{
  "best_metric": 0.890067458152771,
  "best_model_checkpoint": "autotrain-okjuj-buuqs/checkpoint-804",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 804,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014925373134328358,
      "grad_norm": 3.468337297439575,
      "learning_rate": 4.975124378109453e-07,
      "loss": 3.5512,
      "step": 2
    },
    {
      "epoch": 0.029850746268656716,
      "grad_norm": 3.3904953002929688,
      "learning_rate": 9.950248756218907e-07,
      "loss": 3.5495,
      "step": 4
    },
    {
      "epoch": 0.04477611940298507,
      "grad_norm": 8.210745811462402,
      "learning_rate": 1.4925373134328358e-06,
      "loss": 3.5323,
      "step": 6
    },
    {
      "epoch": 0.05970149253731343,
      "grad_norm": 3.321380376815796,
      "learning_rate": 1.9900497512437813e-06,
      "loss": 3.5311,
      "step": 8
    },
    {
      "epoch": 0.07462686567164178,
      "grad_norm": 3.103438377380371,
      "learning_rate": 2.4875621890547264e-06,
      "loss": 3.5312,
      "step": 10
    },
    {
      "epoch": 0.08955223880597014,
      "grad_norm": 3.2276968955993652,
      "learning_rate": 2.9850746268656716e-06,
      "loss": 3.5472,
      "step": 12
    },
    {
      "epoch": 0.1044776119402985,
      "grad_norm": 3.511948585510254,
      "learning_rate": 3.482587064676617e-06,
      "loss": 3.5238,
      "step": 14
    },
    {
      "epoch": 0.11940298507462686,
      "grad_norm": 3.459820508956909,
      "learning_rate": 3.980099502487563e-06,
      "loss": 3.5457,
      "step": 16
    },
    {
      "epoch": 0.13432835820895522,
      "grad_norm": 3.187225103378296,
      "learning_rate": 4.477611940298508e-06,
      "loss": 3.5064,
      "step": 18
    },
    {
      "epoch": 0.14925373134328357,
      "grad_norm": 3.3685667514801025,
      "learning_rate": 4.975124378109453e-06,
      "loss": 3.5019,
      "step": 20
    },
    {
      "epoch": 0.16417910447761194,
      "grad_norm": 3.525916576385498,
      "learning_rate": 5.472636815920398e-06,
      "loss": 3.5202,
      "step": 22
    },
    {
      "epoch": 0.1791044776119403,
      "grad_norm": 3.199352979660034,
      "learning_rate": 5.970149253731343e-06,
      "loss": 3.4878,
      "step": 24
    },
    {
      "epoch": 0.19402985074626866,
      "grad_norm": 3.151981830596924,
      "learning_rate": 6.467661691542288e-06,
      "loss": 3.5254,
      "step": 26
    },
    {
      "epoch": 0.208955223880597,
      "grad_norm": 3.353118896484375,
      "learning_rate": 6.965174129353234e-06,
      "loss": 3.5179,
      "step": 28
    },
    {
      "epoch": 0.22388059701492538,
      "grad_norm": 5.352255344390869,
      "learning_rate": 7.4626865671641785e-06,
      "loss": 3.4951,
      "step": 30
    },
    {
      "epoch": 0.23880597014925373,
      "grad_norm": 4.2452712059021,
      "learning_rate": 7.960199004975125e-06,
      "loss": 3.5144,
      "step": 32
    },
    {
      "epoch": 0.2537313432835821,
      "grad_norm": 3.4914658069610596,
      "learning_rate": 8.45771144278607e-06,
      "loss": 3.4755,
      "step": 34
    },
    {
      "epoch": 0.26865671641791045,
      "grad_norm": 3.413722038269043,
      "learning_rate": 8.955223880597016e-06,
      "loss": 3.5059,
      "step": 36
    },
    {
      "epoch": 0.2835820895522388,
      "grad_norm": 3.5784690380096436,
      "learning_rate": 9.45273631840796e-06,
      "loss": 3.503,
      "step": 38
    },
    {
      "epoch": 0.29850746268656714,
      "grad_norm": 3.3770291805267334,
      "learning_rate": 9.950248756218906e-06,
      "loss": 3.4968,
      "step": 40
    },
    {
      "epoch": 0.31343283582089554,
      "grad_norm": 3.709470748901367,
      "learning_rate": 1.0447761194029851e-05,
      "loss": 3.5453,
      "step": 42
    },
    {
      "epoch": 0.3283582089552239,
      "grad_norm": 3.529064655303955,
      "learning_rate": 1.0945273631840796e-05,
      "loss": 3.4461,
      "step": 44
    },
    {
      "epoch": 0.34328358208955223,
      "grad_norm": 3.4094252586364746,
      "learning_rate": 1.1442786069651743e-05,
      "loss": 3.483,
      "step": 46
    },
    {
      "epoch": 0.3582089552238806,
      "grad_norm": 3.385557174682617,
      "learning_rate": 1.1940298507462686e-05,
      "loss": 3.4522,
      "step": 48
    },
    {
      "epoch": 0.373134328358209,
      "grad_norm": 3.4028918743133545,
      "learning_rate": 1.2437810945273633e-05,
      "loss": 3.4814,
      "step": 50
    },
    {
      "epoch": 0.3880597014925373,
      "grad_norm": 3.482884168624878,
      "learning_rate": 1.2935323383084577e-05,
      "loss": 3.4917,
      "step": 52
    },
    {
      "epoch": 0.40298507462686567,
      "grad_norm": 3.4329397678375244,
      "learning_rate": 1.3432835820895523e-05,
      "loss": 3.4945,
      "step": 54
    },
    {
      "epoch": 0.417910447761194,
      "grad_norm": 3.6114747524261475,
      "learning_rate": 1.3930348258706468e-05,
      "loss": 3.3991,
      "step": 56
    },
    {
      "epoch": 0.43283582089552236,
      "grad_norm": 3.743497848510742,
      "learning_rate": 1.4427860696517415e-05,
      "loss": 3.4314,
      "step": 58
    },
    {
      "epoch": 0.44776119402985076,
      "grad_norm": 13.538641929626465,
      "learning_rate": 1.4925373134328357e-05,
      "loss": 3.4495,
      "step": 60
    },
    {
      "epoch": 0.4626865671641791,
      "grad_norm": 3.536511182785034,
      "learning_rate": 1.5422885572139304e-05,
      "loss": 3.4579,
      "step": 62
    },
    {
      "epoch": 0.47761194029850745,
      "grad_norm": 4.389700889587402,
      "learning_rate": 1.592039800995025e-05,
      "loss": 3.3908,
      "step": 64
    },
    {
      "epoch": 0.4925373134328358,
      "grad_norm": 4.821465969085693,
      "learning_rate": 1.6417910447761194e-05,
      "loss": 3.4305,
      "step": 66
    },
    {
      "epoch": 0.5074626865671642,
      "grad_norm": 3.5865774154663086,
      "learning_rate": 1.691542288557214e-05,
      "loss": 3.4297,
      "step": 68
    },
    {
      "epoch": 0.5223880597014925,
      "grad_norm": 3.3377368450164795,
      "learning_rate": 1.7412935323383088e-05,
      "loss": 3.5506,
      "step": 70
    },
    {
      "epoch": 0.5373134328358209,
      "grad_norm": 3.887446880340576,
      "learning_rate": 1.791044776119403e-05,
      "loss": 3.4623,
      "step": 72
    },
    {
      "epoch": 0.5522388059701493,
      "grad_norm": 3.546586275100708,
      "learning_rate": 1.8407960199004975e-05,
      "loss": 3.4059,
      "step": 74
    },
    {
      "epoch": 0.5671641791044776,
      "grad_norm": 3.8084304332733154,
      "learning_rate": 1.890547263681592e-05,
      "loss": 3.3708,
      "step": 76
    },
    {
      "epoch": 0.582089552238806,
      "grad_norm": 3.5616061687469482,
      "learning_rate": 1.9402985074626868e-05,
      "loss": 3.4297,
      "step": 78
    },
    {
      "epoch": 0.5970149253731343,
      "grad_norm": 3.7048122882843018,
      "learning_rate": 1.990049751243781e-05,
      "loss": 3.3541,
      "step": 80
    },
    {
      "epoch": 0.6119402985074627,
      "grad_norm": 3.7918601036071777,
      "learning_rate": 2.0398009950248755e-05,
      "loss": 3.3734,
      "step": 82
    },
    {
      "epoch": 0.6268656716417911,
      "grad_norm": 4.018467903137207,
      "learning_rate": 2.0895522388059702e-05,
      "loss": 3.3347,
      "step": 84
    },
    {
      "epoch": 0.6417910447761194,
      "grad_norm": 3.6919217109680176,
      "learning_rate": 2.139303482587065e-05,
      "loss": 3.4747,
      "step": 86
    },
    {
      "epoch": 0.6567164179104478,
      "grad_norm": 3.553607940673828,
      "learning_rate": 2.1890547263681592e-05,
      "loss": 3.346,
      "step": 88
    },
    {
      "epoch": 0.6716417910447762,
      "grad_norm": 3.717123508453369,
      "learning_rate": 2.238805970149254e-05,
      "loss": 3.3681,
      "step": 90
    },
    {
      "epoch": 0.6865671641791045,
      "grad_norm": 4.329309463500977,
      "learning_rate": 2.2885572139303486e-05,
      "loss": 3.2841,
      "step": 92
    },
    {
      "epoch": 0.7014925373134329,
      "grad_norm": 3.9699785709381104,
      "learning_rate": 2.338308457711443e-05,
      "loss": 3.3339,
      "step": 94
    },
    {
      "epoch": 0.7164179104477612,
      "grad_norm": 3.9800264835357666,
      "learning_rate": 2.3880597014925373e-05,
      "loss": 3.272,
      "step": 96
    },
    {
      "epoch": 0.7313432835820896,
      "grad_norm": 3.881544589996338,
      "learning_rate": 2.437810945273632e-05,
      "loss": 3.3739,
      "step": 98
    },
    {
      "epoch": 0.746268656716418,
      "grad_norm": 3.7055130004882812,
      "learning_rate": 2.4875621890547266e-05,
      "loss": 3.2847,
      "step": 100
    },
    {
      "epoch": 0.7611940298507462,
      "grad_norm": 4.061646461486816,
      "learning_rate": 2.537313432835821e-05,
      "loss": 3.2573,
      "step": 102
    },
    {
      "epoch": 0.7761194029850746,
      "grad_norm": 4.569900035858154,
      "learning_rate": 2.5870646766169153e-05,
      "loss": 3.2594,
