End of training

all_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 99.0,
+    "eval_accuracy": 0.7685185185185185,
+    "eval_loss": 0.7077056765556335,
+    "eval_runtime": 166.0464,
+    "eval_samples_per_second": 1.301,
+    "eval_steps_per_second": 0.163
+}

test_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 99.0,
+    "eval_accuracy": 0.7685185185185185,
+    "eval_loss": 0.7077056765556335,
+    "eval_runtime": 166.0464,
+    "eval_samples_per_second": 1.301,
+    "eval_steps_per_second": 0.163
+}

trainer_state.json

@@ -0,0 +1,4278 @@
+{
+  "best_metric": 0.815668202764977,
+  "best_model_checkpoint": "videomae-base-finetuned-subset-100epochs/checkpoint-2352",
+  "epoch": 99.00108108108108,
+  "eval_steps": 500,
+  "global_step": 5550,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.00900900900901e-07,
+      "loss": 1.7041,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-06,
+      "loss": 1.6707,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.588,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.603603603603604e-06,
+      "loss": 1.6605,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.504504504504505e-06,
+      "loss": 1.6657,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "eval_accuracy": 0.22580645161290322,
+      "eval_loss": 1.6248142719268799,
+      "eval_runtime": 177.2019,
+      "eval_samples_per_second": 1.225,
+      "eval_steps_per_second": 0.158,
+      "step": 56
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.6433,
+      "step": 60
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.306306306306306e-06,
+      "loss": 1.6323,
+      "step": 70
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.207207207207208e-06,
+      "loss": 1.6449,
+      "step": 80
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.5961,
+      "step": 90
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.00900900900901e-06,
+      "loss": 1.6435,
+      "step": 100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.90990990990991e-06,
+      "loss": 1.6109,
+      "step": 110
+    },
+    {
+      "epoch": 1.01,
+      "eval_accuracy": 0.391705069124424,
+      "eval_loss": 1.5600569248199463,
+      "eval_runtime": 170.9412,
+      "eval_samples_per_second": 1.269,
+      "eval_steps_per_second": 0.164,
+      "step": 112
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 1.5845,
+      "step": 120
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1711711711711713e-05,
+      "loss": 1.6078,
+      "step": 130
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.2612612612612611e-05,
+      "loss": 1.6136,
+      "step": 140
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 1.5919,
+      "step": 150
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.4414414414414416e-05,
+      "loss": 1.5669,
+      "step": 160
+    },
+    {
+      "epoch": 2.01,
+      "eval_accuracy": 0.37327188940092165,
+      "eval_loss": 1.5562978982925415,
+      "eval_runtime": 172.5259,
+      "eval_samples_per_second": 1.258,
+      "eval_steps_per_second": 0.162,
+      "step": 168
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5315315315315316e-05,
+      "loss": 1.5883,
+      "step": 170
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 1.5374,
+      "step": 180
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.7117117117117117e-05,
+      "loss": 1.5314,
+      "step": 190
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.801801801801802e-05,
+      "loss": 1.4725,
+      "step": 200
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 1.4064,
+      "step": 210
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.981981981981982e-05,
+      "loss": 1.45,
+      "step": 220
+    },
+    {
+      "epoch": 3.01,
+      "eval_accuracy": 0.5990783410138248,
+      "eval_loss": 1.0987794399261475,
+      "eval_runtime": 176.949,
+      "eval_samples_per_second": 1.226,
+      "eval_steps_per_second": 0.158,
+      "step": 224
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0720720720720722e-05,
+      "loss": 1.4079,
+      "step": 230
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 1.2881,
+      "step": 240
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.2522522522522523e-05,
+      "loss": 1.2991,
+      "step": 250
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.3423423423423425e-05,
+      "loss": 1.2036,
+      "step": 260
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.4324324324324327e-05,
+      "loss": 1.1179,
+      "step": 270
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.5225225225225222e-05,
+      "loss": 1.1208,
+      "step": 280
+    },
+    {
+      "epoch": 4.01,
+      "eval_accuracy": 0.5714285714285714,
+      "eval_loss": 1.2278647422790527,
+      "eval_runtime": 175.1133,
+      "eval_samples_per_second": 1.239,
+      "eval_steps_per_second": 0.16,
+      "step": 280
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.6126126126126128e-05,
+      "loss": 1.2803,
+      "step": 290
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 1.2979,
+      "step": 300
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.7927927927927926e-05,
+      "loss": 1.2177,
+      "step": 310
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.882882882882883e-05,
+      "loss": 1.1382,
+      "step": 320
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.9729729729729733e-05,
+      "loss": 1.1588,
+      "step": 330
+    },
+    {
+      "epoch": 5.01,
