": 930
+    },
+    {
+      "epoch": 0.2521986199431741,
+      "grad_norm": 0.34838998317718506,
+      "learning_rate": 8.394950405770965e-05,
+      "loss": 1.3748,
+      "step": 932
+    },
+    {
+      "epoch": 0.252739818698417,
+      "grad_norm": 0.3552824556827545,
+      "learning_rate": 8.412984670874662e-05,
+      "loss": 1.3936,
+      "step": 934
+    },
+    {
+      "epoch": 0.25328101745365983,
+      "grad_norm": 0.34918278455734253,
+      "learning_rate": 8.43101893597836e-05,
+      "loss": 1.3733,
+      "step": 936
+    },
+    {
+      "epoch": 0.2538222162089027,
+      "grad_norm": 0.431455135345459,
+      "learning_rate": 8.449053201082057e-05,
+      "loss": 1.3924,
+      "step": 938
+    },
+    {
+      "epoch": 0.2543634149641456,
+      "grad_norm": 0.37811046838760376,
+      "learning_rate": 8.467087466185754e-05,
+      "loss": 1.3861,
+      "step": 940
+    },
+    {
+      "epoch": 0.25490461371938844,
+      "grad_norm": 0.35659778118133545,
+      "learning_rate": 8.48512173128945e-05,
+      "loss": 1.3736,
+      "step": 942
+    },
+    {
+      "epoch": 0.25544581247463133,
+      "grad_norm": 0.4327319264411926,
+      "learning_rate": 8.503155996393147e-05,
+      "loss": 1.3883,
+      "step": 944
+    },
+    {
+      "epoch": 0.25598701122987416,
+      "grad_norm": 0.39134231209754944,
+      "learning_rate": 8.521190261496844e-05,
+      "loss": 1.3704,
+      "step": 946
+    },
+    {
+      "epoch": 0.25652820998511705,
+      "grad_norm": 0.39573270082473755,
+      "learning_rate": 8.53922452660054e-05,
+      "loss": 1.4047,
+      "step": 948
+    },
+    {
+      "epoch": 0.2570694087403599,
+      "grad_norm": 0.3299993872642517,
+      "learning_rate": 8.557258791704237e-05,
+      "loss": 1.3778,
+      "step": 950
+    },
+    {
+      "epoch": 0.2576106074956028,
+      "grad_norm": 0.3559456765651703,
+      "learning_rate": 8.575293056807936e-05,
+      "loss": 1.3794,
+      "step": 952
+    },
+    {
+      "epoch": 0.2581518062508456,
+      "grad_norm": 0.36347028613090515,
+      "learning_rate": 8.593327321911632e-05,
+      "loss": 1.3817,
+      "step": 954
+    },
+    {
+      "epoch": 0.2586930050060885,
+      "grad_norm": 0.39882585406303406,
+      "learning_rate": 8.611361587015329e-05,
+      "loss": 1.3565,
+      "step": 956
+    },
+    {
+      "epoch": 0.2592342037613313,
+      "grad_norm": 0.3932117223739624,
+      "learning_rate": 8.629395852119027e-05,
+      "loss": 1.396,
+      "step": 958
+    },
+    {
+      "epoch": 0.2597754025165742,
+      "grad_norm": 0.3526294231414795,
+      "learning_rate": 8.647430117222724e-05,
+      "loss": 1.3624,
+      "step": 960
+    },
+    {
+      "epoch": 0.26031660127181705,
+      "grad_norm": 0.3804738223552704,
+      "learning_rate": 8.66546438232642e-05,
+      "loss": 1.3616,
+      "step": 962
+    },
+    {
+      "epoch": 0.26085780002705994,
+      "grad_norm": 0.36557725071907043,
+      "learning_rate": 8.683498647430117e-05,
+      "loss": 1.3997,
+      "step": 964
+    },
+    {
+      "epoch": 0.2613989987823028,
+      "grad_norm": 0.3574380874633789,
+      "learning_rate": 8.701532912533815e-05,
+      "loss": 1.3901,
+      "step": 966
+    },
+    {
+      "epoch": 0.26194019753754566,
+      "grad_norm": 0.4025056064128876,
+      "learning_rate": 8.719567177637512e-05,
+      "loss": 1.3707,
+      "step": 968
+    },
+    {
+      "epoch": 0.26248139629278855,
+      "grad_norm": 0.3687063157558441,
+      "learning_rate": 8.737601442741209e-05,
+      "loss": 1.3679,
+      "step": 970
+    },
+    {
+      "epoch": 0.2630225950480314,
+      "grad_norm": 0.3697878420352936,
+      "learning_rate": 8.755635707844906e-05,
+      "loss": 1.3981,
+      "step": 972
+    },
+    {
+      "epoch": 0.26356379380327427,
+      "grad_norm": 0.34241798520088196,
+      "learning_rate": 8.773669972948602e-05,
+      "loss": 1.3728,
+      "step": 974
+    },
+    {
+      "epoch": 0.2641049925585171,
+      "grad_norm": 0.40002745389938354,
+      "learning_rate": 8.791704238052299e-05,
+      "loss": 1.3732,
+      "step": 976
+    },
+    {
+      "epoch": 0.26464619131376,
+      "grad_norm": 0.42943906784057617,
+      "learning_rate": 8.809738503155996e-05,
+      "loss": 1.3731,
+      "step": 978
+    },
+    {
+      "epoch": 0.2651873900690028,
+      "grad_norm": 0.37437063455581665,
+      "learning_rate": 8.827772768259693e-05,
+      "loss": 1.372,
+      "step": 980
+    },
+    {
+      "epoch": 0.2657285888242457,
+      "grad_norm": 0.3378891944885254,
+      "learning_rate": 8.845807033363391e-05,
+      "loss": 1.3777,
+      "step": 982
+    },
+    {
+      "epoch": 0.26626978757948855,
+      "grad_norm": 0.32884734869003296,
+      "learning_rate": 8.863841298467088e-05,
+      "loss": 1.3639,
+      "step": 984
+    },
+    {
+      "epoch": 0.26681098633473144,
+      "grad_norm": 0.3945903480052948,
+      "learning_rate": 8.881875563570786e-05,
+      "loss": 1.3722,
+      "step": 986
+    },
+    {
+      "epoch": 0.26735218508997427,
+      "grad_norm": 0.39569205045700073,
+      "learning_rate": 8.899909828674482e-05,
+      "loss": 1.376,
+      "step": 988
+    },
+    {
+      "epoch": 0.26789338384521716,
+      "grad_norm": 0.31659135222435,
+      "learning_rate": 8.917944093778179e-05,
+      "loss": 1.3807,
+      "step": 990
+    },
+    {
+      "epoch": 0.26843458260046,
+      "grad_norm": 0.44032666087150574,
+      "learning_rate": 8.935978358881876e-05,
+      "loss": 1.3986,
+      "step": 992
+    },
+    {
+      "epoch": 0.2689757813557029,
+      "grad_norm": 0.3445993661880493,
+      "learning_rate": 8.954012623985573e-05,
+      "loss": 1.3589,
+      "step": 994
+    },
+    {
+      "epoch": 0.26951698011094577,
+      "grad_norm": 0.3693557679653168,
+      "learning_rate": 8.97204688908927e-05,
+      "loss": 1.3593,
+      "step": 996
+    },
+    {
+      "epoch": 0.2700581788661886,
+      "grad_norm": 0.3965442478656769,
+      "learning_rate": 8.990081154192968e-05,
+      "loss": 1.3909,
+      "step": 998
+    },
+    {
+      "epoch": 0.2705993776214315,
+      "grad_norm": 0.4038390815258026,
+      "learning_rate": 9.008115419296664e-05,
+      "loss": 1.3629,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2711405763766743,
+      "grad_norm": 0.36394256353378296,
+      "learning_rate": 9.026149684400361e-05,
