ThyroidTumorClassification

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 74.89,
-    "eval_accuracy": 0.8,
-    "eval_loss": 0.5911628603935242,
-    "eval_runtime": 0.4617,
-    "eval_samples_per_second": 151.624,
-    "eval_steps_per_second": 6.498,
-    "total_flos": 5.2149900813992755e+17,
-    "train_loss": 0.47956085205078125,
-    "train_runtime": 368.1446,
-    "train_samples_per_second": 56.432,
-    "train_steps_per_second": 0.407
+    "epoch": 149.89,
+    "eval_f1": 0.9032258064516129,
+    "eval_loss": 0.45062127709388733,
+    "eval_runtime": 0.5727,
+    "eval_samples_per_second": 122.228,
+    "eval_steps_per_second": 5.238,
+    "total_flos": 1.0435256966870508e+18,
+    "train_loss": 0.3680222670237223,
+    "train_runtime": 877.204,
+    "train_samples_per_second": 47.366,
+    "train_steps_per_second": 0.342
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 74.89,
-    "eval_accuracy": 0.8,
-    "eval_loss": 0.5911628603935242,
-    "eval_runtime": 0.4617,
-    "eval_samples_per_second": 151.624,
-    "eval_steps_per_second": 6.498
+    "epoch": 149.89,
+    "eval_f1": 0.9032258064516129,
+    "eval_loss": 0.45062127709388733,
+    "eval_runtime": 0.5727,
+    "eval_samples_per_second": 122.228,
+    "eval_steps_per_second": 5.238
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:99c3b3104392b943a286ff7bca287f28d3567158c999717a61dd2cd9114628a1
+oid sha256:c1111386620231e1358294477e6b15819e1f4b775f95505cb21a90993de59864
 size 111347349

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 74.89,
-    "total_flos": 5.2149900813992755e+17,
-    "train_loss": 0.47956085205078125,
-    "train_runtime": 368.1446,
-    "train_samples_per_second": 56.432,
-    "train_steps_per_second": 0.407
+    "epoch": 149.89,
+    "total_flos": 1.0435256966870508e+18,
+    "train_loss": 0.3680222670237223,
+    "train_runtime": 877.204,
+    "train_samples_per_second": 47.366,
+    "train_steps_per_second": 0.342
 }

trainer_state.json

@@ -1,736 +1,1447 @@
 {
-  "best_metric": 0.8,
-  "best_model_checkpoint": "Cvt-finetuned-thyroid/checkpoint-14",
-  "epoch": 74.88888888888889,
-  "global_step": 150,
+  "best_metric": 0.9032258064516129,
+  "best_model_checkpoint": "Cvt-finetuned-thyroid/checkpoint-218",
+  "epoch": 149.88888888888889,
+  "global_step": 300,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.89,
-      "eval_accuracy": 0.6,
-      "eval_loss": 0.6705443263053894,
-      "eval_runtime": 0.462,
-      "eval_samples_per_second": 151.523,
-      "eval_steps_per_second": 6.494,
+      "eval_f1": 0.8474576271186439,
+      "eval_loss": 0.6517891883850098,
+      "eval_runtime": 0.5116,
+      "eval_samples_per_second": 136.828,
+      "eval_steps_per_second": 5.864,
       "step": 2
     },
     {
       "epoch": 1.89,
-      "eval_accuracy": 0.6285714285714286,
-      "eval_loss": 0.665534257888794,
-      "eval_runtime": 0.4493,
-      "eval_samples_per_second": 155.788,
-      "eval_steps_per_second": 6.677,
+      "eval_f1": 0.8474576271186439,
+      "eval_loss": 0.6492317318916321,
+      "eval_runtime": 0.4952,
+      "eval_samples_per_second": 141.346,
+      "eval_steps_per_second": 6.058,
       "step": 4
     },
     {
       "epoch": 2.89,
-      "eval_accuracy": 0.6714285714285714,
-      "eval_loss": 0.6565709710121155,
-      "eval_runtime": 0.4608,
-      "eval_samples_per_second": 151.906,
-      "eval_steps_per_second": 6.51,
+      "eval_f1": 0.8474576271186439,
+      "eval_loss": 0.6447046995162964,
+      "eval_runtime": 0.6963,
+      "eval_samples_per_second": 100.538,
+      "eval_steps_per_second": 4.309,
       "step": 6
     },
     {
       "epoch": 3.89,
-      "eval_accuracy": 0.7428571428571429,
-      "eval_loss": 0.643831729888916,
-      "eval_runtime": 0.4726,
-      "eval_samples_per_second": 148.132,
-      "eval_steps_per_second": 6.349,
+      "eval_f1": 0.8333333333333334,
+      "eval_loss": 0.6384778022766113,
+      "eval_runtime": 0.5,
+      "eval_samples_per_second": 140.006,
+      "eval_steps_per_second": 6.0,
       "step": 8
     },
     {
       "epoch": 4.89,
-      "eval_accuracy": 0.7571428571428571,
-      "eval_loss": 0.6282520890235901,
-      "eval_runtime": 0.4465,
-      "eval_samples_per_second": 156.767,
-      "eval_steps_per_second": 6.719,
+      "eval_f1": 0.8524590163934426,
+      "eval_loss": 0.6307393312454224,
+      "eval_runtime": 0.5113,
+      "eval_samples_per_second": 136.904,
+      "eval_steps_per_second": 5.867,
       "step": 10
     },
     {
       "epoch": 5.89,
-      "eval_accuracy": 0.7857142857142857,
-      "eval_loss": 0.6104965209960938,
-      "eval_runtime": 0.448,
-      "eval_samples_per_second": 156.239,
-      "eval_steps_per_second": 6.696,
+      "eval_f1": 0.8709677419354839,
+      "eval_loss": 0.621659517288208,
+      "eval_runtime": 0.5005,
+      "eval_samples_per_second": 139.863,
+      "eval_steps_per_second": 5.994,
       "step": 12
     },
     {
       "epoch": 6.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5911628603935242,
-      "eval_runtime": 0.4464,
-      "eval_samples_per_second": 156.804,
-      "eval_steps_per_second": 6.72,
+      "eval_f1": 0.8709677419354839,
+      "eval_loss": 0.6113398671150208,
+      "eval_runtime": 0.5076,
+      "eval_samples_per_second": 137.913,
+      "eval_steps_per_second": 5.911,
       "step": 14
     },
     {
       "epoch": 7.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5708973407745361,
-      "eval_runtime": 0.4463,
-      "eval_samples_per_second": 156.848,
-      "eval_steps_per_second": 6.722,
+      "eval_f1": 0.8799999999999999,
+      "eval_loss": 0.5999341011047363,
+      "eval_runtime": 0.504,
+      "eval_samples_per_second": 138.893,
+      "eval_steps_per_second": 5.953,
       "step": 16
     },
     {
       "epoch": 8.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5531989336013794,
-      "eval_runtime": 0.439,
-      "eval_samples_per_second": 159.442,
-      "eval_steps_per_second": 6.833,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.587955117225647,
+      "eval_runtime": 0.5041,
+      "eval_samples_per_second": 138.872,
+      "eval_steps_per_second": 5.952,
       "step": 18
     },
     {
       "epoch": 9.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5385797619819641,
-      "eval_runtime": 0.4602,
-      "eval_samples_per_second": 152.109,
-      "eval_steps_per_second": 6.519,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5761495232582092,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.618,
+      "eval_steps_per_second": 5.898,
       "step": 20
     },
     {
       "epoch": 10.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5270153880119324,
-      "eval_runtime": 0.4401,
-      "eval_samples_per_second": 159.068,