      "step": 104
    },
    {
      "epoch": 0.7910447761194029,
      "grad_norm": 4.053166389465332,
      "learning_rate": 2.6368159203980103e-05,
      "loss": 3.2739,
      "step": 106
    },
    {
      "epoch": 0.8059701492537313,
      "grad_norm": 5.338119029998779,
      "learning_rate": 2.6865671641791047e-05,
      "loss": 3.1965,
      "step": 108
    },
    {
      "epoch": 0.8208955223880597,
      "grad_norm": 4.987647533416748,
      "learning_rate": 2.736318407960199e-05,
      "loss": 3.1951,
      "step": 110
    },
    {
      "epoch": 0.835820895522388,
      "grad_norm": 4.582657814025879,
      "learning_rate": 2.7860696517412937e-05,
      "loss": 3.1758,
      "step": 112
    },
    {
      "epoch": 0.8507462686567164,
      "grad_norm": 4.130568027496338,
      "learning_rate": 2.835820895522388e-05,
      "loss": 3.2563,
      "step": 114
    },
    {
      "epoch": 0.8656716417910447,
      "grad_norm": 3.6177594661712646,
      "learning_rate": 2.885572139303483e-05,
      "loss": 3.1394,
      "step": 116
    },
    {
      "epoch": 0.8805970149253731,
      "grad_norm": 4.4392242431640625,
      "learning_rate": 2.935323383084577e-05,
      "loss": 2.9928,
      "step": 118
    },
    {
      "epoch": 0.8955223880597015,
      "grad_norm": 4.2055439949035645,
      "learning_rate": 2.9850746268656714e-05,
      "loss": 3.0482,
      "step": 120
    },
    {
      "epoch": 0.9104477611940298,
      "grad_norm": 4.438642978668213,
      "learning_rate": 3.0348258706467664e-05,
      "loss": 3.024,
      "step": 122
    },
    {
      "epoch": 0.9253731343283582,
      "grad_norm": 4.41525936126709,
      "learning_rate": 3.084577114427861e-05,
      "loss": 3.1619,
      "step": 124
    },
    {
      "epoch": 0.9402985074626866,
      "grad_norm": 4.188882827758789,
      "learning_rate": 3.1343283582089554e-05,
      "loss": 3.2697,
      "step": 126
    },
    {
      "epoch": 0.9552238805970149,
      "grad_norm": 3.903090000152588,
      "learning_rate": 3.18407960199005e-05,
      "loss": 3.0515,
      "step": 128
    },
    {
      "epoch": 0.9701492537313433,
      "grad_norm": 4.03048849105835,
      "learning_rate": 3.233830845771145e-05,
      "loss": 2.9312,
      "step": 130
    },
    {
      "epoch": 0.9850746268656716,
      "grad_norm": 3.842494249343872,
      "learning_rate": 3.283582089552239e-05,
      "loss": 3.0714,
      "step": 132
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.9151153564453125,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 2.8809,
      "step": 134
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.375,
      "eval_f1_macro": 0.2794185541385505,
      "eval_f1_micro": 0.375,
      "eval_f1_weighted": 0.29179203240000456,
      "eval_loss": 2.9781627655029297,
      "eval_precision_macro": 0.26769010075416033,
      "eval_precision_micro": 0.375,
      "eval_precision_weighted": 0.28345060265850164,
      "eval_recall_macro": 0.3689075630252101,
      "eval_recall_micro": 0.375,
      "eval_recall_weighted": 0.375,
      "eval_runtime": 38.3444,
      "eval_samples_per_second": 5.216,
      "eval_steps_per_second": 0.183,
      "step": 134
    },
    {
      "epoch": 1.0149253731343284,
      "grad_norm": 3.894672155380249,
      "learning_rate": 3.383084577114428e-05,
      "loss": 2.8126,
      "step": 136
    },
    {
      "epoch": 1.0298507462686568,
      "grad_norm": 4.255545139312744,
      "learning_rate": 3.432835820895522e-05,
      "loss": 2.9291,
      "step": 138
    },
    {
      "epoch": 1.044776119402985,
      "grad_norm": 4.275138854980469,
      "learning_rate": 3.4825870646766175e-05,
      "loss": 2.8673,
      "step": 140
    },
    {
      "epoch": 1.0597014925373134,
      "grad_norm": 4.796700477600098,
      "learning_rate": 3.5323383084577115e-05,
      "loss": 2.8431,
      "step": 142
    },
    {
      "epoch": 1.0746268656716418,
      "grad_norm": 4.217311859130859,
      "learning_rate": 3.582089552238806e-05,
      "loss": 2.8104,
      "step": 144
    },
    {
      "epoch": 1.0895522388059702,
      "grad_norm": 4.417973041534424,
      "learning_rate": 3.631840796019901e-05,
      "loss": 2.7942,
      "step": 146
    },
    {
      "epoch": 1.1044776119402986,
      "grad_norm": 3.8580806255340576,
      "learning_rate": 3.681592039800995e-05,
      "loss": 2.6514,
      "step": 148
    },
    {
      "epoch": 1.1194029850746268,
      "grad_norm": 4.099730014801025,
      "learning_rate": 3.73134328358209e-05,
      "loss": 2.693,
      "step": 150
    },
    {
      "epoch": 1.1343283582089552,
      "grad_norm": 4.598573684692383,
      "learning_rate": 3.781094527363184e-05,
      "loss": 2.7205,
      "step": 152
    },
    {
      "epoch": 1.1492537313432836,
      "grad_norm": 4.880782604217529,
      "learning_rate": 3.830845771144278e-05,
      "loss": 2.8268,
      "step": 154
    },
    {
      "epoch": 1.164179104477612,
      "grad_norm": 3.9455466270446777,
      "learning_rate": 3.8805970149253736e-05,
      "loss": 2.5482,
      "step": 156
    },
    {
      "epoch": 1.1791044776119404,
      "grad_norm": 3.5259761810302734,
      "learning_rate": 3.9303482587064676e-05,
      "loss": 2.4929,
      "step": 158
    },
    {
      "epoch": 1.1940298507462686,
      "grad_norm": 4.351325035095215,
      "learning_rate": 3.980099502487562e-05,
      "loss": 2.6261,
      "step": 160
    },
    {
      "epoch": 1.208955223880597,
      "grad_norm": 3.8126604557037354,
      "learning_rate": 4.029850746268657e-05,
      "loss": 2.3924,
      "step": 162
    },
    {
      "epoch": 1.2238805970149254,
      "grad_norm": 5.261939525604248,
      "learning_rate": 4.079601990049751e-05,
      "loss": 2.3364,
      "step": 164
    },
    {
      "epoch": 1.2388059701492538,
      "grad_norm": 5.208514213562012,
      "learning_rate": 4.1293532338308464e-05,
      "loss": 2.3356,
      "step": 166
    },
    {
      "epoch": 1.2537313432835822,
      "grad_norm": 4.369601249694824,
      "learning_rate": 4.1791044776119404e-05,
      "loss": 2.4191,
      "step": 168
    },
    {
      "epoch": 1.2686567164179103,
      "grad_norm": 5.281370639801025,
      "learning_rate": 4.228855721393035e-05,
      "loss": 2.4544,
      "step": 170
    },
    {
      "epoch": 1.2835820895522387,
      "grad_norm": 4.296084880828857,
      "learning_rate": 4.27860696517413e-05,
      "loss": 2.4446,
      "step": 172
    },
    {
      "epoch": 1.2985074626865671,
      "grad_norm": 3.7815287113189697,
      "learning_rate": 4.328358208955224e-05,
      "loss": 2.3687,
      "step": 174
    },
    {
      "epoch": 1.3134328358208955,
      "grad_norm": 7.161779403686523,
      "learning_rate": 4.3781094527363184e-05,
      "loss": 2.67,
      "step": 176
    },
    {
      "epoch": 1.328358208955224,
      "grad_norm": 4.957999229431152,
      "learning_rate": 4.427860696517413e-05,
      "loss": 2.4644,
      "step": 178
    },
    {
      "epoch": 1.3432835820895521,
      "grad_norm": 4.455275058746338,
      "learning_rate": 4.477611940298508e-05,
      "loss": 2.4389,
      "step": 180
    },
    {
      "epoch": 1.3582089552238805,
      "grad_norm": 5.357694625854492,
      "learning_rate": 4.5273631840796025e-05,
      "loss": 2.1849,
      "step": 182
    },
    {
      "epoch": 1.373134328358209,
      "grad_norm": 6.346301078796387,
      "learning_rate": 4.577114427860697e-05,
      "loss": 2.3134,
      "step": 184
    },
    {
      "epoch": 1.3880597014925373,
      "grad_norm": 6.189851760864258,
      "learning_rate": 4.626865671641791e-05,
      "loss": 2.322,
      "step": 186
    },
    {
      "epoch": 1.4029850746268657,
      "grad_norm": 3.729904890060425,
      "learning_rate": 4.676616915422886e-05,
      "loss": 2.2053,
      "step": 188
    },
    {
      "epoch": 1.417910447761194,
      "grad_norm": 4.210063457489014,
      "learning_rate": 4.7263681592039805e-05,
      "loss": 1.9863,
      "step": 190
    },
    {
      "epoch": 1.4328358208955223,
      "grad_norm": 4.447649002075195,
      "learning_rate": 4.7761194029850745e-05,
      "loss": 2.0302,
      "step": 192
    },
    {
      "epoch": 1.4477611940298507,
      "grad_norm": 5.010406970977783,
      "learning_rate": 4.82587064676617e-05,
      "loss": 2.379,
      "step": 194
    },
    {
      "epoch": 1.462686567164179,
      "grad_norm": 4.134596824645996,
      "learning_rate": 4.875621890547264e-05,
      "loss": 2.1104,
      "step": 196
    },
    {
      "epoch": 1.4776119402985075,
      "grad_norm": 3.678053140640259,
      "learning_rate": 4.9253731343283586e-05,
      "loss": 2.0437,
      "step": 198
    },
    {
      "epoch": 1.4925373134328357,
      "grad_norm": 6.969429016113281,
      "learning_rate": 4.975124378109453e-05,
      "loss": 2.1688,
      "step": 200
    },
    {
      "epoch": 1.5074626865671643,
      "grad_norm": 5.489563941955566,
      "learning_rate": 4.997236042012162e-05,
      "loss": 1.9494,
      "step": 202
    },
    {
      "epoch": 1.5223880597014925,
      "grad_norm": 3.8776581287384033,