+      "eval_accuracy": 0.7096774193548387,
+      "eval_loss": 0.8424109816551208,
+      "eval_runtime": 164.0794,
+      "eval_samples_per_second": 1.323,
+      "eval_steps_per_second": 0.171,
+      "step": 336
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.063063063063063e-05,
+      "loss": 1.2975,
+      "step": 340
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.153153153153153e-05,
+      "loss": 1.1493,
+      "step": 350
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 1.0285,
+      "step": 360
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.9137,
+      "step": 370
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.4234234234234234e-05,
+      "loss": 0.9867,
+      "step": 380
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 3.513513513513514e-05,
+      "loss": 1.0834,
+      "step": 390
+    },
+    {
+      "epoch": 6.01,
+      "eval_accuracy": 0.5345622119815668,
+      "eval_loss": 1.103454351425171,
+      "eval_runtime": 178.4343,
+      "eval_samples_per_second": 1.216,
+      "eval_steps_per_second": 0.157,
+      "step": 392
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.603603603603604e-05,
+      "loss": 1.0298,
+      "step": 400
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 3.693693693693694e-05,
+      "loss": 1.0116,
+      "step": 410
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 0.9508,
+      "step": 420
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.873873873873874e-05,
+      "loss": 1.2421,
+      "step": 430
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 3.963963963963964e-05,
+      "loss": 1.2194,
+      "step": 440
+    },
+    {
+      "epoch": 7.01,
+      "eval_accuracy": 0.4838709677419355,
+      "eval_loss": 1.0748963356018066,
+      "eval_runtime": 175.7888,
+      "eval_samples_per_second": 1.234,
+      "eval_steps_per_second": 0.159,
+      "step": 448
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.0540540540540545e-05,
+      "loss": 1.1793,
+      "step": 450
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.1441441441441444e-05,
+      "loss": 1.0139,
+      "step": 460
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 4.234234234234234e-05,
+      "loss": 1.0531,
+      "step": 470
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 1.098,
+      "step": 480
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.414414414414415e-05,
+      "loss": 0.8256,
+      "step": 490
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 4.5045045045045046e-05,
+      "loss": 0.8462,
+      "step": 500
+    },
+    {
+      "epoch": 8.01,
+      "eval_accuracy": 0.6405529953917051,
+      "eval_loss": 0.8755273222923279,
+      "eval_runtime": 175.5514,
+      "eval_samples_per_second": 1.236,
+      "eval_steps_per_second": 0.159,
+      "step": 504
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.594594594594595e-05,
+      "loss": 1.0328,
+      "step": 510
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.684684684684685e-05,
+      "loss": 1.1518,
+      "step": 520
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.774774774774775e-05,
+      "loss": 1.0959,
+      "step": 530
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 0.9388,
+      "step": 540
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.954954954954955e-05,
+      "loss": 0.937,
+      "step": 550
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.994994994994995e-05,
+      "loss": 1.058,
+      "step": 560
+    },
+    {
+      "epoch": 9.01,
+      "eval_accuracy": 0.6497695852534562,
+      "eval_loss": 0.9025103449821472,
+      "eval_runtime": 177.1283,
+      "eval_samples_per_second": 1.225,
+      "eval_steps_per_second": 0.158,
+      "step": 560
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.984984984984985e-05,
+      "loss": 0.8948,
+      "step": 570
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.974974974974975e-05,
+      "loss": 0.9187,
+      "step": 580
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.964964964964965e-05,
+      "loss": 0.8642,
+      "step": 590
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.954954954954955e-05,
+      "loss": 0.8694,
+      "step": 600
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.944944944944945e-05,
+      "loss": 1.0163,
+      "step": 610
+    },
+    {
+      "epoch": 10.01,
+      "eval_accuracy": 0.4838709677419355,
+      "eval_loss": 1.2587543725967407,
+      "eval_runtime": 175.3167,
+      "eval_samples_per_second": 1.238,
+      "eval_steps_per_second": 0.16,
+      "step": 616
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.9349349349349347e-05,
+      "loss": 1.0813,
+      "step": 620
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.9249249249249253e-05,
+      "loss": 0.8843,
+      "step": 630
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.9149149149149154e-05,
+      "loss": 0.9094,
+      "step": 640
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.9049049049049054e-05,
+      "loss": 0.9094,
+      "step": 650
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.8948948948948954e-05,
+      "loss": 0.7727,
+      "step": 660
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.884884884884885e-05,
+      "loss": 1.0639,
+      "step": 670
+    },
+    {
+      "epoch": 11.01,
+      "eval_accuracy": 0.6359447004608295,
+      "eval_loss": 0.8927530646324158,
+      "eval_runtime": 173.189,
+      "eval_samples_per_second": 1.253,