-      "eval_steps_per_second": 6.817,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5640864372253418,
+      "eval_runtime": 0.5127,
+      "eval_samples_per_second": 136.532,
+      "eval_steps_per_second": 5.851,
       "step": 22
     },
     {
       "epoch": 11.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5181472301483154,
-      "eval_runtime": 0.4475,
-      "eval_samples_per_second": 156.407,
-      "eval_steps_per_second": 6.703,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5520942211151123,
+      "eval_runtime": 0.5104,
+      "eval_samples_per_second": 137.159,
+      "eval_steps_per_second": 5.878,
       "step": 24
     },
     {
       "epoch": 12.44,
-      "learning_rate": 9.25925925925926e-06,
-      "loss": 0.6298,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.6685,
       "step": 25
     },
     {
       "epoch": 12.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5123368501663208,
-      "eval_runtime": 0.4347,
-      "eval_samples_per_second": 161.016,
-      "eval_steps_per_second": 6.901,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5406573414802551,
+      "eval_runtime": 0.5165,
+      "eval_samples_per_second": 135.523,
+      "eval_steps_per_second": 5.808,
       "step": 26
     },
     {
       "epoch": 13.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5075790286064148,
-      "eval_runtime": 0.4347,
-      "eval_samples_per_second": 161.026,
-      "eval_steps_per_second": 6.901,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5298128128051758,
+      "eval_runtime": 0.5102,
+      "eval_samples_per_second": 137.199,
+      "eval_steps_per_second": 5.88,
       "step": 28
     },
     {
       "epoch": 14.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5036987066268921,
-      "eval_runtime": 0.4334,
-      "eval_samples_per_second": 161.512,
-      "eval_steps_per_second": 6.922,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5198087096214294,
+      "eval_runtime": 0.51,
+      "eval_samples_per_second": 137.243,
+      "eval_steps_per_second": 5.882,
       "step": 30
     },
     {
       "epoch": 15.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.5011651515960693,
-      "eval_runtime": 0.4565,
-      "eval_samples_per_second": 153.34,
-      "eval_steps_per_second": 6.572,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5109697580337524,
+      "eval_runtime": 0.5205,
+      "eval_samples_per_second": 134.496,
+      "eval_steps_per_second": 5.764,
       "step": 32
     },
     {
       "epoch": 16.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.498863160610199,
-      "eval_runtime": 0.4641,
-      "eval_samples_per_second": 150.834,
-      "eval_steps_per_second": 6.464,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.5045211911201477,
+      "eval_runtime": 0.514,
+      "eval_samples_per_second": 136.181,
+      "eval_steps_per_second": 5.836,
       "step": 34
     },
     {
       "epoch": 17.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.497054785490036,
-      "eval_runtime": 0.44,
-      "eval_samples_per_second": 159.098,
-      "eval_steps_per_second": 6.818,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4998997747898102,
+      "eval_runtime": 0.5294,
+      "eval_samples_per_second": 132.214,
+      "eval_steps_per_second": 5.666,
       "step": 36
     },
     {
       "epoch": 18.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4955058991909027,
-      "eval_runtime": 0.4462,
-      "eval_samples_per_second": 156.867,
-      "eval_steps_per_second": 6.723,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4965072572231293,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.597,
+      "eval_steps_per_second": 5.897,
       "step": 38
     },
     {
       "epoch": 19.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4940197467803955,
-      "eval_runtime": 0.4416,
-      "eval_samples_per_second": 158.514,
-      "eval_steps_per_second": 6.793,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.493766188621521,
+      "eval_runtime": 0.5102,
+      "eval_samples_per_second": 137.214,
+      "eval_steps_per_second": 5.881,
       "step": 40
     },
     {
       "epoch": 20.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.49268171191215515,
-      "eval_runtime": 0.4504,
-      "eval_samples_per_second": 155.428,
-      "eval_steps_per_second": 6.661,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.49165278673171997,
+      "eval_runtime": 0.5078,
+      "eval_samples_per_second": 137.845,
+      "eval_steps_per_second": 5.908,
       "step": 42
     },
     {
       "epoch": 21.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.49133333563804626,
-      "eval_runtime": 0.4327,
-      "eval_samples_per_second": 161.788,
-      "eval_steps_per_second": 6.934,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4898264706134796,
+      "eval_runtime": 0.5144,
+      "eval_samples_per_second": 136.069,
+      "eval_steps_per_second": 5.832,
       "step": 44
     },
     {
       "epoch": 22.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.489890456199646,
-      "eval_runtime": 0.4454,
-      "eval_samples_per_second": 157.155,
-      "eval_steps_per_second": 6.735,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4883803129196167,
+      "eval_runtime": 0.5035,
+      "eval_samples_per_second": 139.037,
+      "eval_steps_per_second": 5.959,
       "step": 46
     },
     {
       "epoch": 23.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.48838457465171814,
-      "eval_runtime": 0.4388,
-      "eval_samples_per_second": 159.525,
-      "eval_steps_per_second": 6.837,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4872118830680847,
+      "eval_runtime": 0.5072,
+      "eval_samples_per_second": 138.01,
+      "eval_steps_per_second": 5.915,
       "step": 48
     },
     {
       "epoch": 24.89,
-      "learning_rate": 7.4074074074074075e-06,
-      "loss": 0.4947,
+      "learning_rate": 9.25925925925926e-06,
+      "loss": 0.5044,
       "step": 50
     },
     {
       "epoch": 24.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4868685007095337,
-      "eval_runtime": 0.44,
-      "eval_samples_per_second": 159.08,
-      "eval_steps_per_second": 6.818,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4861142933368683,
+      "eval_runtime": 0.5085,
+      "eval_samples_per_second": 137.671,
+      "eval_steps_per_second": 5.9,
       "step": 50
     },
     {
       "epoch": 25.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.48541581630706787,
-      "eval_runtime": 0.4379,
-      "eval_samples_per_second": 159.843,
-      "eval_steps_per_second": 6.85,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4849891662597656,
+      "eval_runtime": 0.5374,