      "learning_rate": 4.991708126036485e-05,
      "loss": 1.7246,
      "step": 204
    },
    {
      "epoch": 1.537313432835821,
      "grad_norm": 3.915743350982666,
      "learning_rate": 4.9861802100608076e-05,
      "loss": 2.1206,
      "step": 206
    },
    {
      "epoch": 1.5522388059701493,
      "grad_norm": 10.498429298400879,
      "learning_rate": 4.9806522940851304e-05,
      "loss": 1.8089,
      "step": 208
    },
    {
      "epoch": 1.5671641791044775,
      "grad_norm": 5.074433326721191,
      "learning_rate": 4.975124378109453e-05,
      "loss": 2.2841,
      "step": 210
    },
    {
      "epoch": 1.582089552238806,
      "grad_norm": 16.675743103027344,
      "learning_rate": 4.969596462133776e-05,
      "loss": 1.9404,
      "step": 212
    },
    {
      "epoch": 1.5970149253731343,
      "grad_norm": 4.692626953125,
      "learning_rate": 4.964068546158099e-05,
      "loss": 1.9563,
      "step": 214
    },
    {
      "epoch": 1.6119402985074627,
      "grad_norm": 4.603738307952881,
      "learning_rate": 4.958540630182422e-05,
      "loss": 2.4004,
      "step": 216
    },
    {
      "epoch": 1.626865671641791,
      "grad_norm": 29.3167781829834,
      "learning_rate": 4.9530127142067445e-05,
      "loss": 2.1461,
      "step": 218
    },
    {
      "epoch": 1.6417910447761193,
      "grad_norm": 4.929641246795654,
      "learning_rate": 4.947484798231067e-05,
      "loss": 1.6009,
      "step": 220
    },
    {
      "epoch": 1.6567164179104479,
      "grad_norm": 7.109385967254639,
      "learning_rate": 4.94195688225539e-05,
      "loss": 1.884,
      "step": 222
    },
    {
      "epoch": 1.671641791044776,
      "grad_norm": 14.075770378112793,
      "learning_rate": 4.936428966279712e-05,
      "loss": 1.6543,
      "step": 224
    },
    {
      "epoch": 1.6865671641791045,
      "grad_norm": 4.957328796386719,
      "learning_rate": 4.930901050304035e-05,
      "loss": 1.6223,
      "step": 226
    },
    {
      "epoch": 1.7014925373134329,
      "grad_norm": 6.562300205230713,
      "learning_rate": 4.9253731343283586e-05,
      "loss": 1.4456,
      "step": 228
    },
    {
      "epoch": 1.716417910447761,
      "grad_norm": 5.4999237060546875,
      "learning_rate": 4.9198452183526814e-05,
      "loss": 1.7814,
      "step": 230
    },
    {
      "epoch": 1.7313432835820897,
      "grad_norm": 5.5423736572265625,
      "learning_rate": 4.914317302377004e-05,
      "loss": 1.884,
      "step": 232
    },
    {
      "epoch": 1.7462686567164178,
      "grad_norm": 5.507659435272217,
      "learning_rate": 4.908789386401327e-05,
      "loss": 1.8302,
      "step": 234
    },
    {
      "epoch": 1.7611940298507462,
      "grad_norm": 7.983649730682373,
      "learning_rate": 4.90326147042565e-05,
      "loss": 1.9323,
      "step": 236
    },
    {
      "epoch": 1.7761194029850746,
      "grad_norm": 4.843520641326904,
      "learning_rate": 4.8977335544499726e-05,
      "loss": 1.6411,
      "step": 238
    },
    {
      "epoch": 1.7910447761194028,
      "grad_norm": 4.717188835144043,
      "learning_rate": 4.8922056384742954e-05,
      "loss": 1.6714,
      "step": 240
    },
    {
      "epoch": 1.8059701492537314,
      "grad_norm": 4.654526233673096,
      "learning_rate": 4.886677722498618e-05,
      "loss": 1.6478,
      "step": 242
    },
    {
      "epoch": 1.8208955223880596,
      "grad_norm": 5.104541301727295,
      "learning_rate": 4.881149806522941e-05,
      "loss": 1.6125,
      "step": 244
    },
    {
      "epoch": 1.835820895522388,
      "grad_norm": 4.356812000274658,
      "learning_rate": 4.875621890547264e-05,
      "loss": 1.9471,
      "step": 246
    },
    {
      "epoch": 1.8507462686567164,
      "grad_norm": 4.166391849517822,
      "learning_rate": 4.870093974571587e-05,
      "loss": 1.4783,
      "step": 248
    },
    {
      "epoch": 1.8656716417910446,
      "grad_norm": 6.2490057945251465,
      "learning_rate": 4.8645660585959095e-05,
      "loss": 1.8157,
      "step": 250
    },
    {
      "epoch": 1.8805970149253732,
      "grad_norm": 5.395322322845459,
      "learning_rate": 4.859038142620232e-05,
      "loss": 1.8702,
      "step": 252
    },
    {
      "epoch": 1.8955223880597014,
      "grad_norm": 4.278140544891357,
      "learning_rate": 4.853510226644555e-05,
      "loss": 1.5588,
      "step": 254
    },
    {
      "epoch": 1.9104477611940298,
      "grad_norm": 5.143709182739258,
      "learning_rate": 4.847982310668878e-05,
      "loss": 1.5783,
      "step": 256
    },
    {
      "epoch": 1.9253731343283582,
      "grad_norm": 8.840173721313477,
      "learning_rate": 4.842454394693201e-05,
      "loss": 1.5149,
      "step": 258
    },
    {
      "epoch": 1.9402985074626866,
      "grad_norm": 5.001150608062744,
      "learning_rate": 4.836926478717524e-05,
      "loss": 1.5529,
      "step": 260
    },
    {
      "epoch": 1.955223880597015,
      "grad_norm": 4.259793281555176,
      "learning_rate": 4.831398562741847e-05,
      "loss": 1.7178,
      "step": 262
    },
    {
      "epoch": 1.9701492537313432,
      "grad_norm": 3.2455430030822754,
      "learning_rate": 4.82587064676617e-05,
      "loss": 1.26,
      "step": 264
    },
    {
      "epoch": 1.9850746268656716,
      "grad_norm": 3.899599313735962,
      "learning_rate": 4.820342730790493e-05,
      "loss": 1.5229,
      "step": 266
    },
    {
      "epoch": 2.0,
      "grad_norm": 6.097535610198975,
      "learning_rate": 4.814814814814815e-05,
      "loss": 1.6973,
      "step": 268
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.63,
      "eval_f1_macro": 0.569843507632617,
      "eval_f1_micro": 0.63,
      "eval_f1_weighted": 0.5873017265672064,
      "eval_loss": 1.6746174097061157,
      "eval_precision_macro": 0.5998444366091426,
      "eval_precision_micro": 0.63,
      "eval_precision_weighted": 0.623490675990676,
      "eval_recall_macro": 0.6212885154061624,
      "eval_recall_micro": 0.63,
      "eval_recall_weighted": 0.63,
      "eval_runtime": 35.3294,
      "eval_samples_per_second": 5.661,
      "eval_steps_per_second": 0.198,
      "step": 268
    },
    {
      "epoch": 2.014925373134328,
      "grad_norm": 2.5601863861083984,
      "learning_rate": 4.8092868988391376e-05,
      "loss": 1.0241,
      "step": 270
    },
    {
      "epoch": 2.029850746268657,
      "grad_norm": 4.754575252532959,
      "learning_rate": 4.8037589828634604e-05,
      "loss": 1.5737,
      "step": 272
    },
    {
      "epoch": 2.044776119402985,
      "grad_norm": 4.370357990264893,
      "learning_rate": 4.798231066887783e-05,
      "loss": 1.4961,
      "step": 274
    },
    {
      "epoch": 2.0597014925373136,
      "grad_norm": 8.790416717529297,
      "learning_rate": 4.792703150912106e-05,
      "loss": 1.6075,
      "step": 276
    },
    {
      "epoch": 2.074626865671642,
      "grad_norm": 4.466557502746582,
      "learning_rate": 4.787175234936429e-05,
      "loss": 1.4832,
      "step": 278
    },
    {
      "epoch": 2.08955223880597,
      "grad_norm": 3.5914487838745117,
      "learning_rate": 4.781647318960752e-05,
      "loss": 1.1338,
      "step": 280
    },
    {
      "epoch": 2.1044776119402986,
      "grad_norm": 3.8444533348083496,
      "learning_rate": 4.7761194029850745e-05,
      "loss": 1.2921,
      "step": 282
    },
    {
      "epoch": 2.1194029850746268,
      "grad_norm": 7.032942295074463,
      "learning_rate": 4.770591487009397e-05,
      "loss": 1.5322,
      "step": 284
    },
    {
      "epoch": 2.1343283582089554,
      "grad_norm": 6.563926696777344,
      "learning_rate": 4.76506357103372e-05,
      "loss": 1.2677,
      "step": 286
    },
    {
      "epoch": 2.1492537313432836,
      "grad_norm": 3.413158655166626,
      "learning_rate": 4.7595356550580436e-05,
      "loss": 1.2462,
      "step": 288
    },
    {
      "epoch": 2.1641791044776117,
      "grad_norm": 4.282972812652588,
      "learning_rate": 4.7540077390823664e-05,
      "loss": 1.1632,
      "step": 290
    },
    {
      "epoch": 2.1791044776119404,
      "grad_norm": 11.003588676452637,
      "learning_rate": 4.748479823106689e-05,
      "loss": 1.2687,
      "step": 292
    },
    {
      "epoch": 2.1940298507462686,
      "grad_norm": 4.725019931793213,
      "learning_rate": 4.742951907131012e-05,
      "loss": 1.2352,
      "step": 294
    },
    {
      "epoch": 2.208955223880597,
      "grad_norm": 3.6079459190368652,
      "learning_rate": 4.737423991155335e-05,
      "loss": 1.4994,
      "step": 296
    },
    {
      "epoch": 2.2238805970149254,
      "grad_norm": 3.576037645339966,
      "learning_rate": 4.731896075179658e-05,
      "loss": 1.1007,
      "step": 298
    },
    {
      "epoch": 2.2388059701492535,
      "grad_norm": 3.0327272415161133,
      "learning_rate": 4.7263681592039805e-05,
      "loss": 1.0624,
      "step": 300
    },
    {
      "epoch": 2.253731343283582,
      "grad_norm": 5.710484504699707,
      "learning_rate": 4.720840243228303e-05,
      "loss": 1.2262,
      "step": 302
    },
    {
      "epoch": 2.2686567164179103,