+      "eval_steps_per_second": 0.162,
+      "step": 672
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.8748748748748754e-05,
+      "loss": 1.0281,
+      "step": 680
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.8648648648648654e-05,
+      "loss": 0.8524,
+      "step": 690
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.854854854854855e-05,
+      "loss": 0.963,
+      "step": 700
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.8448448448448455e-05,
+      "loss": 0.823,
+      "step": 710
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.834834834834835e-05,
+      "loss": 0.9317,
+      "step": 720
+    },
+    {
+      "epoch": 12.01,
+      "eval_accuracy": 0.6221198156682027,
+      "eval_loss": 0.8824858069419861,
+      "eval_runtime": 171.7569,
+      "eval_samples_per_second": 1.263,
+      "eval_steps_per_second": 0.163,
+      "step": 728
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.824824824824825e-05,
+      "loss": 0.8926,
+      "step": 730
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.8433,
+      "step": 740
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.804804804804805e-05,
+      "loss": 1.0372,
+      "step": 750
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.7947947947947955e-05,
+      "loss": 1.0262,
+      "step": 760
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.784784784784785e-05,
+      "loss": 0.8883,
+      "step": 770
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.774774774774775e-05,
+      "loss": 0.9038,
+      "step": 780
+    },
+    {
+      "epoch": 13.01,
+      "eval_accuracy": 0.5622119815668203,
+      "eval_loss": 0.8764956593513489,
+      "eval_runtime": 172.944,
+      "eval_samples_per_second": 1.255,
+      "eval_steps_per_second": 0.162,
+      "step": 784
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.764764764764765e-05,
+      "loss": 0.9196,
+      "step": 790
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.754754754754755e-05,
+      "loss": 1.0407,
+      "step": 800
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.744744744744745e-05,
+      "loss": 0.9383,
+      "step": 810
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.734734734734735e-05,
+      "loss": 0.8965,
+      "step": 820
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.724724724724725e-05,
+      "loss": 0.9297,
+      "step": 830
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.714714714714715e-05,
+      "loss": 0.9155,
+      "step": 840
+    },
+    {
+      "epoch": 14.01,
+      "eval_accuracy": 0.7004608294930875,
+      "eval_loss": 0.8431310653686523,
+      "eval_runtime": 172.9883,
+      "eval_samples_per_second": 1.254,
+      "eval_steps_per_second": 0.162,
+      "step": 840
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.704704704704705e-05,
+      "loss": 0.9725,
+      "step": 850
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.694694694694695e-05,
+      "loss": 0.8736,
+      "step": 860
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.684684684684685e-05,
+      "loss": 0.8519,
+      "step": 870
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.674674674674675e-05,
+      "loss": 0.7844,
+      "step": 880
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.6646646646646644e-05,
+      "loss": 1.0731,
+      "step": 890
+    },
+    {
+      "epoch": 15.01,
+      "eval_accuracy": 0.7004608294930875,
+      "eval_loss": 0.8174832463264465,
+      "eval_runtime": 171.3687,
+      "eval_samples_per_second": 1.266,
+      "eval_steps_per_second": 0.163,
+      "step": 896
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.654654654654655e-05,
+      "loss": 0.7801,
+      "step": 900
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.644644644644645e-05,
+      "loss": 0.7508,
+      "step": 910
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.634634634634635e-05,
+      "loss": 0.8745,
+      "step": 920
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.624624624624625e-05,
+      "loss": 0.786,
+      "step": 930
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.6146146146146144e-05,
+      "loss": 0.8919,
+      "step": 940
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.604604604604605e-05,
+      "loss": 0.6864,
+      "step": 950
+    },
+    {
+      "epoch": 16.01,
+      "eval_accuracy": 0.5852534562211982,
+      "eval_loss": 1.059116005897522,
+      "eval_runtime": 172.8249,
+      "eval_samples_per_second": 1.256,
+      "eval_steps_per_second": 0.162,
+      "step": 952
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.594594594594595e-05,
+      "loss": 0.8259,
+      "step": 960
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.5845845845845845e-05,
+      "loss": 0.7213,
+      "step": 970
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.574574574574575e-05,
+      "loss": 0.6611,
+      "step": 980
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.5645645645645645e-05,
+      "loss": 0.8739,
+      "step": 990
+    },
+    {
+      "epoch": 17.01,
+      "learning_rate": 4.5545545545545545e-05,
+      "loss": 0.9537,
+      "step": 1000
+    },
+    {
+      "epoch": 17.01,
+      "eval_accuracy": 0.6221198156682027,
+      "eval_loss": 0.9702848792076111,
+      "eval_runtime": 172.8825,
+      "eval_samples_per_second": 1.255,
+      "eval_steps_per_second": 0.162,
+      "step": 1008
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 4.544544544544545e-05,
+      "loss": 0.9157,