+      "eval_samples_per_second": 130.249,
+      "eval_steps_per_second": 5.582,
       "step": 52
     },
     {
       "epoch": 26.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4839009940624237,
-      "eval_runtime": 0.4366,
-      "eval_samples_per_second": 160.316,
-      "eval_steps_per_second": 6.871,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.483755499124527,
+      "eval_runtime": 0.51,
+      "eval_samples_per_second": 137.255,
+      "eval_steps_per_second": 5.882,
       "step": 54
     },
     {
       "epoch": 27.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.482656329870224,
-      "eval_runtime": 0.4434,
-      "eval_samples_per_second": 157.854,
-      "eval_steps_per_second": 6.765,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4827323257923126,
+      "eval_runtime": 0.5139,
+      "eval_samples_per_second": 136.21,
+      "eval_steps_per_second": 5.838,
       "step": 56
     },
     {
       "epoch": 28.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4816107749938965,
-      "eval_runtime": 0.4432,
-      "eval_samples_per_second": 157.931,
-      "eval_steps_per_second": 6.768,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4820977449417114,
+      "eval_runtime": 0.5117,
+      "eval_samples_per_second": 136.794,
+      "eval_steps_per_second": 5.863,
       "step": 58
     },
     {
       "epoch": 29.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4808529019355774,
-      "eval_runtime": 0.4449,
-      "eval_samples_per_second": 157.322,
-      "eval_steps_per_second": 6.742,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4817977249622345,
+      "eval_runtime": 0.5176,
+      "eval_samples_per_second": 135.231,
+      "eval_steps_per_second": 5.796,
       "step": 60
     },
     {
       "epoch": 30.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4799829125404358,
-      "eval_runtime": 0.4386,
-      "eval_samples_per_second": 159.609,
-      "eval_steps_per_second": 6.84,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4813471734523773,
+      "eval_runtime": 0.5076,
+      "eval_samples_per_second": 137.901,
+      "eval_steps_per_second": 5.91,
       "step": 62
     },
     {
       "epoch": 31.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.479232519865036,
-      "eval_runtime": 0.4533,
-      "eval_samples_per_second": 154.433,
-      "eval_steps_per_second": 6.619,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4812641739845276,
+      "eval_runtime": 0.5176,
+      "eval_samples_per_second": 135.242,
+      "eval_steps_per_second": 5.796,
       "step": 64
     },
     {
       "epoch": 32.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4783066213130951,
-      "eval_runtime": 0.4417,
-      "eval_samples_per_second": 158.47,
-      "eval_steps_per_second": 6.792,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.48070254921913147,
+      "eval_runtime": 0.5078,
+      "eval_samples_per_second": 137.854,
+      "eval_steps_per_second": 5.908,
       "step": 66
     },
     {
       "epoch": 33.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4775735139846802,
-      "eval_runtime": 0.4472,
-      "eval_samples_per_second": 156.531,
-      "eval_steps_per_second": 6.708,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.48028674721717834,
+      "eval_runtime": 0.5091,
+      "eval_samples_per_second": 137.488,
+      "eval_steps_per_second": 5.892,
       "step": 68
     },
     {
       "epoch": 34.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4768332839012146,
-      "eval_runtime": 0.4342,
-      "eval_samples_per_second": 161.213,
-      "eval_steps_per_second": 6.909,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4801904261112213,
+      "eval_runtime": 0.5047,
+      "eval_samples_per_second": 138.694,
+      "eval_steps_per_second": 5.944,
       "step": 70
     },
     {
       "epoch": 35.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.47618430852890015,
-      "eval_runtime": 0.4501,
-      "eval_samples_per_second": 155.536,
-      "eval_steps_per_second": 6.666,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4799834191799164,
+      "eval_runtime": 0.5083,
+      "eval_samples_per_second": 137.704,
+      "eval_steps_per_second": 5.902,
       "step": 72
     },
     {
       "epoch": 36.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4756462574005127,
-      "eval_runtime": 0.4411,
-      "eval_samples_per_second": 158.688,
-      "eval_steps_per_second": 6.801,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47955021262168884,
+      "eval_runtime": 0.5047,
+      "eval_samples_per_second": 138.688,
+      "eval_steps_per_second": 5.944,
       "step": 74
     },
     {
       "epoch": 37.44,
-      "learning_rate": 5.555555555555557e-06,
-      "loss": 0.4688,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.4434,
       "step": 75
     },
     {
       "epoch": 37.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.47506338357925415,
-      "eval_runtime": 0.4697,
-      "eval_samples_per_second": 149.047,
-      "eval_steps_per_second": 6.388,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47954249382019043,
+      "eval_runtime": 0.5098,
+      "eval_samples_per_second": 137.317,
+      "eval_steps_per_second": 5.885,
       "step": 76
     },
     {
       "epoch": 38.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4744085669517517,
-      "eval_runtime": 0.444,
-      "eval_samples_per_second": 157.644,
-      "eval_steps_per_second": 6.756,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47896233201026917,
+      "eval_runtime": 0.5167,
+      "eval_samples_per_second": 135.481,
+      "eval_steps_per_second": 5.806,
       "step": 78
     },
     {
       "epoch": 39.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4739179313182831,
-      "eval_runtime": 0.4514,
-      "eval_samples_per_second": 155.078,
-      "eval_steps_per_second": 6.646,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4780120849609375,
+      "eval_runtime": 0.5174,
+      "eval_samples_per_second": 135.289,
+      "eval_steps_per_second": 5.798,
       "step": 80
     },
     {
       "epoch": 40.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4733032286167145,
-      "eval_runtime": 0.4549,
-      "eval_samples_per_second": 153.865,
-      "eval_steps_per_second": 6.594,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4772844612598419,
+      "eval_runtime": 0.5162,
+      "eval_samples_per_second": 135.603,
+      "eval_steps_per_second": 5.812,
       "step": 82
     },
     {
       "epoch": 41.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4729520082473755,
-      "eval_runtime": 0.4432,
-      "eval_samples_per_second": 157.928,
-      "eval_steps_per_second": 6.768,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47639530897140503,