      "grad_norm": 3.059814453125,
      "learning_rate": 4.715312327252626e-05,
      "loss": 1.0292,
      "step": 304
    },
    {
      "epoch": 2.283582089552239,
      "grad_norm": 4.122474193572998,
      "learning_rate": 4.709784411276949e-05,
      "loss": 1.1731,
      "step": 306
    },
    {
      "epoch": 2.298507462686567,
      "grad_norm": 8.998103141784668,
      "learning_rate": 4.704256495301272e-05,
      "loss": 1.5075,
      "step": 308
    },
    {
      "epoch": 2.3134328358208958,
      "grad_norm": 3.947554349899292,
      "learning_rate": 4.6987285793255946e-05,
      "loss": 1.1681,
      "step": 310
    },
    {
      "epoch": 2.328358208955224,
      "grad_norm": 6.467041492462158,
      "learning_rate": 4.693200663349917e-05,
      "loss": 1.2688,
      "step": 312
    },
    {
      "epoch": 2.343283582089552,
      "grad_norm": 6.714489936828613,
      "learning_rate": 4.68767274737424e-05,
      "loss": 1.5061,
      "step": 314
    },
    {
      "epoch": 2.3582089552238807,
      "grad_norm": 4.802642822265625,
      "learning_rate": 4.682144831398563e-05,
      "loss": 1.0498,
      "step": 316
    },
    {
      "epoch": 2.373134328358209,
      "grad_norm": 3.7944350242614746,
      "learning_rate": 4.676616915422886e-05,
      "loss": 1.1943,
      "step": 318
    },
    {
      "epoch": 2.388059701492537,
      "grad_norm": 6.245696067810059,
      "learning_rate": 4.6710889994472086e-05,
      "loss": 0.9827,
      "step": 320
    },
    {
      "epoch": 2.4029850746268657,
      "grad_norm": 3.385528326034546,
      "learning_rate": 4.6655610834715315e-05,
      "loss": 1.2411,
      "step": 322
    },
    {
      "epoch": 2.417910447761194,
      "grad_norm": 4.5617828369140625,
      "learning_rate": 4.660033167495854e-05,
      "loss": 0.947,
      "step": 324
    },
    {
      "epoch": 2.4328358208955225,
      "grad_norm": 6.033749580383301,
      "learning_rate": 4.654505251520177e-05,
      "loss": 1.1135,
      "step": 326
    },
    {
      "epoch": 2.4477611940298507,
      "grad_norm": 4.091092586517334,
      "learning_rate": 4.6489773355445e-05,
      "loss": 1.2219,
      "step": 328
    },
    {
      "epoch": 2.4626865671641793,
      "grad_norm": 4.028707027435303,
      "learning_rate": 4.643449419568823e-05,
      "loss": 0.8016,
      "step": 330
    },
    {
      "epoch": 2.4776119402985075,
      "grad_norm": 13.044775009155273,
      "learning_rate": 4.6379215035931455e-05,
      "loss": 1.0265,
      "step": 332
    },
    {
      "epoch": 2.4925373134328357,
      "grad_norm": 3.4753003120422363,
      "learning_rate": 4.632393587617468e-05,
      "loss": 1.1795,
      "step": 334
    },
    {
      "epoch": 2.5074626865671643,
      "grad_norm": 10.754464149475098,
      "learning_rate": 4.626865671641791e-05,
      "loss": 1.0397,
      "step": 336
    },
    {
      "epoch": 2.5223880597014925,
      "grad_norm": 5.801613807678223,
      "learning_rate": 4.621337755666114e-05,
      "loss": 1.0849,
      "step": 338
    },
    {
      "epoch": 2.5373134328358207,
      "grad_norm": 7.185355186462402,
      "learning_rate": 4.615809839690437e-05,
      "loss": 1.3039,
      "step": 340
    },
    {
      "epoch": 2.5522388059701493,
      "grad_norm": 5.9482622146606445,
      "learning_rate": 4.6102819237147596e-05,
      "loss": 0.9354,
      "step": 342
    },
    {
      "epoch": 2.5671641791044775,
      "grad_norm": 5.785240173339844,
      "learning_rate": 4.6047540077390824e-05,
      "loss": 1.1711,
      "step": 344
    },
    {
      "epoch": 2.582089552238806,
      "grad_norm": 4.588739395141602,
      "learning_rate": 4.599226091763406e-05,
      "loss": 1.2397,
      "step": 346
    },
    {
      "epoch": 2.5970149253731343,
      "grad_norm": 9.58119010925293,
      "learning_rate": 4.593698175787729e-05,
      "loss": 1.0885,
      "step": 348
    },
    {
      "epoch": 2.611940298507463,
      "grad_norm": 7.116174221038818,
      "learning_rate": 4.5881702598120515e-05,
      "loss": 1.4472,
      "step": 350
    },
    {
      "epoch": 2.626865671641791,
      "grad_norm": 4.783851146697998,
      "learning_rate": 4.582642343836374e-05,
      "loss": 1.0596,
      "step": 352
    },
    {
      "epoch": 2.6417910447761193,
      "grad_norm": 8.053864479064941,
      "learning_rate": 4.577114427860697e-05,
      "loss": 1.1109,
      "step": 354
    },
    {
      "epoch": 2.656716417910448,
      "grad_norm": 7.039917945861816,
      "learning_rate": 4.571586511885019e-05,
      "loss": 1.0317,
      "step": 356
    },
    {
      "epoch": 2.671641791044776,
      "grad_norm": 10.328742980957031,
      "learning_rate": 4.566058595909342e-05,
      "loss": 1.3927,
      "step": 358
    },
    {
      "epoch": 2.6865671641791042,
      "grad_norm": 6.044647693634033,
      "learning_rate": 4.560530679933665e-05,
      "loss": 1.6115,
      "step": 360
    },
    {
      "epoch": 2.701492537313433,
      "grad_norm": 5.099875450134277,
      "learning_rate": 4.555002763957988e-05,
      "loss": 0.7492,
      "step": 362
    },
    {
      "epoch": 2.716417910447761,
      "grad_norm": 5.763055324554443,
      "learning_rate": 4.5494748479823105e-05,
      "loss": 1.2244,
      "step": 364
    },
    {
      "epoch": 2.7313432835820897,
      "grad_norm": 7.4969987869262695,
      "learning_rate": 4.5439469320066333e-05,
      "loss": 0.9924,
      "step": 366
    },
    {
      "epoch": 2.746268656716418,
      "grad_norm": 11.879778861999512,
      "learning_rate": 4.538419016030956e-05,
      "loss": 1.3022,
      "step": 368
    },
    {
      "epoch": 2.7611940298507465,
      "grad_norm": 2.6563034057617188,
      "learning_rate": 4.532891100055279e-05,
      "loss": 1.0468,
      "step": 370
    },
    {
      "epoch": 2.7761194029850746,
      "grad_norm": 4.352832317352295,
      "learning_rate": 4.5273631840796025e-05,
      "loss": 0.7805,
      "step": 372
    },
    {
      "epoch": 2.791044776119403,
      "grad_norm": 5.752820014953613,
      "learning_rate": 4.521835268103925e-05,
      "loss": 1.0971,
      "step": 374
    },
    {
      "epoch": 2.8059701492537314,
      "grad_norm": 6.028427600860596,
      "learning_rate": 4.516307352128248e-05,
      "loss": 0.9887,
      "step": 376
    },
    {
      "epoch": 2.8208955223880596,
      "grad_norm": 5.541314601898193,
      "learning_rate": 4.510779436152571e-05,
      "loss": 1.0044,
      "step": 378
    },
    {
      "epoch": 2.835820895522388,
      "grad_norm": 6.708818435668945,
      "learning_rate": 4.505251520176894e-05,
      "loss": 0.9756,
      "step": 380
    },
    {
      "epoch": 2.8507462686567164,
      "grad_norm": 5.191244602203369,
      "learning_rate": 4.4997236042012165e-05,
      "loss": 1.0194,
      "step": 382
    },
    {
      "epoch": 2.8656716417910446,
      "grad_norm": 5.064216613769531,
      "learning_rate": 4.4941956882255393e-05,
      "loss": 1.1003,
      "step": 384
    },
    {
      "epoch": 2.8805970149253732,
      "grad_norm": 4.351062774658203,
      "learning_rate": 4.488667772249862e-05,
      "loss": 1.0402,
      "step": 386
    },
    {
      "epoch": 2.8955223880597014,
      "grad_norm": 9.046034812927246,
      "learning_rate": 4.483139856274185e-05,
      "loss": 1.0788,
      "step": 388
    },
    {
      "epoch": 2.91044776119403,
      "grad_norm": 9.672553062438965,
      "learning_rate": 4.477611940298508e-05,
      "loss": 1.1218,
      "step": 390
    },
    {
      "epoch": 2.925373134328358,
      "grad_norm": 6.787928581237793,
      "learning_rate": 4.4720840243228306e-05,
      "loss": 0.9067,
      "step": 392
    },
    {
      "epoch": 2.9402985074626864,
      "grad_norm": 7.9477152824401855,
      "learning_rate": 4.4665561083471534e-05,
      "loss": 1.0954,
      "step": 394
    },
    {
      "epoch": 2.955223880597015,
      "grad_norm": 6.69403600692749,
      "learning_rate": 4.461028192371476e-05,
      "loss": 1.2409,
      "step": 396
    },
    {
      "epoch": 2.970149253731343,
      "grad_norm": 6.419997692108154,
      "learning_rate": 4.455500276395799e-05,
      "loss": 0.7133,
      "step": 398
    },
    {
      "epoch": 2.9850746268656714,
      "grad_norm": 6.382747173309326,
      "learning_rate": 4.449972360420122e-05,
      "loss": 1.3368,
      "step": 400
    },
    {
      "epoch": 3.0,
      "grad_norm": 6.085859298706055,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.8309,
      "step": 402
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.7,
      "eval_f1_macro": 0.6561413388812769,
      "eval_f1_micro": 0.7,
      "eval_f1_weighted": 0.6696798888246258,
      "eval_loss": 1.1434708833694458,
      "eval_precision_macro": 0.6706204141575659,
      "eval_precision_micro": 0.7,
      "eval_precision_weighted": 0.6840218575284365,
      "eval_recall_macro": 0.6900560224089635,
      "eval_recall_micro": 0.7,
      "eval_recall_weighted": 0.7,
      "eval_runtime": 36.7029,
      "eval_samples_per_second": 5.449,
      "eval_steps_per_second": 0.191,
      "step": 402
    },