+      "step": 1010
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 4.5345345345345345e-05,
+      "loss": 0.93,
+      "step": 1020
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 4.524524524524525e-05,
+      "loss": 0.7541,
+      "step": 1030
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.5145145145145146e-05,
+      "loss": 0.9246,
+      "step": 1040
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.5045045045045046e-05,
+      "loss": 0.8477,
+      "step": 1050
+    },
+    {
+      "epoch": 18.01,
+      "learning_rate": 4.4944944944944946e-05,
+      "loss": 0.7499,
+      "step": 1060
+    },
+    {
+      "epoch": 18.01,
+      "eval_accuracy": 0.5806451612903226,
+      "eval_loss": 0.8370996117591858,
+      "eval_runtime": 164.6146,
+      "eval_samples_per_second": 1.318,
+      "eval_steps_per_second": 0.17,
+      "step": 1064
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 4.4844844844844846e-05,
+      "loss": 0.6531,
+      "step": 1070
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 4.4744744744744746e-05,
+      "loss": 1.0249,
+      "step": 1080
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 4.4644644644644646e-05,
+      "loss": 0.8424,
+      "step": 1090
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 4.4544544544544546e-05,
+      "loss": 0.6561,
+      "step": 1100
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.6909,
+      "step": 1110
+    },
+    {
+      "epoch": 19.01,
+      "learning_rate": 4.434434434434435e-05,
+      "loss": 0.7142,
+      "step": 1120
+    },
+    {
+      "epoch": 19.01,
+      "eval_accuracy": 0.663594470046083,
+      "eval_loss": 0.9132143259048462,
+      "eval_runtime": 179.8294,
+      "eval_samples_per_second": 1.207,
+      "eval_steps_per_second": 0.156,
+      "step": 1120
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 4.424424424424425e-05,
+      "loss": 0.7644,
+      "step": 1130
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 4.414414414414415e-05,
+      "loss": 0.7901,
+      "step": 1140
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 4.404404404404405e-05,
+      "loss": 0.751,
+      "step": 1150
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 4.394394394394394e-05,
+      "loss": 0.9573,
+      "step": 1160
+    },
+    {
+      "epoch": 20.01,
+      "learning_rate": 4.384384384384385e-05,
+      "loss": 0.675,
+      "step": 1170
+    },
+    {
+      "epoch": 20.01,
+      "eval_accuracy": 0.6728110599078341,
+      "eval_loss": 0.759711503982544,
+      "eval_runtime": 176.3584,
+      "eval_samples_per_second": 1.23,
+      "eval_steps_per_second": 0.159,
+      "step": 1176
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 4.374374374374375e-05,
+      "loss": 0.6662,
+      "step": 1180
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 4.364364364364365e-05,
+      "loss": 0.6619,
+      "step": 1190
+    },
+    {
+      "epoch": 21.0,
+      "learning_rate": 4.354354354354355e-05,
+      "loss": 0.7366,
+      "step": 1200
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 4.344344344344344e-05,
+      "loss": 0.7486,
+      "step": 1210
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 4.334334334334335e-05,
+      "loss": 0.8793,
+      "step": 1220
+    },
+    {
+      "epoch": 21.01,
+      "learning_rate": 4.324324324324325e-05,
+      "loss": 0.604,
+      "step": 1230
+    },
+    {
+      "epoch": 21.01,
+      "eval_accuracy": 0.5714285714285714,
+      "eval_loss": 1.2003726959228516,
+      "eval_runtime": 174.3155,
+      "eval_samples_per_second": 1.245,
+      "eval_steps_per_second": 0.161,
+      "step": 1232
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 4.314314314314314e-05,
+      "loss": 0.7404,
+      "step": 1240
+    },
+    {
+      "epoch": 22.0,
+      "learning_rate": 4.304304304304305e-05,
+      "loss": 0.631,
+      "step": 1250
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 4.294294294294294e-05,
+      "loss": 0.7146,
+      "step": 1260
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 4.284284284284284e-05,
+      "loss": 0.7736,
+      "step": 1270
+    },
+    {
+      "epoch": 22.01,
+      "learning_rate": 4.274274274274275e-05,
+      "loss": 0.7738,
+      "step": 1280
+    },
+    {
+      "epoch": 22.01,
+      "eval_accuracy": 0.5668202764976958,
+      "eval_loss": 1.0632843971252441,
+      "eval_runtime": 173.5303,
+      "eval_samples_per_second": 1.251,
+      "eval_steps_per_second": 0.161,
+      "step": 1288
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.264264264264264e-05,
+      "loss": 0.6972,
+      "step": 1290
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.254254254254255e-05,
+      "loss": 0.7115,
+      "step": 1300
+    },
+    {
+      "epoch": 23.0,
+      "learning_rate": 4.244244244244244e-05,
+      "loss": 0.5288,
+      "step": 1310
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.234234234234234e-05,
+      "loss": 0.7268,
+      "step": 1320
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.224224224224225e-05,
+      "loss": 0.5663,
+      "step": 1330
+    },
+    {
+      "epoch": 23.01,
+      "learning_rate": 4.214214214214214e-05,
+      "loss": 0.7651,
+      "step": 1340
+    },
+    {
+      "epoch": 23.01,
+      "eval_accuracy": 0.6820276497695853,
+      "eval_loss": 0.6864995360374451,
+      "eval_runtime": 171.8309,
+      "eval_samples_per_second": 1.263,
+      "eval_steps_per_second": 0.163,