+      "eval_runtime": 0.5186,
+      "eval_samples_per_second": 134.98,
+      "eval_steps_per_second": 5.785,
       "step": 84
     },
     {
       "epoch": 42.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4726985991001129,
-      "eval_runtime": 0.4582,
-      "eval_samples_per_second": 152.787,
-      "eval_steps_per_second": 6.548,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47554805874824524,
+      "eval_runtime": 0.5172,
+      "eval_samples_per_second": 135.355,
+      "eval_steps_per_second": 5.801,
       "step": 86
     },
     {
       "epoch": 43.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4726087749004364,
-      "eval_runtime": 0.4498,
-      "eval_samples_per_second": 155.628,
-      "eval_steps_per_second": 6.67,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4745935797691345,
+      "eval_runtime": 0.514,
+      "eval_samples_per_second": 136.19,
+      "eval_steps_per_second": 5.837,
       "step": 88
     },
     {
       "epoch": 44.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.47239962220191956,
-      "eval_runtime": 0.4376,
-      "eval_samples_per_second": 159.981,
-      "eval_steps_per_second": 6.856,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4736994504928589,
+      "eval_runtime": 0.5108,
+      "eval_samples_per_second": 137.03,
+      "eval_steps_per_second": 5.873,
       "step": 90
     },
     {
       "epoch": 45.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4720509648323059,
-      "eval_runtime": 0.4582,
-      "eval_samples_per_second": 152.761,
-      "eval_steps_per_second": 6.547,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4725368618965149,
+      "eval_runtime": 0.5067,
+      "eval_samples_per_second": 138.137,
+      "eval_steps_per_second": 5.92,
       "step": 92
     },
     {
       "epoch": 46.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.47151511907577515,
-      "eval_runtime": 0.4581,
-      "eval_samples_per_second": 152.803,
-      "eval_steps_per_second": 6.549,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.47122785449028015,
+      "eval_runtime": 0.5187,
+      "eval_samples_per_second": 134.95,
+      "eval_steps_per_second": 5.784,
       "step": 94
     },
     {
       "epoch": 47.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4711145758628845,
-      "eval_runtime": 0.4536,
-      "eval_samples_per_second": 154.323,
-      "eval_steps_per_second": 6.614,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4697120189666748,
+      "eval_runtime": 0.5106,
+      "eval_samples_per_second": 137.089,
+      "eval_steps_per_second": 5.875,
       "step": 96
     },
     {
       "epoch": 48.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.47070595622062683,
-      "eval_runtime": 0.4562,
-      "eval_samples_per_second": 153.435,
-      "eval_steps_per_second": 6.576,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4686456322669983,
+      "eval_runtime": 0.5071,
+      "eval_samples_per_second": 138.045,
+      "eval_steps_per_second": 5.916,
       "step": 98
     },
     {
       "epoch": 49.89,
-      "learning_rate": 3.7037037037037037e-06,
-      "loss": 0.4378,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.4,
       "step": 100
     },
     {
       "epoch": 49.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4701813757419586,
-      "eval_runtime": 0.4473,
-      "eval_samples_per_second": 156.496,
-      "eval_steps_per_second": 6.707,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46755868196487427,
+      "eval_runtime": 0.5094,
+      "eval_samples_per_second": 137.403,
+      "eval_steps_per_second": 5.889,
       "step": 100
     },
     {
       "epoch": 50.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46980804204940796,
-      "eval_runtime": 0.4344,
-      "eval_samples_per_second": 161.135,
-      "eval_steps_per_second": 6.906,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4666549265384674,
+      "eval_runtime": 0.5092,
+      "eval_samples_per_second": 137.458,
+      "eval_steps_per_second": 5.891,
       "step": 102
     },
     {
       "epoch": 51.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46947282552719116,
-      "eval_runtime": 0.4638,
-      "eval_samples_per_second": 150.921,
-      "eval_steps_per_second": 6.468,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46638351678848267,
+      "eval_runtime": 0.5124,
+      "eval_samples_per_second": 136.625,
+      "eval_steps_per_second": 5.855,
       "step": 104
     },
     {
       "epoch": 52.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46929818391799927,
-      "eval_runtime": 0.448,
-      "eval_samples_per_second": 156.238,
-      "eval_steps_per_second": 6.696,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4666298031806946,
+      "eval_runtime": 0.5135,
+      "eval_samples_per_second": 136.326,
+      "eval_steps_per_second": 5.843,
       "step": 106
     },
     {
       "epoch": 53.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4691893458366394,
-      "eval_runtime": 0.4537,
-      "eval_samples_per_second": 154.298,
-      "eval_steps_per_second": 6.613,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46660488843917847,
+      "eval_runtime": 0.5175,
+      "eval_samples_per_second": 135.261,
+      "eval_steps_per_second": 5.797,
       "step": 108
     },
     {
       "epoch": 54.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46898388862609863,
-      "eval_runtime": 0.4451,
-      "eval_samples_per_second": 157.257,
-      "eval_steps_per_second": 6.74,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4668794870376587,
+      "eval_runtime": 0.5009,
+      "eval_samples_per_second": 139.747,
+      "eval_steps_per_second": 5.989,
       "step": 110
     },
     {
       "epoch": 55.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46869638562202454,
-      "eval_runtime": 0.4465,
-      "eval_samples_per_second": 156.786,
-      "eval_steps_per_second": 6.719,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46678298711776733,
+      "eval_runtime": 0.5078,
+      "eval_samples_per_second": 137.844,
+      "eval_steps_per_second": 5.908,
       "step": 112
     },
     {
       "epoch": 56.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4685034155845642,
-      "eval_runtime": 0.4553,
-      "eval_samples_per_second": 153.754,
-      "eval_steps_per_second": 6.589,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46638283133506775,
+      "eval_runtime": 0.5157,
+      "eval_samples_per_second": 135.74,
+      "eval_steps_per_second": 5.817,
       "step": 114
     },
     {
       "epoch": 57.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46839845180511475,
-      "eval_runtime": 0.4475,