+      "loss": 1.3812,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2716817751319172,
+      "grad_norm": 0.4527181386947632,
+      "learning_rate": 9.044183949504058e-05,
+      "loss": 1.3692,
+      "step": 1004
+    },
+    {
+      "epoch": 0.27222297388716005,
+      "grad_norm": 0.37700143456459045,
+      "learning_rate": 9.062218214607755e-05,
+      "loss": 1.3652,
+      "step": 1006
+    },
+    {
+      "epoch": 0.27276417264240294,
+      "grad_norm": 0.45016244053840637,
+      "learning_rate": 9.080252479711451e-05,
+      "loss": 1.3657,
+      "step": 1008
+    },
+    {
+      "epoch": 0.27330537139764577,
+      "grad_norm": 0.42159709334373474,
+      "learning_rate": 9.09828674481515e-05,
+      "loss": 1.3702,
+      "step": 1010
+    },
+    {
+      "epoch": 0.27384657015288866,
+      "grad_norm": 0.3884572982788086,
+      "learning_rate": 9.116321009918846e-05,
+      "loss": 1.3535,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2743877689081315,
+      "grad_norm": 0.37507420778274536,
+      "learning_rate": 9.134355275022544e-05,
+      "loss": 1.3659,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2749289676633744,
+      "grad_norm": 0.35269656777381897,
+      "learning_rate": 9.152389540126241e-05,
+      "loss": 1.3623,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2754701664186172,
+      "grad_norm": 0.3543412387371063,
+      "learning_rate": 9.170423805229938e-05,
+      "loss": 1.3695,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2760113651738601,
+      "grad_norm": 0.3173674941062927,
+      "learning_rate": 9.188458070333635e-05,
+      "loss": 1.3572,
+      "step": 1020
+    },
+    {
+      "epoch": 0.276552563929103,
+      "grad_norm": 0.3729746341705322,
+      "learning_rate": 9.206492335437331e-05,
+      "loss": 1.3888,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2770937626843458,
+      "grad_norm": 0.33210429549217224,
+      "learning_rate": 9.224526600541028e-05,
+      "loss": 1.3395,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2776349614395887,
+      "grad_norm": 0.338366836309433,
+      "learning_rate": 9.242560865644725e-05,
+      "loss": 1.3498,
+      "step": 1026
+    },
+    {
+      "epoch": 0.27817616019483155,
+      "grad_norm": 0.3367864191532135,
+      "learning_rate": 9.260595130748423e-05,
+      "loss": 1.3548,
+      "step": 1028
+    },
+    {
+      "epoch": 0.27871735895007443,
+      "grad_norm": 0.40313002467155457,
+      "learning_rate": 9.27862939585212e-05,
+      "loss": 1.4059,
+      "step": 1030
+    },
+    {
+      "epoch": 0.27925855770531727,
+      "grad_norm": 0.3434394299983978,
+      "learning_rate": 9.296663660955816e-05,
+      "loss": 1.3522,
+      "step": 1032
+    },
+    {
+      "epoch": 0.27979975646056016,
+      "grad_norm": 0.35454580187797546,
+      "learning_rate": 9.314697926059513e-05,
+      "loss": 1.3838,
+      "step": 1034
+    },
+    {
+      "epoch": 0.280340955215803,
+      "grad_norm": 0.3280038833618164,
+      "learning_rate": 9.33273219116321e-05,
+      "loss": 1.3753,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2808821539710459,
+      "grad_norm": 0.4306875169277191,
+      "learning_rate": 9.350766456266907e-05,
+      "loss": 1.3807,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2814233527262887,
+      "grad_norm": 0.3500923812389374,
+      "learning_rate": 9.368800721370605e-05,
+      "loss": 1.36,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2819645514815316,
+      "grad_norm": 0.3702130913734436,
+      "learning_rate": 9.386834986474301e-05,
+      "loss": 1.3919,
+      "step": 1042
+    },
+    {
+      "epoch": 0.28250575023677443,
+      "grad_norm": 0.3651416599750519,
+      "learning_rate": 9.404869251578e-05,
+      "loss": 1.3805,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2830469489920173,
+      "grad_norm": 0.35927796363830566,
+      "learning_rate": 9.422903516681696e-05,
+      "loss": 1.3507,
+      "step": 1046
+    },
+    {
+      "epoch": 0.28358814774726016,
+      "grad_norm": 0.36750975251197815,
+      "learning_rate": 9.440937781785393e-05,
+      "loss": 1.3475,
+      "step": 1048
+    },
+    {
+      "epoch": 0.28412934650250304,
+      "grad_norm": 0.31946998834609985,
+      "learning_rate": 9.45897204688909e-05,
+      "loss": 1.3708,
+      "step": 1050
+    },
+    {
+      "epoch": 0.28467054525774593,
+      "grad_norm": 0.3447932302951813,
+      "learning_rate": 9.477006311992787e-05,
+      "loss": 1.3519,
+      "step": 1052
+    },
+    {
+      "epoch": 0.28521174401298877,
+      "grad_norm": 0.31405511498451233,
+      "learning_rate": 9.495040577096483e-05,
+      "loss": 1.3806,
+      "step": 1054
+    },
+    {
+      "epoch": 0.28575294276823165,
+      "grad_norm": 0.3198442757129669,
+      "learning_rate": 9.51307484220018e-05,
+      "loss": 1.368,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2862941415234745,
+      "grad_norm": 0.33328956365585327,
+      "learning_rate": 9.531109107303878e-05,
+      "loss": 1.3429,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2868353402787174,
+      "grad_norm": 0.29432907700538635,
+      "learning_rate": 9.549143372407575e-05,
+      "loss": 1.3698,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2873765390339602,
+      "grad_norm": 0.3468937575817108,
+      "learning_rate": 9.567177637511272e-05,
+      "loss": 1.356,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2879177377892031,
+      "grad_norm": 0.3619658350944519,
+      "learning_rate": 9.585211902614968e-05,
+      "loss": 1.3596,
+      "step": 1064
+    },
+    {
+      "epoch": 0.28845893654444593,
+      "grad_norm": 0.3384917378425598,
+      "learning_rate": 9.603246167718665e-05,
+      "loss": 1.3693,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2890001352996888,
+      "grad_norm": 0.3724029064178467,
+      "learning_rate": 9.621280432822363e-05,
+      "loss": 1.3639,
+      "step": 1068
+    },
+    {
+      "epoch": 0.28954133405493165,
+      "grad_norm": 0.7029115557670593,
+      "learning_rate": 9.63931469792606e-05,
+      "loss": 1.3557,
+      "step": 1070
+    },
+    {
+      "epoch": 0.29008253281017454,
+      "grad_norm": 0.5529230833053589,
+      "learning_rate": 9.657348963029757e-05,
+      "loss": 1.3657,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2906237315654174,
+      "grad_norm": 0.4254820644855499,
+      "learning_rate": 9.675383228133455e-05,