+      "step": 1344
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 4.204204204204204e-05,
+      "loss": 0.5517,
+      "step": 1350
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 4.194194194194194e-05,
+      "loss": 0.6645,
+      "step": 1360
+    },
+    {
+      "epoch": 24.0,
+      "learning_rate": 4.1841841841841843e-05,
+      "loss": 0.5702,
+      "step": 1370
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.1741741741741744e-05,
+      "loss": 0.8262,
+      "step": 1380
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.1641641641641644e-05,
+      "loss": 0.9137,
+      "step": 1390
+    },
+    {
+      "epoch": 24.01,
+      "learning_rate": 4.1541541541541544e-05,
+      "loss": 0.6292,
+      "step": 1400
+    },
+    {
+      "epoch": 24.01,
+      "eval_accuracy": 0.6912442396313364,
+      "eval_loss": 0.7606768012046814,
+      "eval_runtime": 169.2507,
+      "eval_samples_per_second": 1.282,
+      "eval_steps_per_second": 0.165,
+      "step": 1400
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 4.1441441441441444e-05,
+      "loss": 0.6136,
+      "step": 1410
+    },
+    {
+      "epoch": 25.0,
+      "learning_rate": 4.1341341341341344e-05,
+      "loss": 0.5996,
+      "step": 1420
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.124124124124124e-05,
+      "loss": 0.7239,
+      "step": 1430
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.1141141141141144e-05,
+      "loss": 0.5766,
+      "step": 1440
+    },
+    {
+      "epoch": 25.01,
+      "learning_rate": 4.1041041041041045e-05,
+      "loss": 0.7387,
+      "step": 1450
+    },
+    {
+      "epoch": 25.01,
+      "eval_accuracy": 0.5345622119815668,
+      "eval_loss": 1.303768277168274,
+      "eval_runtime": 164.0294,
+      "eval_samples_per_second": 1.323,
+      "eval_steps_per_second": 0.171,
+      "step": 1456
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.0940940940940945e-05,
+      "loss": 0.7438,
+      "step": 1460
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.0840840840840845e-05,
+      "loss": 0.7224,
+      "step": 1470
+    },
+    {
+      "epoch": 26.0,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.9601,
+      "step": 1480
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.0640640640640645e-05,
+      "loss": 0.6726,
+      "step": 1490
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.0540540540540545e-05,
+      "loss": 0.7734,
+      "step": 1500
+    },
+    {
+      "epoch": 26.01,
+      "learning_rate": 4.044044044044044e-05,
+      "loss": 0.7038,
+      "step": 1510
+    },
+    {
+      "epoch": 26.01,
+      "eval_accuracy": 0.5529953917050692,
+      "eval_loss": 1.2832341194152832,
+      "eval_runtime": 163.9119,
+      "eval_samples_per_second": 1.324,
+      "eval_steps_per_second": 0.171,
+      "step": 1512
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.0340340340340346e-05,
+      "loss": 0.4438,
+      "step": 1520
+    },
+    {
+      "epoch": 27.0,
+      "learning_rate": 4.024024024024024e-05,
+      "loss": 0.5555,
+      "step": 1530
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 4.014014014014014e-05,
+      "loss": 0.9817,
+      "step": 1540
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 4.0040040040040046e-05,
+      "loss": 0.8025,
+      "step": 1550
+    },
+    {
+      "epoch": 27.01,
+      "learning_rate": 3.993993993993994e-05,
+      "loss": 0.7565,
+      "step": 1560
+    },
+    {
+      "epoch": 27.01,
+      "eval_accuracy": 0.7004608294930875,
+      "eval_loss": 0.8127643465995789,
+      "eval_runtime": 165.3212,
+      "eval_samples_per_second": 1.313,
+      "eval_steps_per_second": 0.169,
+      "step": 1568
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 3.9839839839839846e-05,
+      "loss": 0.6489,
+      "step": 1570
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 3.973973973973974e-05,
+      "loss": 0.6275,
+      "step": 1580
+    },
+    {
+      "epoch": 28.0,
+      "learning_rate": 3.963963963963964e-05,
+      "loss": 0.7778,
+      "step": 1590
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 3.953953953953955e-05,
+      "loss": 0.7334,
+      "step": 1600
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 3.943943943943944e-05,
+      "loss": 0.839,
+      "step": 1610
+    },
+    {
+      "epoch": 28.01,
+      "learning_rate": 3.933933933933934e-05,
+      "loss": 0.6516,
+      "step": 1620
+    },
+    {
+      "epoch": 28.01,
+      "eval_accuracy": 0.5391705069124424,
+      "eval_loss": 1.0892592668533325,
+      "eval_runtime": 164.7722,
+      "eval_samples_per_second": 1.317,
+      "eval_steps_per_second": 0.17,
+      "step": 1624
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 3.923923923923924e-05,
+      "loss": 0.7291,
+      "step": 1630
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 3.913913913913914e-05,
+      "loss": 0.5831,
+      "step": 1640
+    },
+    {
+      "epoch": 29.0,
+      "learning_rate": 3.903903903903904e-05,
+      "loss": 0.7305,
+      "step": 1650
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 3.893893893893894e-05,
+      "loss": 0.6436,
+      "step": 1660
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 3.883883883883884e-05,
+      "loss": 0.7121,
+      "step": 1670
+    },
+    {
+      "epoch": 29.01,
+      "learning_rate": 3.873873873873874e-05,
+      "loss": 0.7074,
+      "step": 1680
+    },
+    {
+      "epoch": 29.01,