    {
      "epoch": 3.014925373134328,
      "grad_norm": 8.274943351745605,
      "learning_rate": 4.4389165284687675e-05,
      "loss": 0.728,
      "step": 404
    },
    {
      "epoch": 3.029850746268657,
      "grad_norm": 3.896862745285034,
      "learning_rate": 4.43338861249309e-05,
      "loss": 0.8955,
      "step": 406
    },
    {
      "epoch": 3.044776119402985,
      "grad_norm": 3.3480184078216553,
      "learning_rate": 4.427860696517413e-05,
      "loss": 0.84,
      "step": 408
    },
    {
      "epoch": 3.0597014925373136,
      "grad_norm": 9.12331485748291,
      "learning_rate": 4.422332780541736e-05,
      "loss": 0.8026,
      "step": 410
    },
    {
      "epoch": 3.074626865671642,
      "grad_norm": 6.628831386566162,
      "learning_rate": 4.416804864566059e-05,
      "loss": 1.0466,
      "step": 412
    },
    {
      "epoch": 3.08955223880597,
      "grad_norm": 2.638613224029541,
      "learning_rate": 4.4112769485903815e-05,
      "loss": 0.6834,
      "step": 414
    },
    {
      "epoch": 3.1044776119402986,
      "grad_norm": 3.1941006183624268,
      "learning_rate": 4.4057490326147043e-05,
      "loss": 0.9236,
      "step": 416
    },
    {
      "epoch": 3.1194029850746268,
      "grad_norm": 4.337372779846191,
      "learning_rate": 4.400221116639027e-05,
      "loss": 0.8447,
      "step": 418
    },
    {
      "epoch": 3.1343283582089554,
      "grad_norm": 8.465522766113281,
      "learning_rate": 4.39469320066335e-05,
      "loss": 0.8768,
      "step": 420
    },
    {
      "epoch": 3.1492537313432836,
      "grad_norm": 4.252687454223633,
      "learning_rate": 4.389165284687673e-05,
      "loss": 0.7148,
      "step": 422
    },
    {
      "epoch": 3.1641791044776117,
      "grad_norm": 5.22479772567749,
      "learning_rate": 4.3836373687119956e-05,
      "loss": 0.7375,
      "step": 424
    },
    {
      "epoch": 3.1791044776119404,
      "grad_norm": 3.702615737915039,
      "learning_rate": 4.3781094527363184e-05,
      "loss": 0.8488,
      "step": 426
    },
    {
      "epoch": 3.1940298507462686,
      "grad_norm": 3.2345383167266846,
      "learning_rate": 4.372581536760641e-05,
      "loss": 0.7861,
      "step": 428
    },
    {
      "epoch": 3.208955223880597,
      "grad_norm": 7.831496238708496,
      "learning_rate": 4.367053620784965e-05,
      "loss": 1.117,
      "step": 430
    },
    {
      "epoch": 3.2238805970149254,
      "grad_norm": 2.40031361579895,
      "learning_rate": 4.3615257048092875e-05,
      "loss": 0.7079,
      "step": 432
    },
    {
      "epoch": 3.2388059701492535,
      "grad_norm": 4.2407050132751465,
      "learning_rate": 4.3559977888336103e-05,
      "loss": 0.7778,
      "step": 434
    },
    {
      "epoch": 3.253731343283582,
      "grad_norm": 3.885312795639038,
      "learning_rate": 4.350469872857933e-05,
      "loss": 1.0815,
      "step": 436
    },
    {
      "epoch": 3.2686567164179103,
      "grad_norm": 3.0180788040161133,
      "learning_rate": 4.344941956882256e-05,
      "loss": 0.6417,
      "step": 438
    },
    {
      "epoch": 3.283582089552239,
      "grad_norm": 6.578612804412842,
      "learning_rate": 4.339414040906579e-05,
      "loss": 0.9152,
      "step": 440
    },
    {
      "epoch": 3.298507462686567,
      "grad_norm": 4.349058151245117,
      "learning_rate": 4.3338861249309016e-05,
      "loss": 0.9936,
      "step": 442
    },
    {
      "epoch": 3.3134328358208958,
      "grad_norm": 7.187576770782471,
      "learning_rate": 4.328358208955224e-05,
      "loss": 0.9757,
      "step": 444
    },
    {
      "epoch": 3.328358208955224,
      "grad_norm": 3.981358528137207,
      "learning_rate": 4.3228302929795465e-05,
      "loss": 0.6527,
      "step": 446
    },
    {
      "epoch": 3.343283582089552,
      "grad_norm": 4.734504222869873,
      "learning_rate": 4.3173023770038694e-05,
      "loss": 0.917,
      "step": 448
    },
    {
      "epoch": 3.3582089552238807,
      "grad_norm": 7.110349655151367,
      "learning_rate": 4.311774461028192e-05,
      "loss": 0.7495,
      "step": 450
    },
    {
      "epoch": 3.373134328358209,
      "grad_norm": 4.323439121246338,
      "learning_rate": 4.306246545052515e-05,
      "loss": 0.7577,
      "step": 452
    },
    {
      "epoch": 3.388059701492537,
      "grad_norm": 6.730025768280029,
      "learning_rate": 4.300718629076838e-05,
      "loss": 1.2703,
      "step": 454
    },
    {
      "epoch": 3.4029850746268657,
      "grad_norm": 10.35930347442627,
      "learning_rate": 4.2951907131011606e-05,
      "loss": 0.4539,
      "step": 456
    },
    {
      "epoch": 3.417910447761194,
      "grad_norm": 3.9190192222595215,
      "learning_rate": 4.289662797125484e-05,
      "loss": 0.8863,
      "step": 458
    },
    {
      "epoch": 3.4328358208955225,
      "grad_norm": 6.045052528381348,
      "learning_rate": 4.284134881149807e-05,
      "loss": 0.8221,
      "step": 460
    },
    {
      "epoch": 3.4477611940298507,
      "grad_norm": 9.53062629699707,
      "learning_rate": 4.27860696517413e-05,
      "loss": 0.9652,
      "step": 462
    },
    {
      "epoch": 3.4626865671641793,
      "grad_norm": 5.4133710861206055,
      "learning_rate": 4.2730790491984525e-05,
      "loss": 0.6179,
      "step": 464
    },
    {
      "epoch": 3.4776119402985075,
      "grad_norm": 5.08922004699707,
      "learning_rate": 4.2675511332227754e-05,
      "loss": 0.8349,
      "step": 466
    },
    {
      "epoch": 3.4925373134328357,
      "grad_norm": 9.440414428710938,
      "learning_rate": 4.262023217247098e-05,
      "loss": 0.7352,
      "step": 468
    },
    {
      "epoch": 3.5074626865671643,
      "grad_norm": 5.086453437805176,
      "learning_rate": 4.256495301271421e-05,
      "loss": 0.6674,
      "step": 470
    },
    {
      "epoch": 3.5223880597014925,
      "grad_norm": 5.283850193023682,
      "learning_rate": 4.250967385295744e-05,
      "loss": 0.9129,
      "step": 472
    },
    {
      "epoch": 3.5373134328358207,
      "grad_norm": 6.586909770965576,
      "learning_rate": 4.2454394693200666e-05,
      "loss": 0.4512,
      "step": 474
    },
    {
      "epoch": 3.5522388059701493,
      "grad_norm": 6.328141212463379,
      "learning_rate": 4.2399115533443894e-05,
      "loss": 0.9986,
      "step": 476
    },
    {
      "epoch": 3.5671641791044775,
      "grad_norm": 11.822186470031738,
      "learning_rate": 4.234383637368712e-05,
      "loss": 0.6238,
      "step": 478
    },
    {
      "epoch": 3.582089552238806,
      "grad_norm": 12.303596496582031,
      "learning_rate": 4.228855721393035e-05,
      "loss": 0.666,
      "step": 480
    },
    {
      "epoch": 3.5970149253731343,
      "grad_norm": 10.991721153259277,
      "learning_rate": 4.223327805417358e-05,
      "loss": 0.8176,
      "step": 482
    },
    {
      "epoch": 3.611940298507463,
      "grad_norm": 4.936182022094727,
      "learning_rate": 4.217799889441681e-05,
      "loss": 0.6964,
      "step": 484
    },
    {
      "epoch": 3.626865671641791,
      "grad_norm": 4.455245494842529,
      "learning_rate": 4.2122719734660035e-05,
      "loss": 0.8681,
      "step": 486
    },
    {
      "epoch": 3.6417910447761193,
      "grad_norm": 4.1669206619262695,
      "learning_rate": 4.206744057490326e-05,
      "loss": 0.8556,
      "step": 488
    },
    {
      "epoch": 3.656716417910448,
      "grad_norm": 2.757246732711792,
      "learning_rate": 4.201216141514649e-05,
      "loss": 0.6557,
      "step": 490
    },
    {
      "epoch": 3.671641791044776,
      "grad_norm": 4.065201282501221,
      "learning_rate": 4.195688225538972e-05,
      "loss": 0.7899,
      "step": 492
    },
    {
      "epoch": 3.6865671641791042,
      "grad_norm": 5.666295528411865,
      "learning_rate": 4.190160309563295e-05,
      "loss": 0.7589,
      "step": 494
    },
    {
      "epoch": 3.701492537313433,
      "grad_norm": 4.929719924926758,
      "learning_rate": 4.1846323935876176e-05,
      "loss": 0.6709,
      "step": 496
    },
    {
      "epoch": 3.716417910447761,
      "grad_norm": 6.261268138885498,
      "learning_rate": 4.1791044776119404e-05,
      "loss": 0.4849,
      "step": 498
    },
    {
      "epoch": 3.7313432835820897,
      "grad_norm": 1.9559329748153687,
      "learning_rate": 4.173576561636263e-05,
      "loss": 0.4897,
      "step": 500
    },
    {
      "epoch": 3.746268656716418,
      "grad_norm": 3.6171419620513916,
      "learning_rate": 4.168048645660586e-05,
      "loss": 0.4541,
      "step": 502
    },
    {
      "epoch": 3.7611940298507465,
      "grad_norm": 5.1620988845825195,
      "learning_rate": 4.162520729684909e-05,
      "loss": 0.4952,
      "step": 504
    },
    {
      "epoch": 3.7761194029850746,
      "grad_norm": 6.901045322418213,
      "learning_rate": 4.1569928137092316e-05,
      "loss": 0.8728,
      "step": 506
    },
    {
      "epoch": 3.791044776119403,
      "grad_norm": 6.784924507141113,
      "learning_rate": 4.1514648977335544e-05,
      "loss": 1.0379,
      "step": 508
    },
    {