+      "loss": 1.3633,
+      "step": 1074
+    },
+    {
+      "epoch": 0.29116493032066026,
+      "grad_norm": 0.4930615723133087,
+      "learning_rate": 9.693417493237152e-05,
+      "loss": 1.3714,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2917061290759031,
+      "grad_norm": 0.4455857574939728,
+      "learning_rate": 9.711451758340848e-05,
+      "loss": 1.3615,
+      "step": 1078
+    },
+    {
+      "epoch": 0.292247327831146,
+      "grad_norm": 0.4171796441078186,
+      "learning_rate": 9.729486023444545e-05,
+      "loss": 1.3673,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2927885265863889,
+      "grad_norm": 0.37810683250427246,
+      "learning_rate": 9.747520288548242e-05,
+      "loss": 1.3683,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2933297253416317,
+      "grad_norm": 0.4057900905609131,
+      "learning_rate": 9.765554553651939e-05,
+      "loss": 1.3674,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2938709240968746,
+      "grad_norm": 0.40583640336990356,
+      "learning_rate": 9.783588818755635e-05,
+      "loss": 1.3566,
+      "step": 1086
+    },
+    {
+      "epoch": 0.29441212285211743,
+      "grad_norm": 0.39454150199890137,
+      "learning_rate": 9.801623083859334e-05,
+      "loss": 1.3611,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2949533216073603,
+      "grad_norm": 0.42229679226875305,
+      "learning_rate": 9.81965734896303e-05,
+      "loss": 1.3726,
+      "step": 1090
+    },
+    {
+      "epoch": 0.29549452036260315,
+      "grad_norm": 0.3274170160293579,
+      "learning_rate": 9.837691614066727e-05,
+      "loss": 1.3375,
+      "step": 1092
+    },
+    {
+      "epoch": 0.29603571911784604,
+      "grad_norm": 0.40999388694763184,
+      "learning_rate": 9.855725879170424e-05,
+      "loss": 1.3548,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2965769178730889,
+      "grad_norm": 0.33515796065330505,
+      "learning_rate": 9.873760144274122e-05,
+      "loss": 1.3903,
+      "step": 1096
+    },
+    {
+      "epoch": 0.29711811662833176,
+      "grad_norm": 0.3834095597267151,
+      "learning_rate": 9.891794409377819e-05,
+      "loss": 1.3653,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2976593153835746,
+      "grad_norm": 0.34850651025772095,
+      "learning_rate": 9.909828674481515e-05,
+      "loss": 1.3573,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2982005141388175,
+      "grad_norm": 0.3811749815940857,
+      "learning_rate": 9.927862939585212e-05,
+      "loss": 1.3843,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2987417128940603,
+      "grad_norm": 0.3308597803115845,
+      "learning_rate": 9.94589720468891e-05,
+      "loss": 1.3492,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2992829116493032,
+      "grad_norm": 0.31952470541000366,
+      "learning_rate": 9.963931469792607e-05,
+      "loss": 1.3586,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2998241104045461,
+      "grad_norm": 0.3433592915534973,
+      "learning_rate": 9.981965734896304e-05,
+      "loss": 1.3524,
+      "step": 1108
+    },
+    {
+      "epoch": 0.30036530915978893,
+      "grad_norm": 0.4547680914402008,
+      "learning_rate": 0.0001,
+      "loss": 1.3562,
+      "step": 1110
+    },
+    {
+      "epoch": 0.3009065079150318,
+      "grad_norm": 0.4963592290878296,
+      "learning_rate": 9.999999008881264e-05,
+      "loss": 1.3452,
+      "step": 1112
+    },
+    {
+      "epoch": 0.30144770667027465,
+      "grad_norm": 1.1111193895339966,
+      "learning_rate": 9.999996035525452e-05,
+      "loss": 1.3732,
+      "step": 1114
+    },
+    {
+      "epoch": 0.30198890542551754,
+      "grad_norm": 0.6860964298248291,
+      "learning_rate": 9.999991079933739e-05,
+      "loss": 1.3689,
+      "step": 1116
+    },
+    {
+      "epoch": 0.3025301041807604,
+      "grad_norm": 0.7344204783439636,
+      "learning_rate": 9.999984142108093e-05,
+      "loss": 1.3575,
+      "step": 1118
+    },
+    {
+      "epoch": 0.30307130293600326,
+      "grad_norm": 0.6534725427627563,
+      "learning_rate": 9.999975222051263e-05,
+      "loss": 1.376,
+      "step": 1120
+    },
+    {
+      "epoch": 0.3036125016912461,
+      "grad_norm": 0.5108229517936707,
+      "learning_rate": 9.999964319766785e-05,
+      "loss": 1.3741,
+      "step": 1122
+    },
+    {
+      "epoch": 0.304153700446489,
+      "grad_norm": 0.4888688325881958,
+      "learning_rate": 9.99995143525898e-05,
+      "loss": 1.3555,
+      "step": 1124
+    },
+    {
+      "epoch": 0.3046948992017318,
+      "grad_norm": 0.42808806896209717,
+      "learning_rate": 9.999936568532962e-05,
+      "loss": 1.3548,
+      "step": 1126
+    },
+    {
+      "epoch": 0.3052360979569747,
+      "grad_norm": 0.3921727240085602,
+      "learning_rate": 9.999919719594617e-05,
+      "loss": 1.3559,
+      "step": 1128
+    },
+    {
+      "epoch": 0.30577729671221754,
+      "grad_norm": 0.3473529517650604,
+      "learning_rate": 9.999900888450628e-05,
+      "loss": 1.3603,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3063184954674604,
+      "grad_norm": 0.3337381184101105,
+      "learning_rate": 9.999880075108464e-05,
+      "loss": 1.3642,
+      "step": 1132
+    },
+    {
+      "epoch": 0.30685969422270326,
+      "grad_norm": 0.3363231122493744,
+      "learning_rate": 9.99985727957637e-05,
+      "loss": 1.3606,
+      "step": 1134
+    },
+    {
+      "epoch": 0.30740089297794615,
+      "grad_norm": 0.32726484537124634,
+      "learning_rate": 9.999832501863386e-05,
+      "loss": 1.3493,
+      "step": 1136
+    },
+    {
+      "epoch": 0.30794209173318904,
+      "grad_norm": 0.3190646767616272,
+      "learning_rate": 9.999805741979338e-05,
+      "loss": 1.3518,
+      "step": 1138
+    },
+    {
+      "epoch": 0.30848329048843187,
+      "grad_norm": 0.31244540214538574,
+      "learning_rate": 9.999776999934831e-05,
+      "loss": 1.3495,
+      "step": 1140
+    },
+    {
+      "epoch": 0.30902448924367476,
+      "grad_norm": 0.3286384344100952,
+      "learning_rate": 9.999746275741261e-05,
+      "loss": 1.3517,
+      "step": 1142
+    },
+    {
+      "epoch": 0.3095656879989176,
+      "grad_norm": 0.3630046546459198,
+      "learning_rate": 9.99971356941081e-05,
+      "loss": 1.3641,
+      "step": 1144
+    },
+    {
+      "epoch": 0.3101068867541605,
+      "grad_norm": 0.30771151185035706,