+      "eval_accuracy": 0.5990783410138248,
+      "eval_loss": 1.08941650390625,
+      "eval_runtime": 163.3245,
+      "eval_samples_per_second": 1.329,
+      "eval_steps_per_second": 0.171,
+      "step": 1680
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3.863863863863864e-05,
+      "loss": 0.6485,
+      "step": 1690
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 3.8538538538538534e-05,
+      "loss": 0.672,
+      "step": 1700
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3.843843843843844e-05,
+      "loss": 0.6947,
+      "step": 1710
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3.833833833833834e-05,
+      "loss": 0.7492,
+      "step": 1720
+    },
+    {
+      "epoch": 30.01,
+      "learning_rate": 3.823823823823824e-05,
+      "loss": 0.4902,
+      "step": 1730
+    },
+    {
+      "epoch": 30.01,
+      "eval_accuracy": 0.5622119815668203,
+      "eval_loss": 1.0694791078567505,
+      "eval_runtime": 166.2845,
+      "eval_samples_per_second": 1.305,
+      "eval_steps_per_second": 0.168,
+      "step": 1736
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.813813813813814e-05,
+      "loss": 0.6269,
+      "step": 1740
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.8038038038038035e-05,
+      "loss": 0.871,
+      "step": 1750
+    },
+    {
+      "epoch": 31.0,
+      "learning_rate": 3.793793793793794e-05,
+      "loss": 0.4449,
+      "step": 1760
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3.783783783783784e-05,
+      "loss": 0.8107,
+      "step": 1770
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3.7737737737737736e-05,
+      "loss": 0.7082,
+      "step": 1780
+    },
+    {
+      "epoch": 31.01,
+      "learning_rate": 3.763763763763764e-05,
+      "loss": 0.4563,
+      "step": 1790
+    },
+    {
+      "epoch": 31.01,
+      "eval_accuracy": 0.5299539170506913,
+      "eval_loss": 1.2921912670135498,
+      "eval_runtime": 164.2241,
+      "eval_samples_per_second": 1.321,
+      "eval_steps_per_second": 0.17,
+      "step": 1792
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.7537537537537536e-05,
+      "loss": 0.6189,
+      "step": 1800
+    },
+    {
+      "epoch": 32.0,
+      "learning_rate": 3.7437437437437436e-05,
+      "loss": 0.689,
+      "step": 1810
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.733733733733734e-05,
+      "loss": 0.6324,
+      "step": 1820
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.7237237237237236e-05,
+      "loss": 0.6537,
+      "step": 1830
+    },
+    {
+      "epoch": 32.01,
+      "learning_rate": 3.713713713713714e-05,
+      "loss": 0.7543,
+      "step": 1840
+    },
+    {
+      "epoch": 32.01,
+      "eval_accuracy": 0.6820276497695853,
+      "eval_loss": 0.8960239887237549,
+      "eval_runtime": 174.6812,
+      "eval_samples_per_second": 1.242,
+      "eval_steps_per_second": 0.16,
+      "step": 1848
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.8928,
+      "step": 1850
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.693693693693694e-05,
+      "loss": 0.6618,
+      "step": 1860
+    },
+    {
+      "epoch": 33.0,
+      "learning_rate": 3.6836836836836844e-05,
+      "loss": 0.6464,
+      "step": 1870
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3.673673673673674e-05,
+      "loss": 0.5765,
+      "step": 1880
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3.663663663663664e-05,
+      "loss": 0.5994,
+      "step": 1890
+    },
+    {
+      "epoch": 33.01,
+      "learning_rate": 3.653653653653654e-05,
+      "loss": 0.7467,
+      "step": 1900
+    },
+    {
+      "epoch": 33.01,
+      "eval_accuracy": 0.7465437788018433,
+      "eval_loss": 0.7861269116401672,
+      "eval_runtime": 174.8536,
+      "eval_samples_per_second": 1.241,
+      "eval_steps_per_second": 0.16,
+      "step": 1904
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.643643643643644e-05,
+      "loss": 0.5162,
+      "step": 1910
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.633633633633634e-05,
+      "loss": 0.5848,
+      "step": 1920
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.623623623623624e-05,
+      "loss": 0.5114,
+      "step": 1930
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.613613613613614e-05,
+      "loss": 0.5392,
+      "step": 1940
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.603603603603604e-05,
+      "loss": 0.5633,
+      "step": 1950
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.593593593593594e-05,
+      "loss": 0.6459,
+      "step": 1960
+    },
+    {
+      "epoch": 34.01,
+      "eval_accuracy": 0.5622119815668203,
+      "eval_loss": 1.283542275428772,
+      "eval_runtime": 176.7914,
+      "eval_samples_per_second": 1.227,
+      "eval_steps_per_second": 0.158,
+      "step": 1960
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.583583583583583e-05,
+      "loss": 0.533,
+      "step": 1970
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.573573573573574e-05,
+      "loss": 0.7035,
+      "step": 1980
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.563563563563564e-05,
+      "loss": 0.6015,
+      "step": 1990
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.553553553553554e-05,
+      "loss": 0.7309,
+      "step": 2000
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.543543543543544e-05,
+      "loss": 0.7296,
+      "step": 2010
+    },
+    {
+      "epoch": 35.01,
+      "eval_accuracy": 0.5806451612903226,