-      "eval_samples_per_second": 156.417,
-      "eval_steps_per_second": 6.704,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46611693501472473,
+      "eval_runtime": 0.5257,
+      "eval_samples_per_second": 133.154,
+      "eval_steps_per_second": 5.707,
       "step": 116
     },
     {
       "epoch": 58.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46825066208839417,
-      "eval_runtime": 0.446,
-      "eval_samples_per_second": 156.968,
-      "eval_steps_per_second": 6.727,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4656626582145691,
+      "eval_runtime": 0.5108,
+      "eval_samples_per_second": 137.041,
+      "eval_steps_per_second": 5.873,
       "step": 118
     },
     {
       "epoch": 59.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46806439757347107,
-      "eval_runtime": 0.4552,
-      "eval_samples_per_second": 153.787,
-      "eval_steps_per_second": 6.591,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4653801918029785,
+      "eval_runtime": 0.5184,
+      "eval_samples_per_second": 135.019,
+      "eval_steps_per_second": 5.787,
       "step": 120
     },
     {
       "epoch": 60.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4679650366306305,
-      "eval_runtime": 0.45,
-      "eval_samples_per_second": 155.565,
-      "eval_steps_per_second": 6.667,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46494531631469727,
+      "eval_runtime": 0.5265,
+      "eval_samples_per_second": 132.948,
+      "eval_steps_per_second": 5.698,
       "step": 122
     },
     {
       "epoch": 61.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4678630232810974,
-      "eval_runtime": 0.4524,
-      "eval_samples_per_second": 154.721,
-      "eval_steps_per_second": 6.631,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4643649458885193,
+      "eval_runtime": 0.5176,
+      "eval_samples_per_second": 135.236,
+      "eval_steps_per_second": 5.796,
       "step": 124
     },
     {
       "epoch": 62.44,
-      "learning_rate": 1.8518518518518519e-06,
-      "loss": 0.4274,
+      "learning_rate": 6.481481481481482e-06,
+      "loss": 0.3712,
       "step": 125
     },
     {
       "epoch": 62.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.467722088098526,
-      "eval_runtime": 0.4455,
-      "eval_samples_per_second": 157.131,
-      "eval_steps_per_second": 6.734,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46361738443374634,
+      "eval_runtime": 0.5177,
+      "eval_samples_per_second": 135.21,
+      "eval_steps_per_second": 5.795,
       "step": 126
     },
     {
       "epoch": 63.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4675312638282776,
-      "eval_runtime": 0.4574,
-      "eval_samples_per_second": 153.047,
-      "eval_steps_per_second": 6.559,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.46249672770500183,
+      "eval_runtime": 0.5106,
+      "eval_samples_per_second": 137.097,
+      "eval_steps_per_second": 5.876,
       "step": 128
     },
     {
       "epoch": 64.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46731725335121155,
-      "eval_runtime": 0.4688,
-      "eval_samples_per_second": 149.332,
-      "eval_steps_per_second": 6.4,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4611242711544037,
+      "eval_runtime": 0.5134,
+      "eval_samples_per_second": 136.352,
+      "eval_steps_per_second": 5.844,
       "step": 130
     },
     {
       "epoch": 65.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.467153936624527,
-      "eval_runtime": 0.4613,
-      "eval_samples_per_second": 151.749,
-      "eval_steps_per_second": 6.504,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4598376154899597,
+      "eval_runtime": 0.5045,
+      "eval_samples_per_second": 138.759,
+      "eval_steps_per_second": 5.947,
       "step": 132
     },
     {
       "epoch": 66.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46700313687324524,
-      "eval_runtime": 0.4516,
-      "eval_samples_per_second": 155.016,
-      "eval_steps_per_second": 6.644,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45868563652038574,
+      "eval_runtime": 0.5061,
+      "eval_samples_per_second": 138.303,
+      "eval_steps_per_second": 5.927,
       "step": 134
     },
     {
       "epoch": 67.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46686962246894836,
-      "eval_runtime": 0.4449,
-      "eval_samples_per_second": 157.335,
-      "eval_steps_per_second": 6.743,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4579010605812073,
+      "eval_runtime": 0.5154,
+      "eval_samples_per_second": 135.819,
+      "eval_steps_per_second": 5.821,
       "step": 136
     },
     {
       "epoch": 68.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46674275398254395,
-      "eval_runtime": 0.4512,
-      "eval_samples_per_second": 155.137,
-      "eval_steps_per_second": 6.649,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4574340879917145,
+      "eval_runtime": 0.5107,
+      "eval_samples_per_second": 137.062,
+      "eval_steps_per_second": 5.874,
       "step": 138
     },
     {
       "epoch": 69.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4666597545146942,
-      "eval_runtime": 0.4478,
-      "eval_samples_per_second": 156.329,
-      "eval_steps_per_second": 6.7,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45691609382629395,
+      "eval_runtime": 0.5186,
+      "eval_samples_per_second": 134.99,
+      "eval_steps_per_second": 5.785,
       "step": 140
     },
     {
       "epoch": 70.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4665738344192505,
-      "eval_runtime": 0.4419,
-      "eval_samples_per_second": 158.423,
-      "eval_steps_per_second": 6.79,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45680859684944153,
+      "eval_runtime": 0.5147,
+      "eval_samples_per_second": 136.006,
+      "eval_steps_per_second": 5.829,
       "step": 142
     },
     {
       "epoch": 71.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46650615334510803,
-      "eval_runtime": 0.45,
-      "eval_samples_per_second": 155.562,
-      "eval_steps_per_second": 6.667,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4564048945903778,
+      "eval_runtime": 0.5157,
+      "eval_samples_per_second": 135.741,
+      "eval_steps_per_second": 5.817,
       "step": 144
     },
     {
       "epoch": 72.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4664507806301117,
-      "eval_runtime": 0.4559,
-      "eval_samples_per_second": 153.528,
-      "eval_steps_per_second": 6.58,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4561113119125366,
+      "eval_runtime": 0.5039,
+      "eval_samples_per_second": 138.917,
+      "eval_steps_per_second": 5.954,
       "step": 146
     },
     {