+      "learning_rate": 9.999678880956443e-05,
+      "loss": 1.3571,
+      "step": 1146
+    },
+    {
+      "epoch": 0.3106480855094033,
+      "grad_norm": 0.30026301741600037,
+      "learning_rate": 9.99964221039191e-05,
+      "loss": 1.3541,
+      "step": 1148
+    },
+    {
+      "epoch": 0.3111892842646462,
+      "grad_norm": 0.3128298223018646,
+      "learning_rate": 9.999603557731754e-05,
+      "loss": 1.3556,
+      "step": 1150
+    },
+    {
+      "epoch": 0.31173048301988904,
+      "grad_norm": 0.30185452103614807,
+      "learning_rate": 9.999562922991293e-05,
+      "loss": 1.3484,
+      "step": 1152
+    },
+    {
+      "epoch": 0.3122716817751319,
+      "grad_norm": 0.3274635076522827,
+      "learning_rate": 9.99952030618664e-05,
+      "loss": 1.3729,
+      "step": 1154
+    },
+    {
+      "epoch": 0.31281288053037476,
+      "grad_norm": 0.30549076199531555,
+      "learning_rate": 9.999475707334692e-05,
+      "loss": 1.3642,
+      "step": 1156
+    },
+    {
+      "epoch": 0.31335407928561765,
+      "grad_norm": 0.3147718906402588,
+      "learning_rate": 9.999429126453126e-05,
+      "loss": 1.3493,
+      "step": 1158
+    },
+    {
+      "epoch": 0.3138952780408605,
+      "grad_norm": 0.6205586791038513,
+      "learning_rate": 9.99938056356041e-05,
+      "loss": 1.3623,
+      "step": 1160
+    },
+    {
+      "epoch": 0.31443647679610337,
+      "grad_norm": 0.3471706211566925,
+      "learning_rate": 9.999330018675798e-05,
+      "loss": 1.3533,
+      "step": 1162
+    },
+    {
+      "epoch": 0.31497767555134626,
+      "grad_norm": 1.3515815734863281,
+      "learning_rate": 9.999277491819328e-05,
+      "loss": 1.3565,
+      "step": 1164
+    },
+    {
+      "epoch": 0.3155188743065891,
+      "grad_norm": 733.9155883789062,
+      "learning_rate": 9.999222983011824e-05,
+      "loss": 5.2143,
+      "step": 1166
+    },
+    {
+      "epoch": 0.316060073061832,
+      "grad_norm": 2.9439170360565186,
+      "learning_rate": 9.999166492274894e-05,
+      "loss": 1.4438,
+      "step": 1168
+    },
+    {
+      "epoch": 0.3166012718170748,
+      "grad_norm": 1.5871142148971558,
+      "learning_rate": 9.999108019630938e-05,
+      "loss": 1.4426,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3171424705723177,
+      "grad_norm": 711.9217529296875,
+      "learning_rate": 9.999047565103132e-05,
+      "loss": 3.6935,
+      "step": 1172
+    },
+    {
+      "epoch": 0.31768366932756054,
+      "grad_norm": 100.76264953613281,
+      "learning_rate": 9.998985128715448e-05,
+      "loss": 4.2396,
+      "step": 1174
+    },
+    {
+      "epoch": 0.3182248680828034,
+      "grad_norm": 108.88189697265625,
+      "learning_rate": 9.998920710492634e-05,
+      "loss": 4.9929,
+      "step": 1176
+    },
+    {
+      "epoch": 0.31876606683804626,
+      "grad_norm": 72.18595123291016,
+      "learning_rate": 9.998854310460233e-05,
+      "loss": 6.0375,
+      "step": 1178
+    },
+    {
+      "epoch": 0.31930726559328915,
+      "grad_norm": 59.48538589477539,
+      "learning_rate": 9.998785928644567e-05,
+      "loss": 5.8932,
+      "step": 1180
+    },
+    {
+      "epoch": 0.319848464348532,
+      "grad_norm": 36.32703399658203,
+      "learning_rate": 9.998715565072744e-05,
+      "loss": 6.5369,
+      "step": 1182
+    },
+    {
+      "epoch": 0.32038966310377487,
+      "grad_norm": 18.565351486206055,
+      "learning_rate": 9.998643219772664e-05,
+      "loss": 6.1671,
+      "step": 1184
+    },
+    {
+      "epoch": 0.3209308618590177,
+      "grad_norm": 45.84898376464844,
+      "learning_rate": 9.998568892773003e-05,
+      "loss": 5.9379,
+      "step": 1186
+    },
+    {
+      "epoch": 0.3214720606142606,
+      "grad_norm": 66.2480239868164,
+      "learning_rate": 9.998492584103232e-05,
+      "loss": 5.7071,
+      "step": 1188
+    },
+    {
+      "epoch": 0.3220132593695034,
+      "grad_norm": 41.693092346191406,
+      "learning_rate": 9.998414293793599e-05,
+      "loss": 6.3198,
+      "step": 1190
+    },
+    {
+      "epoch": 0.3225544581247463,
+      "grad_norm": 19.323413848876953,
+      "learning_rate": 9.998334021875147e-05,
+      "loss": 5.377,
+      "step": 1192
+    },
+    {
+      "epoch": 0.3230956568799892,
+      "grad_norm": 15.907301902770996,
+      "learning_rate": 9.998251768379696e-05,
+      "loss": 4.5293,
+      "step": 1194
+    },
+    {
+      "epoch": 0.32363685563523203,
+      "grad_norm": 80.1374740600586,
+      "learning_rate": 9.998167533339857e-05,
+      "loss": 4.3471,
+      "step": 1196
+    },
+    {
+      "epoch": 0.3241780543904749,
+      "grad_norm": 23.298336029052734,
+      "learning_rate": 9.998081316789024e-05,
+      "loss": 3.7461,
+      "step": 1198
+    },
+    {
+      "epoch": 0.32471925314571776,
+      "grad_norm": 82.48027801513672,
+      "learning_rate": 9.997993118761378e-05,
+      "loss": 4.1647,
+      "step": 1200
+    },
+    {
+      "epoch": 0.32526045190096065,
+      "grad_norm": 27.916913986206055,
+      "learning_rate": 9.997902939291883e-05,
+      "loss": 3.9092,
+      "step": 1202
+    },
+    {
+      "epoch": 0.3258016506562035,
+      "grad_norm": 15.70148754119873,
+      "learning_rate": 9.997810778416293e-05,
+      "loss": 3.1628,
+      "step": 1204
+    },
+    {
+      "epoch": 0.32634284941144637,
+      "grad_norm": 18.33330535888672,
+      "learning_rate": 9.997716636171142e-05,
+      "loss": 2.8777,
+      "step": 1206
+    },
+    {
+      "epoch": 0.3268840481666892,
+      "grad_norm": 10.6620512008667,
+      "learning_rate": 9.997620512593755e-05,
+      "loss": 2.3009,
+      "step": 1208
+    },
+    {
+      "epoch": 0.3274252469219321,
+      "grad_norm": 32.01799011230469,
+      "learning_rate": 9.99752240772224e-05,
+      "loss": 1.9617,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3279664456771749,
+      "grad_norm": 5.677090644836426,
+      "learning_rate": 9.997422321595488e-05,
+      "loss": 1.8401,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3285076444324178,
+      "grad_norm": 8.914667129516602,
+      "learning_rate": 9.997320254253179e-05,
+      "loss": 1.6707,
+      "step": 1214
+    },
+    {
+      "epoch": 0.32904884318766064,
+      "grad_norm": 2.3725008964538574,
+      "learning_rate": 9.997216205735779e-05,
+      "loss": 1.5757,
+      "step": 1216
+    },
+    {
+      "epoch": 0.32959004194290353,