+      "eval_loss": 1.0303059816360474,
+      "eval_runtime": 175.2838,
+      "eval_samples_per_second": 1.238,
+      "eval_steps_per_second": 0.16,
+      "step": 2016
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.533533533533533e-05,
+      "loss": 0.6262,
+      "step": 2020
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.523523523523524e-05,
+      "loss": 0.6559,
+      "step": 2030
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.513513513513514e-05,
+      "loss": 0.5336,
+      "step": 2040
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.503503503503503e-05,
+      "loss": 0.6212,
+      "step": 2050
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.493493493493494e-05,
+      "loss": 0.8769,
+      "step": 2060
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.483483483483483e-05,
+      "loss": 0.5,
+      "step": 2070
+    },
+    {
+      "epoch": 36.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 0.8923883438110352,
+      "eval_runtime": 172.7445,
+      "eval_samples_per_second": 1.256,
+      "eval_steps_per_second": 0.162,
+      "step": 2072
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.473473473473473e-05,
+      "loss": 0.3772,
+      "step": 2080
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.463463463463464e-05,
+      "loss": 0.5472,
+      "step": 2090
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.453453453453453e-05,
+      "loss": 0.3361,
+      "step": 2100
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.443443443443444e-05,
+      "loss": 0.8774,
+      "step": 2110
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.4334334334334334e-05,
+      "loss": 0.5181,
+      "step": 2120
+    },
+    {
+      "epoch": 37.01,
+      "eval_accuracy": 0.7235023041474654,
+      "eval_loss": 0.8768814206123352,
+      "eval_runtime": 169.9168,
+      "eval_samples_per_second": 1.277,
+      "eval_steps_per_second": 0.165,
+      "step": 2128
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.4234234234234234e-05,
+      "loss": 0.667,
+      "step": 2130
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.413413413413414e-05,
+      "loss": 0.5336,
+      "step": 2140
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.4034034034034034e-05,
+      "loss": 0.7924,
+      "step": 2150
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.3933933933933934e-05,
+      "loss": 0.6568,
+      "step": 2160
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.3833833833833834e-05,
+      "loss": 0.666,
+      "step": 2170
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.3733733733733734e-05,
+      "loss": 0.5225,
+      "step": 2180
+    },
+    {
+      "epoch": 38.01,
+      "eval_accuracy": 0.7511520737327189,
+      "eval_loss": 0.7287940382957458,
+      "eval_runtime": 165.8928,
+      "eval_samples_per_second": 1.308,
+      "eval_steps_per_second": 0.169,
+      "step": 2184
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.3633633633633635e-05,
+      "loss": 0.6175,
+      "step": 2190
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.3533533533533535e-05,
+      "loss": 0.4297,
+      "step": 2200
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.3433433433433435e-05,
+      "loss": 0.4564,
+      "step": 2210
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.7067,
+      "step": 2220
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.3233233233233235e-05,
+      "loss": 0.7556,
+      "step": 2230
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.3133133133133135e-05,
+      "loss": 0.5617,
+      "step": 2240
+    },
+    {
+      "epoch": 39.01,
+      "eval_accuracy": 0.7926267281105991,
+      "eval_loss": 0.6330269575119019,
+      "eval_runtime": 164.5501,
+      "eval_samples_per_second": 1.319,
+      "eval_steps_per_second": 0.17,
+      "step": 2240
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3.3033033033033035e-05,
+      "loss": 0.4753,
+      "step": 2250
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3.2932932932932935e-05,
+      "loss": 0.5791,
+      "step": 2260
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 3.2832832832832836e-05,
+      "loss": 0.4268,
+      "step": 2270
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 3.2732732732732736e-05,
+      "loss": 0.5892,
+      "step": 2280
+    },
+    {
+      "epoch": 40.01,
+      "learning_rate": 3.263263263263263e-05,
+      "loss": 0.677,
+      "step": 2290
+    },
+    {
+      "epoch": 40.01,
+      "eval_accuracy": 0.7419354838709677,
+      "eval_loss": 0.7732899785041809,
+      "eval_runtime": 164.8066,
+      "eval_samples_per_second": 1.317,
+      "eval_steps_per_second": 0.17,
+      "step": 2296
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 3.2532532532532536e-05,
+      "loss": 0.6502,
+      "step": 2300
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 3.2432432432432436e-05,
+      "loss": 0.5512,
+      "step": 2310
+    },
+    {
+      "epoch": 41.0,
+      "learning_rate": 3.233233233233233e-05,
+      "loss": 0.6401,
+      "step": 2320
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 3.2232232232232236e-05,
+      "loss": 0.3983,
+      "step": 2330
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 3.213213213213213e-05,
+      "loss": 0.7081,
+      "step": 2340
+    },
+    {
+      "epoch": 41.01,
+      "learning_rate": 3.203203203203203e-05,
+      "loss": 0.6891,
+      "step": 2350
+    },
+    {