       "epoch": 73.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.46641966700553894,
-      "eval_runtime": 0.4546,
-      "eval_samples_per_second": 153.981,
-      "eval_steps_per_second": 6.599,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45622721314430237,
+      "eval_runtime": 0.514,
+      "eval_samples_per_second": 136.175,
+      "eval_steps_per_second": 5.836,
       "step": 148
     },
     {
       "epoch": 74.89,
-      "learning_rate": 0.0,
-      "loss": 0.4187,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 0.3419,
       "step": 150
     },
     {
       "epoch": 74.89,
-      "eval_accuracy": 0.8,
-      "eval_loss": 0.4664078950881958,
-      "eval_runtime": 0.444,
-      "eval_samples_per_second": 157.667,
-      "eval_steps_per_second": 6.757,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4559585452079773,
+      "eval_runtime": 0.5107,
+      "eval_samples_per_second": 137.074,
+      "eval_steps_per_second": 5.875,
       "step": 150
     },
     {
-      "epoch": 74.89,
-      "step": 150,
-      "total_flos": 5.2149900813992755e+17,
-      "train_loss": 0.47956085205078125,
-      "train_runtime": 368.1446,
-      "train_samples_per_second": 56.432,
-      "train_steps_per_second": 0.407
+      "epoch": 75.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4556769132614136,
+      "eval_runtime": 0.5134,
+      "eval_samples_per_second": 136.339,
+      "eval_steps_per_second": 5.843,
+      "step": 152
+    },
+    {
+      "epoch": 76.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45586857199668884,
+      "eval_runtime": 0.5119,
+      "eval_samples_per_second": 136.757,
+      "eval_steps_per_second": 5.861,
+      "step": 154
+    },
+    {
+      "epoch": 77.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45571208000183105,
+      "eval_runtime": 0.5116,
+      "eval_samples_per_second": 136.818,
+      "eval_steps_per_second": 5.864,
+      "step": 156
+    },
+    {
+      "epoch": 78.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4558698832988739,
+      "eval_runtime": 0.5082,
+      "eval_samples_per_second": 137.744,
+      "eval_steps_per_second": 5.903,
+      "step": 158
+    },
+    {
+      "epoch": 79.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4559585154056549,
+      "eval_runtime": 0.5127,
+      "eval_samples_per_second": 136.542,
+      "eval_steps_per_second": 5.852,
+      "step": 160
+    },
+    {
+      "epoch": 80.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4561022222042084,
+      "eval_runtime": 0.5136,
+      "eval_samples_per_second": 136.283,
+      "eval_steps_per_second": 5.841,
+      "step": 162
+    },
+    {
+      "epoch": 81.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.4561418294906616,
+      "eval_runtime": 0.5185,
+      "eval_samples_per_second": 135.002,
+      "eval_steps_per_second": 5.786,
+      "step": 164
+    },
+    {
+      "epoch": 82.89,
+      "eval_f1": 0.888888888888889,
+      "eval_loss": 0.45633062720298767,
+      "eval_runtime": 0.5583,
+      "eval_samples_per_second": 125.383,
+      "eval_steps_per_second": 5.374,
+      "step": 166
+    },
+    {
+      "epoch": 83.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45609140396118164,
+      "eval_runtime": 0.5096,
+      "eval_samples_per_second": 137.376,
+      "eval_steps_per_second": 5.888,
+      "step": 168
+    },
+    {
+      "epoch": 84.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4557640850543976,
+      "eval_runtime": 0.5089,
+      "eval_samples_per_second": 137.544,
+      "eval_steps_per_second": 5.895,
+      "step": 170
+    },
+    {
+      "epoch": 85.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4552680552005768,
+      "eval_runtime": 0.5124,
+      "eval_samples_per_second": 136.601,
+      "eval_steps_per_second": 5.854,
+      "step": 172
+    },
+    {
+      "epoch": 86.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45485004782676697,
+      "eval_runtime": 0.5155,
+      "eval_samples_per_second": 135.799,
+      "eval_steps_per_second": 5.82,
+      "step": 174
+    },
+    {
+      "epoch": 87.44,
+      "learning_rate": 4.62962962962963e-06,
+      "loss": 0.3221,
+      "step": 175
+    },
+    {
+      "epoch": 87.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45451802015304565,
+      "eval_runtime": 0.5105,
+      "eval_samples_per_second": 137.109,
+      "eval_steps_per_second": 5.876,
+      "step": 176
+    },
+    {
+      "epoch": 88.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4541896879673004,
+      "eval_runtime": 0.5149,
+      "eval_samples_per_second": 135.937,
+      "eval_steps_per_second": 5.826,
+      "step": 178
+    },
+    {
+      "epoch": 89.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45373955368995667,
+      "eval_runtime": 0.5183,
+      "eval_samples_per_second": 135.051,
+      "eval_steps_per_second": 5.788,
+      "step": 180
+    },
+    {
+      "epoch": 90.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4535920023918152,
+      "eval_runtime": 0.5109,
+      "eval_samples_per_second": 137.024,
+      "eval_steps_per_second": 5.872,
+      "step": 182
+    },
+    {
+      "epoch": 91.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45348674058914185,
+      "eval_runtime": 0.5161,
+      "eval_samples_per_second": 135.628,
+      "eval_steps_per_second": 5.813,
+      "step": 184
+    },
+    {
+      "epoch": 92.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45325374603271484,
+      "eval_runtime": 0.5128,
+      "eval_samples_per_second": 136.516,
+      "eval_steps_per_second": 5.851,
+      "step": 186
+    },
+    {
+      "epoch": 93.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4530419111251831,
+      "eval_runtime": 0.5198,
+      "eval_samples_per_second": 134.677,
+      "eval_steps_per_second": 5.772,
+      "step": 188
+    },
+    {
+      "epoch": 94.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.452932745218277,
+      "eval_runtime": 0.5123,
+      "eval_samples_per_second": 136.651,
+      "eval_steps_per_second": 5.856,
+      "step": 190
+    },
+    {
+      "epoch": 95.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45291921496391296,
+      "eval_runtime": 0.5108,
+      "eval_samples_per_second": 137.03,
+      "eval_steps_per_second": 5.873,
+      "step": 192
+    },
+    {
+      "epoch": 96.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45280784368515015,
+      "eval_runtime": 0.5125,
+      "eval_samples_per_second": 136.577,
+      "eval_steps_per_second": 5.853,
+      "step": 194
+    },
+    {
+      "epoch": 97.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4528333246707916,
+      "eval_runtime": 0.5077,
+      "eval_samples_per_second": 137.871,
+      "eval_steps_per_second": 5.909,