+      "grad_norm": 2.418389320373535,
+      "learning_rate": 9.997110176084538e-05,
+      "loss": 1.5154,
+      "step": 1218
+    },
+    {
+      "epoch": 0.33013124069814637,
+      "grad_norm": 2.802185297012329,
+      "learning_rate": 9.997002165341487e-05,
+      "loss": 1.4883,
+      "step": 1220
+    },
+    {
+      "epoch": 0.33067243945338926,
+      "grad_norm": 2.1769211292266846,
+      "learning_rate": 9.996892173549452e-05,
+      "loss": 1.445,
+      "step": 1222
+    },
+    {
+      "epoch": 0.33121363820863214,
+      "grad_norm": 1.799670934677124,
+      "learning_rate": 9.996780200752035e-05,
+      "loss": 1.4276,
+      "step": 1224
+    },
+    {
+      "epoch": 0.331754836963875,
+      "grad_norm": 3.2545313835144043,
+      "learning_rate": 9.996666246993627e-05,
+      "loss": 1.4394,
+      "step": 1226
+    },
+    {
+      "epoch": 0.33229603571911787,
+      "grad_norm": 1.1922351121902466,
+      "learning_rate": 9.996550312319408e-05,
+      "loss": 1.4359,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3328372344743607,
+      "grad_norm": 2.6813228130340576,
+      "learning_rate": 9.996432396775339e-05,
+      "loss": 1.4229,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3333784332296036,
+      "grad_norm": 1.6968843936920166,
+      "learning_rate": 9.996312500408165e-05,
+      "loss": 1.4281,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3339196319848464,
+      "grad_norm": 1.3502254486083984,
+      "learning_rate": 9.996190623265421e-05,
+      "loss": 1.408,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3344608307400893,
+      "grad_norm": 1.2809518575668335,
+      "learning_rate": 9.996066765395424e-05,
+      "loss": 1.4176,
+      "step": 1236
+    },
+    {
+      "epoch": 0.33500202949533214,
+      "grad_norm": 1.0455057621002197,
+      "learning_rate": 9.995940926847279e-05,
+      "loss": 1.4056,
+      "step": 1238
+    },
+    {
+      "epoch": 0.33554322825057503,
+      "grad_norm": 1.3292824029922485,
+      "learning_rate": 9.99581310767087e-05,
+      "loss": 1.4033,
+      "step": 1240
+    },
+    {
+      "epoch": 0.33608442700581787,
+      "grad_norm": 1.5960067510604858,
+      "learning_rate": 9.995683307916875e-05,
+      "loss": 1.379,
+      "step": 1242
+    },
+    {
+      "epoch": 0.33662562576106075,
+      "grad_norm": 1.0471105575561523,
+      "learning_rate": 9.99555152763675e-05,
+      "loss": 1.3823,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3371668245163036,
+      "grad_norm": 2.339273452758789,
+      "learning_rate": 9.99541776688274e-05,
+      "loss": 1.3698,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3377080232715465,
+      "grad_norm": 0.81674724817276,
+      "learning_rate": 9.995282025707875e-05,
+      "loss": 1.4154,
+      "step": 1248
+    },
+    {
+      "epoch": 0.33824922202678936,
+      "grad_norm": 0.6240290999412537,
+      "learning_rate": 9.995144304165968e-05,
+      "loss": 1.4035,
+      "step": 1250
+    },
+    {
+      "epoch": 0.3387904207820322,
+      "grad_norm": 2.281787872314453,
+      "learning_rate": 9.995004602311619e-05,
+      "loss": 1.3906,
+      "step": 1252
+    },
+    {
+      "epoch": 0.3393316195372751,
+      "grad_norm": 0.6818395853042603,
+      "learning_rate": 9.99486292020021e-05,
+      "loss": 1.3853,
+      "step": 1254
+    },
+    {
+      "epoch": 0.3398728182925179,
+      "grad_norm": 6.299881935119629,
+      "learning_rate": 9.994719257887915e-05,
+      "loss": 1.3856,
+      "step": 1256
+    },
+    {
+      "epoch": 0.3404140170477608,
+      "grad_norm": 0.8173750638961792,
+      "learning_rate": 9.994573615431686e-05,
+      "loss": 1.3871,
+      "step": 1258
+    },
+    {
+      "epoch": 0.34095521580300364,
+      "grad_norm": 2.155395746231079,
+      "learning_rate": 9.994425992889262e-05,
+      "loss": 1.3382,
+      "step": 1260
+    },
+    {
+      "epoch": 0.34149641455824653,
+      "grad_norm": 0.5846114754676819,
+      "learning_rate": 9.99427639031917e-05,
+      "loss": 1.3978,
+      "step": 1262
+    },
+    {
+      "epoch": 0.34203761331348936,
+      "grad_norm": 0.6624069213867188,
+      "learning_rate": 9.994124807780717e-05,
+      "loss": 1.3792,
+      "step": 1264
+    },
+    {
+      "epoch": 0.34257881206873225,
+      "grad_norm": 0.5708588361740112,
+      "learning_rate": 9.993971245333998e-05,
+      "loss": 1.3677,
+      "step": 1266
+    },
+    {
+      "epoch": 0.3431200108239751,
+      "grad_norm": 0.5245474576950073,
+      "learning_rate": 9.993815703039894e-05,
+      "loss": 1.3672,
+      "step": 1268
+    },
+    {
+      "epoch": 0.343661209579218,
+      "grad_norm": 0.501871645450592,
+      "learning_rate": 9.993658180960069e-05,
+      "loss": 1.3674,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3442024083344608,
+      "grad_norm": 0.5990382432937622,
+      "learning_rate": 9.993498679156969e-05,
+      "loss": 1.3804,
+      "step": 1272
+    },
+    {
+      "epoch": 0.3447436070897037,
+      "grad_norm": 0.42392146587371826,
+      "learning_rate": 9.993337197693833e-05,
+      "loss": 1.3628,
+      "step": 1274
+    },
+    {
+      "epoch": 0.34528480584494653,
+      "grad_norm": 0.46936917304992676,
+      "learning_rate": 9.993173736634676e-05,
+      "loss": 1.3696,
+      "step": 1276
+    },
+    {
+      "epoch": 0.3458260046001894,
+      "grad_norm": 0.52222740650177,
+      "learning_rate": 9.993008296044304e-05,
+      "loss": 1.3697,
+      "step": 1278
+    },
+    {
+      "epoch": 0.3463672033554323,
+      "grad_norm": 0.3582518398761749,
+      "learning_rate": 9.992840875988305e-05,
+      "loss": 1.3825,
+      "step": 1280
+    },
+    {
+      "epoch": 0.34690840211067514,
+      "grad_norm": 0.3533988296985626,
+      "learning_rate": 9.99267147653305e-05,
+      "loss": 1.361,
+      "step": 1282
+    },
+    {
+      "epoch": 0.34744960086591803,
+      "grad_norm": 0.35905274748802185,
+      "learning_rate": 9.992500097745702e-05,
+      "loss": 1.3721,
+      "step": 1284
+    },
+    {
+      "epoch": 0.34799079962116086,
+      "grad_norm": 0.3057416081428528,
+      "learning_rate": 9.9923267396942e-05,
+      "loss": 1.369,
+      "step": 1286
+    },
+    {
+      "epoch": 0.34853199837640375,
+      "grad_norm": 0.3299311101436615,
+      "learning_rate": 9.992151402447272e-05,
+      "loss": 1.358,
+      "step": 1288
+    },
+    {