      "epoch": 3.8059701492537314,
      "grad_norm": 3.829332113265991,
      "learning_rate": 4.145936981757877e-05,
      "loss": 0.8488,
      "step": 510
    },
    {
      "epoch": 3.8208955223880596,
      "grad_norm": 8.228715896606445,
      "learning_rate": 4.1404090657822e-05,
      "loss": 0.4737,
      "step": 512
    },
    {
      "epoch": 3.835820895522388,
      "grad_norm": 7.089122295379639,
      "learning_rate": 4.134881149806523e-05,
      "loss": 0.8644,
      "step": 514
    },
    {
      "epoch": 3.8507462686567164,
      "grad_norm": 6.432941913604736,
      "learning_rate": 4.1293532338308464e-05,
      "loss": 0.5153,
      "step": 516
    },
    {
      "epoch": 3.8656716417910446,
      "grad_norm": 8.107361793518066,
      "learning_rate": 4.123825317855169e-05,
      "loss": 0.6265,
      "step": 518
    },
    {
      "epoch": 3.8805970149253732,
      "grad_norm": 4.366127014160156,
      "learning_rate": 4.118297401879492e-05,
      "loss": 0.9007,
      "step": 520
    },
    {
      "epoch": 3.8955223880597014,
      "grad_norm": 3.596374988555908,
      "learning_rate": 4.112769485903815e-05,
      "loss": 0.7653,
      "step": 522
    },
    {
      "epoch": 3.91044776119403,
      "grad_norm": 6.282411098480225,
      "learning_rate": 4.1072415699281376e-05,
      "loss": 0.8757,
      "step": 524
    },
    {
      "epoch": 3.925373134328358,
      "grad_norm": 6.818746566772461,
      "learning_rate": 4.1017136539524604e-05,
      "loss": 0.7569,
      "step": 526
    },
    {
      "epoch": 3.9402985074626864,
      "grad_norm": 6.6983113288879395,
      "learning_rate": 4.096185737976783e-05,
      "loss": 0.79,
      "step": 528
    },
    {
      "epoch": 3.955223880597015,
      "grad_norm": 3.532710075378418,
      "learning_rate": 4.090657822001106e-05,
      "loss": 0.6981,
      "step": 530
    },
    {
      "epoch": 3.970149253731343,
      "grad_norm": 6.195641040802002,
      "learning_rate": 4.085129906025428e-05,
      "loss": 0.8261,
      "step": 532
    },
    {
      "epoch": 3.9850746268656714,
      "grad_norm": 8.55378246307373,
      "learning_rate": 4.079601990049751e-05,
      "loss": 0.7682,
      "step": 534
    },
    {
      "epoch": 4.0,
      "grad_norm": 7.6657891273498535,
      "learning_rate": 4.074074074074074e-05,
      "loss": 0.6755,
      "step": 536
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.74,
      "eval_f1_macro": 0.6978798162621692,
      "eval_f1_micro": 0.74,
      "eval_f1_weighted": 0.7101990093240095,
      "eval_loss": 0.93646240234375,
      "eval_precision_macro": 0.7079046770223241,
      "eval_precision_micro": 0.74,
      "eval_precision_weighted": 0.719476911976912,
      "eval_recall_macro": 0.7278711484593837,
      "eval_recall_micro": 0.74,
      "eval_recall_weighted": 0.74,
      "eval_runtime": 40.7905,
      "eval_samples_per_second": 4.903,
      "eval_steps_per_second": 0.172,
      "step": 536
    },
    {
      "epoch": 4.014925373134329,
      "grad_norm": 3.920659065246582,
      "learning_rate": 4.0685461580983966e-05,
      "loss": 0.5528,
      "step": 538
    },
    {
      "epoch": 4.029850746268656,
      "grad_norm": 5.147222995758057,
      "learning_rate": 4.0630182421227194e-05,
      "loss": 1.0055,
      "step": 540
    },
    {
      "epoch": 4.044776119402985,
      "grad_norm": 2.260542869567871,
      "learning_rate": 4.057490326147043e-05,
      "loss": 0.6422,
      "step": 542
    },
    {
      "epoch": 4.059701492537314,
      "grad_norm": 5.286315441131592,
      "learning_rate": 4.051962410171366e-05,
      "loss": 0.5399,
      "step": 544
    },
    {
      "epoch": 4.074626865671641,
      "grad_norm": 8.412203788757324,
      "learning_rate": 4.0464344941956886e-05,
      "loss": 0.6129,
      "step": 546
    },
    {
      "epoch": 4.08955223880597,
      "grad_norm": 6.18348503112793,
      "learning_rate": 4.0409065782200114e-05,
      "loss": 0.5879,
      "step": 548
    },
    {
      "epoch": 4.104477611940299,
      "grad_norm": 2.4661457538604736,
      "learning_rate": 4.035378662244334e-05,
      "loss": 0.5941,
      "step": 550
    },
    {
      "epoch": 4.119402985074627,
      "grad_norm": 4.261096477508545,
      "learning_rate": 4.029850746268657e-05,
      "loss": 0.9773,
      "step": 552
    },
    {
      "epoch": 4.134328358208955,
      "grad_norm": 6.717421054840088,
      "learning_rate": 4.02432283029298e-05,
      "loss": 0.7255,
      "step": 554
    },
    {
      "epoch": 4.149253731343284,
      "grad_norm": 4.847568511962891,
      "learning_rate": 4.0187949143173026e-05,
      "loss": 0.6345,
      "step": 556
    },
    {
      "epoch": 4.164179104477612,
      "grad_norm": 3.6066529750823975,
      "learning_rate": 4.0132669983416254e-05,
      "loss": 0.4784,
      "step": 558
    },
    {
      "epoch": 4.17910447761194,
      "grad_norm": 1.8522593975067139,
      "learning_rate": 4.007739082365948e-05,
      "loss": 0.4678,
      "step": 560
    },
    {
      "epoch": 4.1940298507462686,
      "grad_norm": 7.491433620452881,
      "learning_rate": 4.002211166390271e-05,
      "loss": 0.6955,
      "step": 562
    },
    {
      "epoch": 4.208955223880597,
      "grad_norm": 3.9711689949035645,
      "learning_rate": 3.996683250414594e-05,
      "loss": 0.7432,
      "step": 564
    },
    {
      "epoch": 4.223880597014926,
      "grad_norm": 3.5602264404296875,
      "learning_rate": 3.991155334438917e-05,
      "loss": 0.6618,
      "step": 566
    },
    {
      "epoch": 4.2388059701492535,
      "grad_norm": 3.383795976638794,
      "learning_rate": 3.9856274184632395e-05,
      "loss": 0.6079,
      "step": 568
    },
    {
      "epoch": 4.253731343283582,
      "grad_norm": 2.320563554763794,
      "learning_rate": 3.980099502487562e-05,
      "loss": 0.5248,
      "step": 570
    },
    {
      "epoch": 4.268656716417911,
      "grad_norm": 3.110429048538208,
      "learning_rate": 3.974571586511885e-05,
      "loss": 0.3517,
      "step": 572
    },
    {
      "epoch": 4.2835820895522385,
      "grad_norm": 6.524468421936035,
      "learning_rate": 3.9690436705362086e-05,
      "loss": 0.6481,
      "step": 574
    },
    {
      "epoch": 4.298507462686567,
      "grad_norm": 3.3383541107177734,
      "learning_rate": 3.9635157545605314e-05,
      "loss": 0.6297,
      "step": 576
    },
    {
      "epoch": 4.313432835820896,
      "grad_norm": 1.686207890510559,
      "learning_rate": 3.9579878385848536e-05,
      "loss": 0.5142,
      "step": 578
    },
    {
      "epoch": 4.3283582089552235,
      "grad_norm": 9.0740385055542,
      "learning_rate": 3.9524599226091764e-05,
      "loss": 0.478,
      "step": 580
    },
    {
      "epoch": 4.343283582089552,
      "grad_norm": 8.83989143371582,
      "learning_rate": 3.946932006633499e-05,
      "loss": 0.7348,
      "step": 582
    },
    {
      "epoch": 4.358208955223881,
      "grad_norm": 9.292248725891113,
      "learning_rate": 3.941404090657822e-05,
      "loss": 0.6026,
      "step": 584
    },
    {
      "epoch": 4.373134328358209,
      "grad_norm": 4.79518461227417,
      "learning_rate": 3.935876174682145e-05,
      "loss": 0.818,
      "step": 586
    },
    {
      "epoch": 4.388059701492537,
      "grad_norm": 6.292643070220947,
      "learning_rate": 3.9303482587064676e-05,
      "loss": 0.5522,
      "step": 588
    },
    {
      "epoch": 4.402985074626866,
      "grad_norm": 4.075753688812256,
      "learning_rate": 3.9248203427307905e-05,
      "loss": 0.5259,
      "step": 590
    },
    {
      "epoch": 4.417910447761194,
      "grad_norm": 2.985786199569702,
      "learning_rate": 3.919292426755113e-05,
      "loss": 0.4218,
      "step": 592
    },
    {
      "epoch": 4.432835820895522,
      "grad_norm": 4.570261478424072,
      "learning_rate": 3.913764510779436e-05,
      "loss": 0.4735,
      "step": 594
    },
    {
      "epoch": 4.447761194029851,
      "grad_norm": 2.8057727813720703,
      "learning_rate": 3.908236594803759e-05,
      "loss": 0.4241,
      "step": 596
    },
    {
      "epoch": 4.462686567164179,
      "grad_norm": 7.44076681137085,
      "learning_rate": 3.902708678828082e-05,
      "loss": 0.4751,
      "step": 598
    },
    {
      "epoch": 4.477611940298507,
      "grad_norm": 4.970695972442627,
      "learning_rate": 3.8971807628524045e-05,
      "loss": 0.5297,
      "step": 600
    },
    {
      "epoch": 4.492537313432836,
      "grad_norm": 3.2911581993103027,
      "learning_rate": 3.891652846876728e-05,
      "loss": 0.4756,
      "step": 602
    },
    {
      "epoch": 4.507462686567164,
      "grad_norm": 6.148906230926514,
      "learning_rate": 3.886124930901051e-05,
      "loss": 0.5251,
      "step": 604
    },
    {
      "epoch": 4.522388059701493,
      "grad_norm": 4.234598636627197,
      "learning_rate": 3.8805970149253736e-05,
      "loss": 0.3342,
      "step": 606
    },
    {
      "epoch": 4.537313432835821,
      "grad_norm": 4.786794662475586,
      "learning_rate": 3.8750690989496964e-05,
      "loss": 0.6528,