+      "step": 196
+    },
+    {
+      "epoch": 98.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45277661085128784,
+      "eval_runtime": 0.508,
+      "eval_samples_per_second": 137.803,
+      "eval_steps_per_second": 5.906,
+      "step": 198
+    },
+    {
+      "epoch": 99.89,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.297,
+      "step": 200
+    },
+    {
+      "epoch": 99.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45276370644569397,
+      "eval_runtime": 0.5103,
+      "eval_samples_per_second": 137.186,
+      "eval_steps_per_second": 5.879,
+      "step": 200
+    },
+    {
+      "epoch": 100.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4528166651725769,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.613,
+      "eval_steps_per_second": 5.898,
+      "step": 202
+    },
+    {
+      "epoch": 101.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45277735590934753,
+      "eval_runtime": 0.5107,
+      "eval_samples_per_second": 137.056,
+      "eval_steps_per_second": 5.874,
+      "step": 204
+    },
+    {
+      "epoch": 102.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4523409307003021,
+      "eval_runtime": 0.519,
+      "eval_samples_per_second": 134.862,
+      "eval_steps_per_second": 5.78,
+      "step": 206
+    },
+    {
+      "epoch": 103.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4522373080253601,
+      "eval_runtime": 0.5234,
+      "eval_samples_per_second": 133.744,
+      "eval_steps_per_second": 5.732,
+      "step": 208
+    },
+    {
+      "epoch": 104.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4519382417201996,
+      "eval_runtime": 0.5099,
+      "eval_samples_per_second": 137.277,
+      "eval_steps_per_second": 5.883,
+      "step": 210
+    },
+    {
+      "epoch": 105.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4515880048274994,
+      "eval_runtime": 0.5111,
+      "eval_samples_per_second": 136.947,
+      "eval_steps_per_second": 5.869,
+      "step": 212
+    },
+    {
+      "epoch": 106.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.4515409767627716,
+      "eval_runtime": 0.5212,
+      "eval_samples_per_second": 134.295,
+      "eval_steps_per_second": 5.755,
+      "step": 214
+    },
+    {
+      "epoch": 107.89,
+      "eval_f1": 0.896,
+      "eval_loss": 0.45120465755462646,
+      "eval_runtime": 0.5234,
+      "eval_samples_per_second": 133.737,
+      "eval_steps_per_second": 5.732,
+      "step": 216
+    },
+    {
+      "epoch": 108.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.45062127709388733,
+      "eval_runtime": 0.5116,
+      "eval_samples_per_second": 136.817,
+      "eval_steps_per_second": 5.864,
+      "step": 218
+    },
+    {
+      "epoch": 109.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.450234055519104,
+      "eval_runtime": 0.5074,
+      "eval_samples_per_second": 137.954,
+      "eval_steps_per_second": 5.912,
+      "step": 220
+    },
+    {
+      "epoch": 110.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.45027095079421997,
+      "eval_runtime": 0.5078,
+      "eval_samples_per_second": 137.84,
+      "eval_steps_per_second": 5.907,
+      "step": 222
+    },
+    {
+      "epoch": 111.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4502160847187042,
+      "eval_runtime": 0.5143,
+      "eval_samples_per_second": 136.105,
+      "eval_steps_per_second": 5.833,
+      "step": 224
+    },
+    {
+      "epoch": 112.44,
+      "learning_rate": 2.7777777777777783e-06,
+      "loss": 0.2809,
+      "step": 225
+    },
+    {
+      "epoch": 112.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.45027267932891846,
+      "eval_runtime": 0.5062,
+      "eval_samples_per_second": 138.288,
+      "eval_steps_per_second": 5.927,
+      "step": 226
+    },
+    {
+      "epoch": 113.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.45001256465911865,
+      "eval_runtime": 0.5176,
+      "eval_samples_per_second": 135.243,
+      "eval_steps_per_second": 5.796,
+      "step": 228
+    },
+    {
+      "epoch": 114.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44984087347984314,
+      "eval_runtime": 0.5057,
+      "eval_samples_per_second": 138.435,
+      "eval_steps_per_second": 5.933,
+      "step": 230
+    },
+    {
+      "epoch": 115.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4496540129184723,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.605,
+      "eval_steps_per_second": 5.897,
+      "step": 232
+    },
+    {
+      "epoch": 116.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495893120765686,
+      "eval_runtime": 0.5162,
+      "eval_samples_per_second": 135.612,
+      "eval_steps_per_second": 5.812,
+      "step": 234
+    },
+    {
+      "epoch": 117.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495932459831238,
+      "eval_runtime": 0.5162,
+      "eval_samples_per_second": 135.596,
+      "eval_steps_per_second": 5.811,
+      "step": 236
+    },
+    {
+      "epoch": 118.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495743215084076,
+      "eval_runtime": 0.5143,
+      "eval_samples_per_second": 136.098,
+      "eval_steps_per_second": 5.833,
+      "step": 238
+    },
+    {
+      "epoch": 119.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495793282985687,
+      "eval_runtime": 0.5201,
+      "eval_samples_per_second": 134.594,
+      "eval_steps_per_second": 5.768,
+      "step": 240
+    },
+    {
+      "epoch": 120.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44972047209739685,
+      "eval_runtime": 0.517,
+      "eval_samples_per_second": 135.393,
+      "eval_steps_per_second": 5.803,
+      "step": 242
+    },
+    {
+      "epoch": 121.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4496006965637207,
+      "eval_runtime": 0.5201,
+      "eval_samples_per_second": 134.597,
+      "eval_steps_per_second": 5.768,
+      "step": 244
+    },
+    {
+      "epoch": 122.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495578706264496,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.594,
+      "eval_steps_per_second": 5.897,
+      "step": 246
+    },
+    {
+      "epoch": 123.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44961830973625183,
+      "eval_runtime": 0.5091,
+      "eval_samples_per_second": 137.492,
+      "eval_steps_per_second": 5.893,
+      "step": 248
+    },
+    {
+      "epoch": 124.89,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.2717,
+      "step": 250
+    },
+    {
+      "epoch": 124.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4495493769645691,
+      "eval_runtime": 0.5126,
+      "eval_samples_per_second": 136.564,
+      "eval_steps_per_second": 5.853,
+      "step": 250
+    },
+    {