+      "epoch": 0.3490731971316466,
+      "grad_norm": 0.3086453080177307,
+      "learning_rate": 9.99197408607443e-05,
+      "loss": 1.3534,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3496143958868895,
+      "grad_norm": 0.3111782968044281,
+      "learning_rate": 9.991794790645969e-05,
+      "loss": 1.3605,
+      "step": 1292
+    },
+    {
+      "epoch": 0.3501555946421323,
+      "grad_norm": 0.3231568932533264,
+      "learning_rate": 9.991613516232974e-05,
+      "loss": 1.3543,
+      "step": 1294
+    },
+    {
+      "epoch": 0.3506967933973752,
+      "grad_norm": 0.3288814425468445,
+      "learning_rate": 9.991430262907309e-05,
+      "loss": 1.3521,
+      "step": 1296
+    },
+    {
+      "epoch": 0.35123799215261803,
+      "grad_norm": 0.3239436745643616,
+      "learning_rate": 9.991245030741622e-05,
+      "loss": 1.3335,
+      "step": 1298
+    },
+    {
+      "epoch": 0.3517791909078609,
+      "grad_norm": 0.3560773730278015,
+      "learning_rate": 9.991057819809353e-05,
+      "loss": 1.3487,
+      "step": 1300
+    },
+    {
+      "epoch": 0.35232038966310375,
+      "grad_norm": 0.4387347400188446,
+      "learning_rate": 9.990868630184716e-05,
+      "loss": 1.3548,
+      "step": 1302
+    },
+    {
+      "epoch": 0.35286158841834664,
+      "grad_norm": 0.32067278027534485,
+      "learning_rate": 9.990677461942717e-05,
+      "loss": 1.3471,
+      "step": 1304
+    },
+    {
+      "epoch": 0.3534027871735895,
+      "grad_norm": 0.4399580955505371,
+      "learning_rate": 9.990484315159146e-05,
+      "loss": 1.3588,
+      "step": 1306
+    },
+    {
+      "epoch": 0.35394398592883236,
+      "grad_norm": 0.9175602793693542,
+      "learning_rate": 9.990289189910571e-05,
+      "loss": 1.3432,
+      "step": 1308
+    },
+    {
+      "epoch": 0.35448518468407525,
+      "grad_norm": 0.45273318886756897,
+      "learning_rate": 9.990092086274352e-05,
+      "loss": 1.3434,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3550263834393181,
+      "grad_norm": 0.3346487879753113,
+      "learning_rate": 9.989893004328632e-05,
+      "loss": 1.3339,
+      "step": 1312
+    },
+    {
+      "epoch": 0.35556758219456097,
+      "grad_norm": 0.4779951870441437,
+      "learning_rate": 9.989691944152333e-05,
+      "loss": 1.3561,
+      "step": 1314
+    },
+    {
+      "epoch": 0.3561087809498038,
+      "grad_norm": 0.6359366774559021,
+      "learning_rate": 9.989488905825166e-05,
+      "loss": 1.3499,
+      "step": 1316
+    },
+    {
+      "epoch": 0.3566499797050467,
+      "grad_norm": 0.5867050290107727,
+      "learning_rate": 9.989283889427625e-05,
+      "loss": 1.3791,
+      "step": 1318
+    },
+    {
+      "epoch": 0.3571911784602895,
+      "grad_norm": 1.869691014289856,
+      "learning_rate": 9.989076895040989e-05,
+      "loss": 1.3663,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3577323772155324,
+      "grad_norm": 2.7147843837738037,
+      "learning_rate": 9.98886792274732e-05,
+      "loss": 1.358,
+      "step": 1322
+    },
+    {
+      "epoch": 0.35827357597077525,
+      "grad_norm": 0.8717885613441467,
+      "learning_rate": 9.988656972629465e-05,
+      "loss": 1.34,
+      "step": 1324
+    },
+    {
+      "epoch": 0.35881477472601814,
+      "grad_norm": 0.7126337885856628,
+      "learning_rate": 9.988444044771054e-05,
+      "loss": 1.3281,
+      "step": 1326
+    },
+    {
+      "epoch": 0.35935597348126097,
+      "grad_norm": 0.7409217357635498,
+      "learning_rate": 9.988229139256502e-05,
+      "loss": 1.3571,
+      "step": 1328
+    },
+    {
+      "epoch": 0.35989717223650386,
+      "grad_norm": 0.5892549157142639,
+      "learning_rate": 9.988012256171006e-05,
+      "loss": 1.3269,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3604383709917467,
+      "grad_norm": 0.4858717620372772,
+      "learning_rate": 9.98779339560055e-05,
+      "loss": 1.3506,
+      "step": 1332
+    },
+    {
+      "epoch": 0.3609795697469896,
+      "grad_norm": 0.37409740686416626,
+      "learning_rate": 9.987572557631903e-05,
+      "loss": 1.3339,
+      "step": 1334
+    },
+    {
+      "epoch": 0.36152076850223247,
+      "grad_norm": 0.38315168023109436,
+      "learning_rate": 9.987349742352611e-05,
+      "loss": 1.3404,
+      "step": 1336
+    },
+    {
+      "epoch": 0.3620619672574753,
+      "grad_norm": 0.32702726125717163,
+      "learning_rate": 9.987124949851014e-05,
+      "loss": 1.3595,
+      "step": 1338
+    },
+    {
+      "epoch": 0.3626031660127182,
+      "grad_norm": 0.3133656680583954,
+      "learning_rate": 9.986898180216226e-05,
+      "loss": 1.3428,
+      "step": 1340
+    },
+    {
+      "epoch": 0.363144364767961,
+      "grad_norm": 0.2916230857372284,
+      "learning_rate": 9.986669433538152e-05,
+      "loss": 1.3381,
+      "step": 1342
+    },
+    {
+      "epoch": 0.3636855635232039,
+      "grad_norm": 0.28036215901374817,
+      "learning_rate": 9.986438709907476e-05,
+      "loss": 1.3447,
+      "step": 1344
+    },
+    {
+      "epoch": 0.36422676227844675,
+      "grad_norm": 0.30352699756622314,
+      "learning_rate": 9.98620600941567e-05,
+      "loss": 1.3427,
+      "step": 1346
+    },
+    {
+      "epoch": 0.36476796103368964,
+      "grad_norm": 0.3100769519805908,
+      "learning_rate": 9.985971332154984e-05,
+      "loss": 1.3603,
+      "step": 1348
+    },
+    {
+      "epoch": 0.36530915978893247,
+      "grad_norm": 0.2933647930622101,
+      "learning_rate": 9.98573467821846e-05,
+      "loss": 1.3646,
+      "step": 1350
+    },
+    {
+      "epoch": 0.36585035854417536,
+      "grad_norm": 0.2938663959503174,
+      "learning_rate": 9.985496047699916e-05,
+      "loss": 1.3763,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3663915572994182,
+      "grad_norm": 0.2916519343852997,
+      "learning_rate": 9.985255440693955e-05,
+      "loss": 1.3431,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3669327560546611,
+      "grad_norm": 0.2954147756099701,
+      "learning_rate": 9.985012857295968e-05,
+      "loss": 1.338,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3674739548099039,
+      "grad_norm": 0.2839341163635254,
+      "learning_rate": 9.984768297602125e-05,
+      "loss": 1.3653,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3680151535651468,
+      "grad_norm": 0.2878473699092865,
+      "learning_rate": 9.984521761709382e-05,