      "step": 608
    },
    {
      "epoch": 4.552238805970149,
      "grad_norm": 8.322453498840332,
      "learning_rate": 3.869541182974019e-05,
      "loss": 0.5658,
      "step": 610
    },
    {
      "epoch": 4.567164179104478,
      "grad_norm": 6.955239772796631,
      "learning_rate": 3.864013266998342e-05,
      "loss": 0.5496,
      "step": 612
    },
    {
      "epoch": 4.582089552238806,
      "grad_norm": 4.89069938659668,
      "learning_rate": 3.858485351022665e-05,
      "loss": 0.4191,
      "step": 614
    },
    {
      "epoch": 4.597014925373134,
      "grad_norm": 9.381403923034668,
      "learning_rate": 3.852957435046988e-05,
      "loss": 0.5993,
      "step": 616
    },
    {
      "epoch": 4.611940298507463,
      "grad_norm": 4.977263450622559,
      "learning_rate": 3.8474295190713105e-05,
      "loss": 0.569,
      "step": 618
    },
    {
      "epoch": 4.6268656716417915,
      "grad_norm": 3.183396816253662,
      "learning_rate": 3.841901603095633e-05,
      "loss": 0.5183,
      "step": 620
    },
    {
      "epoch": 4.641791044776119,
      "grad_norm": 3.1676368713378906,
      "learning_rate": 3.8363736871199555e-05,
      "loss": 0.4647,
      "step": 622
    },
    {
      "epoch": 4.656716417910448,
      "grad_norm": 9.969623565673828,
      "learning_rate": 3.830845771144278e-05,
      "loss": 0.4468,
      "step": 624
    },
    {
      "epoch": 4.6716417910447765,
      "grad_norm": 11.773579597473145,
      "learning_rate": 3.825317855168601e-05,
      "loss": 0.477,
      "step": 626
    },
    {
      "epoch": 4.686567164179104,
      "grad_norm": 3.361043691635132,
      "learning_rate": 3.8197899391929246e-05,
      "loss": 0.4533,
      "step": 628
    },
    {
      "epoch": 4.701492537313433,
      "grad_norm": 6.481675148010254,
      "learning_rate": 3.8142620232172474e-05,
      "loss": 0.5998,
      "step": 630
    },
    {
      "epoch": 4.7164179104477615,
      "grad_norm": 10.961061477661133,
      "learning_rate": 3.80873410724157e-05,
      "loss": 0.4246,
      "step": 632
    },
    {
      "epoch": 4.731343283582089,
      "grad_norm": 11.879002571105957,
      "learning_rate": 3.803206191265893e-05,
      "loss": 0.4113,
      "step": 634
    },
    {
      "epoch": 4.746268656716418,
      "grad_norm": 6.598226547241211,
      "learning_rate": 3.797678275290216e-05,
      "loss": 0.4986,
      "step": 636
    },
    {
      "epoch": 4.7611940298507465,
      "grad_norm": 9.366608619689941,
      "learning_rate": 3.7921503593145386e-05,
      "loss": 0.3439,
      "step": 638
    },
    {
      "epoch": 4.776119402985074,
      "grad_norm": 11.350964546203613,
      "learning_rate": 3.7866224433388615e-05,
      "loss": 0.8422,
      "step": 640
    },
    {
      "epoch": 4.791044776119403,
      "grad_norm": 8.099052429199219,
      "learning_rate": 3.781094527363184e-05,
      "loss": 0.687,
      "step": 642
    },
    {
      "epoch": 4.8059701492537314,
      "grad_norm": 5.031801223754883,
      "learning_rate": 3.775566611387507e-05,
      "loss": 0.484,
      "step": 644
    },
    {
      "epoch": 4.82089552238806,
      "grad_norm": 3.5851833820343018,
      "learning_rate": 3.77003869541183e-05,
      "loss": 0.6037,
      "step": 646
    },
    {
      "epoch": 4.835820895522388,
      "grad_norm": 14.049162864685059,
      "learning_rate": 3.764510779436153e-05,
      "loss": 0.5521,
      "step": 648
    },
    {
      "epoch": 4.850746268656716,
      "grad_norm": 5.086363315582275,
      "learning_rate": 3.7589828634604755e-05,
      "loss": 0.5173,
      "step": 650
    },
    {
      "epoch": 4.865671641791045,
      "grad_norm": 2.833794355392456,
      "learning_rate": 3.753454947484798e-05,
      "loss": 0.7762,
      "step": 652
    },
    {
      "epoch": 4.880597014925373,
      "grad_norm": 2.7670750617980957,
      "learning_rate": 3.747927031509121e-05,
      "loss": 0.5592,
      "step": 654
    },
    {
      "epoch": 4.895522388059701,
      "grad_norm": 2.6060335636138916,
      "learning_rate": 3.742399115533444e-05,
      "loss": 0.2484,
      "step": 656
    },
    {
      "epoch": 4.91044776119403,
      "grad_norm": 5.139571666717529,
      "learning_rate": 3.736871199557767e-05,
      "loss": 0.5157,
      "step": 658
    },
    {
      "epoch": 4.925373134328359,
      "grad_norm": 3.047414779663086,
      "learning_rate": 3.73134328358209e-05,
      "loss": 0.4967,
      "step": 660
    },
    {
      "epoch": 4.940298507462686,
      "grad_norm": 7.455503940582275,
      "learning_rate": 3.725815367606413e-05,
      "loss": 0.4034,
      "step": 662
    },
    {
      "epoch": 4.955223880597015,
      "grad_norm": 12.100015640258789,
      "learning_rate": 3.720287451630736e-05,
      "loss": 0.5245,
      "step": 664
    },
    {
      "epoch": 4.970149253731344,
      "grad_norm": 3.1610209941864014,
      "learning_rate": 3.714759535655058e-05,
      "loss": 0.6381,
      "step": 666
    },
    {
      "epoch": 4.985074626865671,
      "grad_norm": 3.781731367111206,
      "learning_rate": 3.709231619679381e-05,
      "loss": 0.5727,
      "step": 668
    },
    {
      "epoch": 5.0,
      "grad_norm": 1.7648042440414429,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.3107,
      "step": 670
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.725,
      "eval_f1_macro": 0.67604038118744,
      "eval_f1_micro": 0.725,
      "eval_f1_weighted": 0.6927145770895771,
      "eval_loss": 1.0104252099990845,
      "eval_precision_macro": 0.7111726508785332,
      "eval_precision_micro": 0.725,
      "eval_precision_weighted": 0.7247546897546897,
      "eval_recall_macro": 0.7092436974789916,
      "eval_recall_micro": 0.725,
      "eval_recall_weighted": 0.725,
      "eval_runtime": 39.3596,
      "eval_samples_per_second": 5.081,
      "eval_steps_per_second": 0.178,
      "step": 670
    },
    {
      "epoch": 5.014925373134329,
      "grad_norm": 1.6856744289398193,
      "learning_rate": 3.6981757877280265e-05,
      "loss": 0.289,
      "step": 672
    },
    {
      "epoch": 5.029850746268656,
      "grad_norm": 2.8615660667419434,
      "learning_rate": 3.692647871752349e-05,
      "loss": 0.4746,
      "step": 674
    },
    {
      "epoch": 5.044776119402985,
      "grad_norm": 6.883438587188721,
      "learning_rate": 3.687119955776672e-05,
      "loss": 0.6167,
      "step": 676
    },
    {
      "epoch": 5.059701492537314,
      "grad_norm": 7.769782543182373,
      "learning_rate": 3.681592039800995e-05,
      "loss": 0.3471,
      "step": 678
    },
    {
      "epoch": 5.074626865671641,
      "grad_norm": 3.5327107906341553,
      "learning_rate": 3.676064123825318e-05,
      "loss": 0.4453,
      "step": 680
    },
    {
      "epoch": 5.08955223880597,
      "grad_norm": 3.9162540435791016,
      "learning_rate": 3.6705362078496405e-05,
      "loss": 0.4471,
      "step": 682
    },
    {
      "epoch": 5.104477611940299,
      "grad_norm": 2.9827005863189697,
      "learning_rate": 3.6650082918739633e-05,
      "loss": 0.344,
      "step": 684
    },
    {
      "epoch": 5.119402985074627,
      "grad_norm": 5.053178310394287,
      "learning_rate": 3.659480375898287e-05,
      "loss": 0.3973,
      "step": 686
    },
    {
      "epoch": 5.134328358208955,
      "grad_norm": 5.1265716552734375,
      "learning_rate": 3.6539524599226097e-05,
      "loss": 0.3783,
      "step": 688
    },
    {
      "epoch": 5.149253731343284,
      "grad_norm": 3.368522882461548,
      "learning_rate": 3.6484245439469325e-05,
      "loss": 0.5253,
      "step": 690
    },
    {
      "epoch": 5.164179104477612,
      "grad_norm": 5.44252872467041,
      "learning_rate": 3.642896627971255e-05,
      "loss": 0.54,
      "step": 692
    },
    {
      "epoch": 5.17910447761194,
      "grad_norm": 7.313326358795166,
      "learning_rate": 3.637368711995578e-05,
      "loss": 0.3734,
      "step": 694
    },
    {
      "epoch": 5.1940298507462686,
      "grad_norm": 3.947737693786621,
      "learning_rate": 3.631840796019901e-05,
      "loss": 0.4204,
      "step": 696
    },
    {
      "epoch": 5.208955223880597,
      "grad_norm": 2.3334567546844482,
      "learning_rate": 3.626312880044224e-05,
      "loss": 0.3622,
      "step": 698
    },
    {
      "epoch": 5.223880597014926,
      "grad_norm": 4.052284240722656,
      "learning_rate": 3.6207849640685465e-05,
      "loss": 0.2778,
      "step": 700
    },
    {
      "epoch": 5.2388059701492535,
      "grad_norm": 11.430601119995117,
      "learning_rate": 3.6152570480928693e-05,
      "loss": 0.6734,
      "step": 702
    },
    {
      "epoch": 5.253731343283582,
      "grad_norm": 5.606107234954834,
      "learning_rate": 3.609729132117192e-05,
      "loss": 0.4028,
      "step": 704
    },
    {
      "epoch": 5.268656716417911,
      "grad_norm": 13.239790916442871,
      "learning_rate": 3.604201216141515e-05,
      "loss": 0.6932,
      "step": 706
    },
    {
      "epoch": 5.2835820895522385,
      "grad_norm": 2.727085590362549,
      "learning_rate": 3.598673300165838e-05,