+      "epoch": 125.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44953247904777527,
+      "eval_runtime": 0.5166,
+      "eval_samples_per_second": 135.508,
+      "eval_steps_per_second": 5.807,
+      "step": 252
+    },
+    {
+      "epoch": 126.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44945859909057617,
+      "eval_runtime": 0.5087,
+      "eval_samples_per_second": 137.598,
+      "eval_steps_per_second": 5.897,
+      "step": 254
+    },
+    {
+      "epoch": 127.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4493277072906494,
+      "eval_runtime": 0.5129,
+      "eval_samples_per_second": 136.49,
+      "eval_steps_per_second": 5.85,
+      "step": 256
+    },
+    {
+      "epoch": 128.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4492017924785614,
+      "eval_runtime": 0.5132,
+      "eval_samples_per_second": 136.411,
+      "eval_steps_per_second": 5.846,
+      "step": 258
+    },
+    {
+      "epoch": 129.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44910600781440735,
+      "eval_runtime": 0.5158,
+      "eval_samples_per_second": 135.724,
+      "eval_steps_per_second": 5.817,
+      "step": 260
+    },
+    {
+      "epoch": 130.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44898271560668945,
+      "eval_runtime": 0.5171,
+      "eval_samples_per_second": 135.359,
+      "eval_steps_per_second": 5.801,
+      "step": 262
+    },
+    {
+      "epoch": 131.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44882574677467346,
+      "eval_runtime": 0.5104,
+      "eval_samples_per_second": 137.159,
+      "eval_steps_per_second": 5.878,
+      "step": 264
+    },
+    {
+      "epoch": 132.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44872909784317017,
+      "eval_runtime": 0.5186,
+      "eval_samples_per_second": 134.967,
+      "eval_steps_per_second": 5.784,
+      "step": 266
+    },
+    {
+      "epoch": 133.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4485660791397095,
+      "eval_runtime": 0.511,
+      "eval_samples_per_second": 136.99,
+      "eval_steps_per_second": 5.871,
+      "step": 268
+    },
+    {
+      "epoch": 134.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44834819436073303,
+      "eval_runtime": 0.5105,
+      "eval_samples_per_second": 137.112,
+      "eval_steps_per_second": 5.876,
+      "step": 270
+    },
+    {
+      "epoch": 135.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.448197603225708,
+      "eval_runtime": 0.5111,
+      "eval_samples_per_second": 136.954,
+      "eval_steps_per_second": 5.869,
+      "step": 272
+    },
+    {
+      "epoch": 136.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4480949342250824,
+      "eval_runtime": 0.5102,
+      "eval_samples_per_second": 137.192,
+      "eval_steps_per_second": 5.88,
+      "step": 274
+    },
+    {
+      "epoch": 137.44,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 0.2597,
+      "step": 275
+    },
+    {
+      "epoch": 137.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.447935551404953,
+      "eval_runtime": 0.5123,
+      "eval_samples_per_second": 136.645,
+      "eval_steps_per_second": 5.856,
+      "step": 276
+    },
+    {
+      "epoch": 138.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44773900508880615,
+      "eval_runtime": 0.5066,
+      "eval_samples_per_second": 138.167,
+      "eval_steps_per_second": 5.921,
+      "step": 278
+    },
+    {
+      "epoch": 139.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44755300879478455,
+      "eval_runtime": 0.5128,
+      "eval_samples_per_second": 136.51,
+      "eval_steps_per_second": 5.85,
+      "step": 280
+    },
+    {
+      "epoch": 140.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44738340377807617,
+      "eval_runtime": 0.5088,
+      "eval_samples_per_second": 137.582,
+      "eval_steps_per_second": 5.896,
+      "step": 282
+    },
+    {
+      "epoch": 141.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4472770690917969,
+      "eval_runtime": 0.5106,
+      "eval_samples_per_second": 137.086,
+      "eval_steps_per_second": 5.875,
+      "step": 284
+    },
+    {
+      "epoch": 142.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44719263911247253,
+      "eval_runtime": 0.5111,
+      "eval_samples_per_second": 136.971,
+      "eval_steps_per_second": 5.87,
+      "step": 286
+    },
+    {
+      "epoch": 143.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44712573289871216,
+      "eval_runtime": 0.519,
+      "eval_samples_per_second": 134.866,
+      "eval_steps_per_second": 5.78,
+      "step": 288
+    },
+    {
+      "epoch": 144.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44704240560531616,
+      "eval_runtime": 0.5125,
+      "eval_samples_per_second": 136.589,
+      "eval_steps_per_second": 5.854,
+      "step": 290
+    },
+    {
+      "epoch": 145.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4469843804836273,
+      "eval_runtime": 0.5194,
+      "eval_samples_per_second": 134.777,
+      "eval_steps_per_second": 5.776,
+      "step": 292
+    },
+    {
+      "epoch": 146.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44694092869758606,
+      "eval_runtime": 0.5135,
+      "eval_samples_per_second": 136.31,
+      "eval_steps_per_second": 5.842,
+      "step": 294
+    },
+    {
+      "epoch": 147.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.4469132423400879,
+      "eval_runtime": 0.5162,
+      "eval_samples_per_second": 135.603,
+      "eval_steps_per_second": 5.812,
+      "step": 296
+    },
+    {
+      "epoch": 148.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44689762592315674,
+      "eval_runtime": 0.514,
+      "eval_samples_per_second": 136.176,
+      "eval_steps_per_second": 5.836,
+      "step": 298
+    },
+    {
+      "epoch": 149.89,
+      "learning_rate": 0.0,
+      "loss": 0.2556,
+      "step": 300
+    },
+    {
+      "epoch": 149.89,
+      "eval_f1": 0.9032258064516129,
+      "eval_loss": 0.44689178466796875,
+      "eval_runtime": 0.5161,
+      "eval_samples_per_second": 135.631,
+      "eval_steps_per_second": 5.813,
+      "step": 300
+    },
+    {
+      "epoch": 149.89,
+      "step": 300,
+      "total_flos": 1.0435256966870508e+18,
+      "train_loss": 0.3680222670237223,
+      "train_runtime": 877.204,
+      "train_samples_per_second": 47.366,
+      "train_steps_per_second": 0.342
     }
   ],
-  "max_steps": 150,
-  "num_train_epochs": 75,
-  "total_flos": 5.2149900813992755e+17,
+  "max_steps": 300,
+  "num_train_epochs": 150,
+  "total_flos": 1.0435256966870508e+18,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:267ba35e6adf6eb6c01f73e02dabe9f9040c5d7d0b2abeece88e0e9424fc738b
+oid sha256:40d1617033179273eb9bed1677a63638edc820431ce580d577ec90836a4cfcb0
 size 3311