+      "loss": 1.3302,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3685563523203897,
+      "grad_norm": 0.2859325408935547,
+      "learning_rate": 9.984273249715478e-05,
+      "loss": 1.3273,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3690975510756325,
+      "grad_norm": 0.28399959206581116,
+      "learning_rate": 9.984022761718933e-05,
+      "loss": 1.3516,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3696387498308754,
+      "grad_norm": 0.29740169644355774,
+      "learning_rate": 9.983770297819052e-05,
+      "loss": 1.3389,
+      "step": 1366
+    },
+    {
+      "epoch": 0.37017994858611825,
+      "grad_norm": 0.3143361806869507,
+      "learning_rate": 9.983515858115928e-05,
+      "loss": 1.3557,
+      "step": 1368
+    },
+    {
+      "epoch": 0.37072114734136113,
+      "grad_norm": 0.30783936381340027,
+      "learning_rate": 9.983259442710429e-05,
+      "loss": 1.3498,
+      "step": 1370
+    },
+    {
+      "epoch": 0.37126234609660397,
+      "grad_norm": 0.297091543674469,
+      "learning_rate": 9.983001051704211e-05,
+      "loss": 1.3308,
+      "step": 1372
+    },
+    {
+      "epoch": 0.37180354485184686,
+      "grad_norm": 0.3118893504142761,
+      "learning_rate": 9.982740685199712e-05,
+      "loss": 1.3372,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3723447436070897,
+      "grad_norm": 0.2826865017414093,
+      "learning_rate": 9.982478343300155e-05,
+      "loss": 1.3488,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3728859423623326,
+      "grad_norm": 0.2829175889492035,
+      "learning_rate": 9.982214026109544e-05,
+      "loss": 1.3693,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3734271411175754,
+      "grad_norm": 0.3026389479637146,
+      "learning_rate": 9.981947733732668e-05,
+      "loss": 1.3276,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3739683398728183,
+      "grad_norm": 0.30112889409065247,
+      "learning_rate": 9.981679466275096e-05,
+      "loss": 1.3441,
+      "step": 1382
+    },
+    {
+      "epoch": 0.37450953862806113,
+      "grad_norm": 0.27241262793540955,
+      "learning_rate": 9.981409223843183e-05,
+      "loss": 1.3373,
+      "step": 1384
+    },
+    {
+      "epoch": 0.375050737383304,
+      "grad_norm": 0.2804114520549774,
+      "learning_rate": 9.981137006544066e-05,
+      "loss": 1.344,
+      "step": 1386
+    },
+    {
+      "epoch": 0.37559193613854686,
+      "grad_norm": 0.27698764204978943,
+      "learning_rate": 9.980862814485665e-05,
+      "loss": 1.3543,
+      "step": 1388
+    },
+    {
+      "epoch": 0.37613313489378974,
+      "grad_norm": 0.29283177852630615,
+      "learning_rate": 9.980586647776681e-05,
+      "loss": 1.3332,
+      "step": 1390
+    },
+    {
+      "epoch": 0.37667433364903263,
+      "grad_norm": 0.2896028459072113,
+      "learning_rate": 9.980308506526604e-05,
+      "loss": 1.3392,
+      "step": 1392
+    },
+    {
+      "epoch": 0.37721553240427547,
+      "grad_norm": 0.27882838249206543,
+      "learning_rate": 9.980028390845697e-05,
+      "loss": 1.336,
+      "step": 1394
+    },
+    {
+      "epoch": 0.37775673115951836,
+      "grad_norm": 0.2886262834072113,
+      "learning_rate": 9.979746300845015e-05,
+      "loss": 1.3331,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3782979299147612,
+      "grad_norm": 0.3085189163684845,
+      "learning_rate": 9.97946223663639e-05,
+      "loss": 1.3296,
+      "step": 1398
+    },
+    {
+      "epoch": 0.3788391286700041,
+      "grad_norm": 0.3342386484146118,
+      "learning_rate": 9.97917619833244e-05,
+      "loss": 1.351,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3793803274252469,
+      "grad_norm": 0.3263756036758423,
+      "learning_rate": 9.978888186046562e-05,
+      "loss": 1.3526,
+      "step": 1402
+    },
+    {
+      "epoch": 0.3799215261804898,
+      "grad_norm": 0.292346715927124,
+      "learning_rate": 9.97859819989294e-05,
+      "loss": 1.3498,
+      "step": 1404
+    },
+    {
+      "epoch": 0.38046272493573263,
+      "grad_norm": 0.29072263836860657,
+      "learning_rate": 9.978306239986536e-05,
+      "loss": 1.3423,
+      "step": 1406
+    },
+    {
+      "epoch": 0.3810039236909755,
+      "grad_norm": 0.3350834548473358,
+      "learning_rate": 9.978012306443101e-05,
+      "loss": 1.3559,
+      "step": 1408
+    },
+    {
+      "epoch": 0.38154512244621835,
+      "grad_norm": 0.28721559047698975,
+      "learning_rate": 9.977716399379157e-05,
+      "loss": 1.3294,
+      "step": 1410
+    },
+    {
+      "epoch": 0.38208632120146124,
+      "grad_norm": 0.3062276244163513,
+      "learning_rate": 9.977418518912023e-05,
+      "loss": 1.3457,
+      "step": 1412
+    },
+    {
+      "epoch": 0.3826275199567041,
+      "grad_norm": 0.30255332589149475,
+      "learning_rate": 9.977118665159791e-05,
+      "loss": 1.3371,
+      "step": 1414
+    },
+    {
+      "epoch": 0.38316871871194697,
+      "grad_norm": 0.2800199091434479,
+      "learning_rate": 9.976816838241334e-05,
+      "loss": 1.3439,
+      "step": 1416
+    },
+    {
+      "epoch": 0.3837099174671898,
+      "grad_norm": 0.2754746675491333,
+      "learning_rate": 9.976513038276312e-05,
+      "loss": 1.3303,
+      "step": 1418
+    },
+    {
+      "epoch": 0.3842511162224327,
+      "grad_norm": 0.29933616518974304,
+      "learning_rate": 9.976207265385168e-05,
+      "loss": 1.3365,
+      "step": 1420
+    },
+    {
+      "epoch": 0.3847923149776756,
+      "grad_norm": 0.3023386001586914,
+      "learning_rate": 9.975899519689122e-05,
+      "loss": 1.3164,
+      "step": 1422
+    },
+    {
+      "epoch": 0.3853335137329184,
+      "grad_norm": 0.2901383936405182,
+      "learning_rate": 9.975589801310181e-05,
+      "loss": 1.3209,
+      "step": 1424
+    },
+    {
+      "epoch": 0.3858747124881613,
+      "grad_norm": 0.28566035628318787,
+      "learning_rate": 9.975278110371131e-05,
+      "loss": 1.3301,
+      "step": 1426
+    },
+    {
+      "epoch": 0.38641591124340413,
+      "grad_norm": 0.3010505735874176,
+      "learning_rate": 9.974964446995543e-05,
+      "loss": 1.319,
+      "step": 1428
+    },
+    {
+      "epoch": 0.386957109998647,
+      "grad_norm": 0.2977135479450226,
+      "learning_rate": 9.974648811307766e-05,
+      "loss": 1.3311,
+      "step": 1430
+    },
+    {
+      "epoch": 0.38749830875388985,
+      "grad_norm": 0.28914034366607666,