      "loss": 0.6325,
      "step": 708
    },
    {
      "epoch": 5.298507462686567,
      "grad_norm": 4.920370101928711,
      "learning_rate": 3.59314538419016e-05,
      "loss": 0.215,
      "step": 710
    },
    {
      "epoch": 5.313432835820896,
      "grad_norm": 3.629782199859619,
      "learning_rate": 3.5876174682144834e-05,
      "loss": 0.5572,
      "step": 712
    },
    {
      "epoch": 5.3283582089552235,
      "grad_norm": 5.6093668937683105,
      "learning_rate": 3.582089552238806e-05,
      "loss": 0.7121,
      "step": 714
    },
    {
      "epoch": 5.343283582089552,
      "grad_norm": 6.661886215209961,
      "learning_rate": 3.576561636263129e-05,
      "loss": 0.4479,
      "step": 716
    },
    {
      "epoch": 5.358208955223881,
      "grad_norm": 10.350639343261719,
      "learning_rate": 3.571033720287452e-05,
      "loss": 0.5578,
      "step": 718
    },
    {
      "epoch": 5.373134328358209,
      "grad_norm": 6.086325168609619,
      "learning_rate": 3.565505804311775e-05,
      "loss": 0.4269,
      "step": 720
    },
    {
      "epoch": 5.388059701492537,
      "grad_norm": 6.263159275054932,
      "learning_rate": 3.5599778883360975e-05,
      "loss": 0.5466,
      "step": 722
    },
    {
      "epoch": 5.402985074626866,
      "grad_norm": 6.2910990715026855,
      "learning_rate": 3.55444997236042e-05,
      "loss": 0.3659,
      "step": 724
    },
    {
      "epoch": 5.417910447761194,
      "grad_norm": 4.90096378326416,
      "learning_rate": 3.548922056384743e-05,
      "loss": 0.6204,
      "step": 726
    },
    {
      "epoch": 5.432835820895522,
      "grad_norm": 3.6040308475494385,
      "learning_rate": 3.543394140409066e-05,
      "loss": 0.5589,
      "step": 728
    },
    {
      "epoch": 5.447761194029851,
      "grad_norm": 14.020445823669434,
      "learning_rate": 3.537866224433389e-05,
      "loss": 0.5483,
      "step": 730
    },
    {
      "epoch": 5.462686567164179,
      "grad_norm": 2.3177435398101807,
      "learning_rate": 3.5323383084577115e-05,
      "loss": 0.3787,
      "step": 732
    },
    {
      "epoch": 5.477611940298507,
      "grad_norm": 2.891357183456421,
      "learning_rate": 3.5268103924820344e-05,
      "loss": 0.347,
      "step": 734
    },
    {
      "epoch": 5.492537313432836,
      "grad_norm": 11.490731239318848,
      "learning_rate": 3.521282476506357e-05,
      "loss": 0.5475,
      "step": 736
    },
    {
      "epoch": 5.507462686567164,
      "grad_norm": 2.3620495796203613,
      "learning_rate": 3.51575456053068e-05,
      "loss": 0.4033,
      "step": 738
    },
    {
      "epoch": 5.522388059701493,
      "grad_norm": 3.9230000972747803,
      "learning_rate": 3.510226644555003e-05,
      "loss": 0.4246,
      "step": 740
    },
    {
      "epoch": 5.537313432835821,
      "grad_norm": 3.205672025680542,
      "learning_rate": 3.5046987285793256e-05,
      "loss": 0.4462,
      "step": 742
    },
    {
      "epoch": 5.552238805970149,
      "grad_norm": 5.685773849487305,
      "learning_rate": 3.499170812603649e-05,
      "loss": 0.379,
      "step": 744
    },
    {
      "epoch": 5.567164179104478,
      "grad_norm": 5.982370376586914,
      "learning_rate": 3.493642896627972e-05,
      "loss": 0.4962,
      "step": 746
    },
    {
      "epoch": 5.582089552238806,
      "grad_norm": 13.228385925292969,
      "learning_rate": 3.488114980652295e-05,
      "loss": 0.3345,
      "step": 748
    },
    {
      "epoch": 5.597014925373134,
      "grad_norm": 5.765232086181641,
      "learning_rate": 3.4825870646766175e-05,
      "loss": 0.4973,
      "step": 750
    },
    {
      "epoch": 5.611940298507463,
      "grad_norm": 1.701096534729004,
      "learning_rate": 3.4770591487009404e-05,
      "loss": 0.3608,
      "step": 752
    },
    {
      "epoch": 5.6268656716417915,
      "grad_norm": 14.079666137695312,
      "learning_rate": 3.4715312327252625e-05,
      "loss": 0.5854,
      "step": 754
    },
    {
      "epoch": 5.641791044776119,
      "grad_norm": 3.6387109756469727,
      "learning_rate": 3.466003316749585e-05,
      "loss": 0.2784,
      "step": 756
    },
    {
      "epoch": 5.656716417910448,
      "grad_norm": 1.6772111654281616,
      "learning_rate": 3.460475400773908e-05,
      "loss": 0.2205,
      "step": 758
    },
    {
      "epoch": 5.6716417910447765,
      "grad_norm": 5.0109405517578125,
      "learning_rate": 3.454947484798231e-05,
      "loss": 0.4189,
      "step": 760
    },
    {
      "epoch": 5.686567164179104,
      "grad_norm": 4.438418865203857,
      "learning_rate": 3.449419568822554e-05,
      "loss": 0.1976,
      "step": 762
    },
    {
      "epoch": 5.701492537313433,
      "grad_norm": 4.297623157501221,
      "learning_rate": 3.4438916528468766e-05,
      "loss": 0.3569,
      "step": 764
    },
    {
      "epoch": 5.7164179104477615,
      "grad_norm": 7.163605213165283,
      "learning_rate": 3.4383637368711994e-05,
      "loss": 0.2972,
      "step": 766
    },
    {
      "epoch": 5.731343283582089,
      "grad_norm": 3.6532185077667236,
      "learning_rate": 3.432835820895522e-05,
      "loss": 0.1953,
      "step": 768
    },
    {
      "epoch": 5.746268656716418,
      "grad_norm": 10.143959045410156,
      "learning_rate": 3.427307904919845e-05,
      "loss": 0.4292,
      "step": 770
    },
    {
      "epoch": 5.7611940298507465,
      "grad_norm": 7.471317768096924,
      "learning_rate": 3.4217799889441685e-05,
      "loss": 0.472,
      "step": 772
    },
    {
      "epoch": 5.776119402985074,
      "grad_norm": 3.14628529548645,
      "learning_rate": 3.416252072968491e-05,
      "loss": 0.4863,
      "step": 774
    },
    {
      "epoch": 5.791044776119403,
      "grad_norm": 2.2698168754577637,
      "learning_rate": 3.410724156992814e-05,
      "loss": 0.2093,
      "step": 776
    },
    {
      "epoch": 5.8059701492537314,
      "grad_norm": 3.6853973865509033,
      "learning_rate": 3.405196241017137e-05,
      "loss": 0.5749,
      "step": 778
    },
    {
      "epoch": 5.82089552238806,
      "grad_norm": 9.027390480041504,
      "learning_rate": 3.39966832504146e-05,
      "loss": 0.8224,
      "step": 780
    },
    {
      "epoch": 5.835820895522388,
      "grad_norm": 4.245992660522461,
      "learning_rate": 3.3941404090657825e-05,
      "loss": 0.2923,
      "step": 782
    },
    {
      "epoch": 5.850746268656716,
      "grad_norm": 10.722375869750977,
      "learning_rate": 3.3886124930901054e-05,
      "loss": 0.7797,
      "step": 784
    },
    {
      "epoch": 5.865671641791045,
      "grad_norm": 4.297248363494873,
      "learning_rate": 3.383084577114428e-05,
      "loss": 0.5619,
      "step": 786
    },
    {
      "epoch": 5.880597014925373,
      "grad_norm": 4.837419509887695,
      "learning_rate": 3.377556661138751e-05,
      "loss": 0.3356,
      "step": 788
    },
    {
      "epoch": 5.895522388059701,
      "grad_norm": 1.2375975847244263,
      "learning_rate": 3.372028745163074e-05,
      "loss": 0.1544,
      "step": 790
    },
    {
      "epoch": 5.91044776119403,
      "grad_norm": 12.756596565246582,
      "learning_rate": 3.3665008291873966e-05,
      "loss": 0.3809,
      "step": 792
    },
    {
      "epoch": 5.925373134328359,
      "grad_norm": 3.4860129356384277,
      "learning_rate": 3.3609729132117194e-05,
      "loss": 0.3144,
      "step": 794
    },
    {
      "epoch": 5.940298507462686,
      "grad_norm": 7.757724761962891,
      "learning_rate": 3.355444997236042e-05,
      "loss": 0.3695,
      "step": 796
    },
    {
      "epoch": 5.955223880597015,
      "grad_norm": 2.4491782188415527,
      "learning_rate": 3.349917081260365e-05,
      "loss": 0.275,
      "step": 798
    },
    {
      "epoch": 5.970149253731344,
      "grad_norm": 3.097658395767212,
      "learning_rate": 3.344389165284688e-05,
      "loss": 0.5223,
      "step": 800
    },
    {
      "epoch": 5.985074626865671,
      "grad_norm": 3.9302120208740234,
      "learning_rate": 3.338861249309011e-05,
      "loss": 0.2202,
      "step": 802
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.732783794403076,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.4422,
      "step": 804
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.76,
      "eval_f1_macro": 0.7178212262554823,
      "eval_f1_micro": 0.76,
      "eval_f1_weighted": 0.7341365415630122,
      "eval_loss": 0.890067458152771,
      "eval_precision_macro": 0.7468954248366012,
      "eval_precision_micro": 0.76,
      "eval_precision_weighted": 0.766861111111111,
      "eval_recall_macro": 0.7448179271708684,
      "eval_recall_micro": 0.76,
      "eval_recall_weighted": 0.76,
      "eval_runtime": 37.7852,
      "eval_samples_per_second": 5.293,
      "eval_steps_per_second": 0.185,
      "step": 804
    }
  ],
  "logging_steps": 2,
  "max_steps": 2010,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 5,
        "early_stopping_threshold": 0.01
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 844829150146560.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}