+      "learning_rate": 9.974331203432932e-05,
+      "loss": 1.343,
+      "step": 1432
+    },
+    {
+      "epoch": 0.38803950750913274,
+      "grad_norm": 0.2842980623245239,
+      "learning_rate": 9.974011623496958e-05,
+      "loss": 1.3162,
+      "step": 1434
+    },
+    {
+      "epoch": 0.3885807062643756,
+      "grad_norm": 0.3048929274082184,
+      "learning_rate": 9.97369007162654e-05,
+      "loss": 1.3166,
+      "step": 1436
+    },
+    {
+      "epoch": 0.38912190501961846,
+      "grad_norm": 0.3024531304836273,
+      "learning_rate": 9.973366547949157e-05,
+      "loss": 1.3156,
+      "step": 1438
+    },
+    {
+      "epoch": 0.3896631037748613,
+      "grad_norm": 0.2911103367805481,
+      "learning_rate": 9.973041052593068e-05,
+      "loss": 1.3314,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3902043025301042,
+      "grad_norm": 0.30932334065437317,
+      "learning_rate": 9.972713585687317e-05,
+      "loss": 1.3144,
+      "step": 1442
+    },
+    {
+      "epoch": 0.390745501285347,
+      "grad_norm": 0.302971750497818,
+      "learning_rate": 9.972384147361725e-05,
+      "loss": 1.3431,
+      "step": 1444
+    },
+    {
+      "epoch": 0.3912867000405899,
+      "grad_norm": 0.32412296533584595,
+      "learning_rate": 9.972052737746898e-05,
+      "loss": 1.3167,
+      "step": 1446
+    },
+    {
+      "epoch": 0.3918278987958328,
+      "grad_norm": 0.4637945890426636,
+      "learning_rate": 9.97171935697422e-05,
+      "loss": 1.3433,
+      "step": 1448
+    },
+    {
+      "epoch": 0.39236909755107563,
+      "grad_norm": 0.32690081000328064,
+      "learning_rate": 9.971384005175864e-05,
+      "loss": 1.3327,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3929102963063185,
+      "grad_norm": 0.3049994111061096,
+      "learning_rate": 9.971046682484776e-05,
+      "loss": 1.3401,
+      "step": 1452
+    },
+    {
+      "epoch": 0.39345149506156135,
+      "grad_norm": 0.306095689535141,
+      "learning_rate": 9.970707389034688e-05,
+      "loss": 1.3205,
+      "step": 1454
+    },
+    {
+      "epoch": 0.39399269381680424,
+      "grad_norm": 0.3375592529773712,
+      "learning_rate": 9.970366124960111e-05,
+      "loss": 1.3243,
+      "step": 1456
+    },
+    {
+      "epoch": 0.3945338925720471,
+      "grad_norm": 0.30508387088775635,
+      "learning_rate": 9.970022890396338e-05,
+      "loss": 1.3342,
+      "step": 1458
+    },
+    {
+      "epoch": 0.39507509132728996,
+      "grad_norm": 0.2996918261051178,
+      "learning_rate": 9.969677685479444e-05,
+      "loss": 1.3457,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3956162900825328,
+      "grad_norm": 0.29500269889831543,
+      "learning_rate": 9.969330510346286e-05,
+      "loss": 1.3306,
+      "step": 1462
+    },
+    {
+      "epoch": 0.3961574888377757,
+      "grad_norm": 0.28392598032951355,
+      "learning_rate": 9.9689813651345e-05,
+      "loss": 1.3347,
+      "step": 1464
+    },
+    {
+      "epoch": 0.3966986875930185,
+      "grad_norm": 0.2859434485435486,
+      "learning_rate": 9.968630249982503e-05,
+      "loss": 1.3342,
+      "step": 1466
+    },
+    {
+      "epoch": 0.3972398863482614,
+      "grad_norm": 0.3038876950740814,
+      "learning_rate": 9.968277165029494e-05,
+      "loss": 1.3248,
+      "step": 1468
+    },
+    {
+      "epoch": 0.39778108510350424,
+      "grad_norm": 0.3060581088066101,
+      "learning_rate": 9.967922110415454e-05,
+      "loss": 1.3403,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39832228385874713,
+      "grad_norm": 0.30475133657455444,
+      "learning_rate": 9.96756508628114e-05,
+      "loss": 1.3338,
+      "step": 1472
+    },
+    {
+      "epoch": 0.39886348261398996,
+      "grad_norm": 0.33263343572616577,
+      "learning_rate": 9.967206092768095e-05,
+      "loss": 1.3209,
+      "step": 1474
+    },
+    {
+      "epoch": 0.39940468136923285,
+      "grad_norm": 0.2895435094833374,
+      "learning_rate": 9.966845130018645e-05,
+      "loss": 1.3352,
+      "step": 1476
+    },
+    {
+      "epoch": 0.39994588012447574,
+      "grad_norm": 0.27237775921821594,
+      "learning_rate": 9.966482198175886e-05,
+      "loss": 1.3239,
+      "step": 1478
+    },
+    {
+      "epoch": 0.40048707887971857,
+      "grad_norm": 0.2740168571472168,
+      "learning_rate": 9.966117297383707e-05,
+      "loss": 1.3371,
+      "step": 1480
+    },
+    {
+      "epoch": 0.40102827763496146,
+      "grad_norm": 0.30601269006729126,
+      "learning_rate": 9.965750427786768e-05,
+      "loss": 1.343,
+      "step": 1482
+    },
+    {
+      "epoch": 0.4015694763902043,
+      "grad_norm": 0.28768840432167053,
+      "learning_rate": 9.965381589530518e-05,
+      "loss": 1.3442,
+      "step": 1484
+    },
+    {
+      "epoch": 0.4021106751454472,
+      "grad_norm": 0.28244882822036743,
+      "learning_rate": 9.965010782761177e-05,
+      "loss": 1.3336,
+      "step": 1486
+    },
+    {
+      "epoch": 0.40265187390069,
+      "grad_norm": 0.2694818079471588,
+      "learning_rate": 9.964638007625754e-05,
+      "loss": 1.3448,
+      "step": 1488
+    },
+    {
+      "epoch": 0.4031930726559329,
+      "grad_norm": 0.29507288336753845,
+      "learning_rate": 9.964263264272033e-05,
+      "loss": 1.327,
+      "step": 1490
+    },
+    {
+      "epoch": 0.40373427141117574,
+      "grad_norm": 0.3036315143108368,
+      "learning_rate": 9.963886552848581e-05,
+      "loss": 1.3289,
+      "step": 1492
+    },
+    {
+      "epoch": 0.4042754701664186,
+      "grad_norm": 0.2737107574939728,
+      "learning_rate": 9.963507873504744e-05,
+      "loss": 1.3281,
+      "step": 1494
+    },
+    {
+      "epoch": 0.40481666892166146,
+      "grad_norm": 0.29833105206489563,
+      "learning_rate": 9.963127226390647e-05,
+      "loss": 1.3378,
+      "step": 1496
+    },
+    {
+      "epoch": 0.40535786767690435,
+      "grad_norm": 0.32203689217567444,
+      "learning_rate": 9.9627446116572e-05,
+      "loss": 1.3158,
+      "step": 1498
+    },
+    {
+      "epoch": 0.4058990664321472,
+      "grad_norm": 0.27837038040161133,
+      "learning_rate": 9.962360029456086e-05,
+      "loss": 1.3051,
+      "step": 1500
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 11088,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 300,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.235471745541734e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}