+      "epoch": 41.01,
+      "eval_accuracy": 0.815668202764977,
+      "eval_loss": 0.7462968826293945,
+      "eval_runtime": 163.9014,
+      "eval_samples_per_second": 1.324,
+      "eval_steps_per_second": 0.171,
+      "step": 2352
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 3.193193193193194e-05,
+      "loss": 0.3909,
+      "step": 2360
+    },
+    {
+      "epoch": 42.0,
+      "learning_rate": 3.183183183183183e-05,
+      "loss": 0.641,
+      "step": 2370
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 3.173173173173174e-05,
+      "loss": 0.5696,
+      "step": 2380
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 3.163163163163163e-05,
+      "loss": 0.419,
+      "step": 2390
+    },
+    {
+      "epoch": 42.01,
+      "learning_rate": 3.153153153153153e-05,
+      "loss": 0.6662,
+      "step": 2400
+    },
+    {
+      "epoch": 42.01,
+      "eval_accuracy": 0.7235023041474654,
+      "eval_loss": 0.9304085969924927,
+      "eval_runtime": 165.0958,
+      "eval_samples_per_second": 1.314,
+      "eval_steps_per_second": 0.17,
+      "step": 2408
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 3.143143143143144e-05,
+      "loss": 0.5243,
+      "step": 2410
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 3.133133133133133e-05,
+      "loss": 0.4339,
+      "step": 2420
+    },
+    {
+      "epoch": 43.0,
+      "learning_rate": 3.123123123123123e-05,
+      "loss": 0.4849,
+      "step": 2430
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 3.113113113113113e-05,
+      "loss": 0.85,
+      "step": 2440
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 3.103103103103103e-05,
+      "loss": 0.4044,
+      "step": 2450
+    },
+    {
+      "epoch": 43.01,
+      "learning_rate": 3.093093093093093e-05,
+      "loss": 0.4602,
+      "step": 2460
+    },
+    {
+      "epoch": 43.01,
+      "eval_accuracy": 0.5207373271889401,
+      "eval_loss": 1.5115013122558594,
+      "eval_runtime": 163.1033,
+      "eval_samples_per_second": 1.33,
+      "eval_steps_per_second": 0.172,
+      "step": 2464
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 3.083083083083083e-05,
+      "loss": 0.5138,
+      "step": 2470
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 3.073073073073073e-05,
+      "loss": 0.3286,
+      "step": 2480
+    },
+    {
+      "epoch": 44.0,
+      "learning_rate": 3.063063063063063e-05,
+      "loss": 0.4418,
+      "step": 2490
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 3.053053053053053e-05,
+      "loss": 0.4038,
+      "step": 2500
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 3.0430430430430436e-05,
+      "loss": 0.7055,
+      "step": 2510
+    },
+    {
+      "epoch": 44.01,
+      "learning_rate": 3.0330330330330332e-05,
+      "loss": 0.581,
+      "step": 2520
+    },
+    {
+      "epoch": 44.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.229552984237671,
+      "eval_runtime": 179.9416,
+      "eval_samples_per_second": 1.206,
+      "eval_steps_per_second": 0.156,
+      "step": 2520
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 3.0230230230230232e-05,
+      "loss": 0.5623,
+      "step": 2530
+    },
+    {
+      "epoch": 45.0,
+      "learning_rate": 3.013013013013013e-05,
+      "loss": 0.6803,
+      "step": 2540
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 3.0030030030030033e-05,
+      "loss": 0.6378,
+      "step": 2550
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 2.9929929929929933e-05,
+      "loss": 0.6153,
+      "step": 2560
+    },
+    {
+      "epoch": 45.01,
+      "learning_rate": 2.982982982982983e-05,
+      "loss": 0.5418,
+      "step": 2570
+    },
+    {
+      "epoch": 45.01,
+      "eval_accuracy": 0.6221198156682027,
+      "eval_loss": 1.0069782733917236,
+      "eval_runtime": 171.9564,
+      "eval_samples_per_second": 1.262,
+      "eval_steps_per_second": 0.163,
+      "step": 2576
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 2.9729729729729733e-05,
+      "loss": 0.489,
+      "step": 2580
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.4833,
+      "step": 2590
+    },
+    {
+      "epoch": 46.0,
+      "learning_rate": 2.952952952952953e-05,
+      "loss": 0.8491,
+      "step": 2600
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 2.9429429429429427e-05,
+      "loss": 0.4682,
+      "step": 2610
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 2.932932932932933e-05,
+      "loss": 0.6626,
+      "step": 2620
+    },
+    {
+      "epoch": 46.01,
+      "learning_rate": 2.9229229229229234e-05,
+      "loss": 0.5199,
+      "step": 2630
+    },
+    {
+      "epoch": 46.01,
+      "eval_accuracy": 0.6082949308755761,
+      "eval_loss": 1.1344462633132935,
+      "eval_runtime": 173.0896,
+      "eval_samples_per_second": 1.254,
+      "eval_steps_per_second": 0.162,
+      "step": 2632
+    },
+    {
+      "epoch": 47.0,
+      "learning_rate": 2.912912912912913e-05,
+      "loss": 0.6701,
+      "step": 2640
+    },
+    {
+      "epoch": 47.0,
+      "learning_rate": 2.902902902902903e-05,
+      "loss": 0.504,
+      "step": 2650
+    },
+    {
+      "epoch": 47.01,
+      "learning_rate": 2.8928928928928928e-05,
+      "loss": 0.5013,
+      "step": 2660
+    },
+    {
+      "epoch": 47.01,
+      "learning_rate": 2.882882882882883e-05,
+      "loss": 0.5753,
+      "step": 2670
+    },
+    {
+      "epoch": 47.01,
+      "learning_rate": 2.872872872872873e-05,
+      "loss": 0.6876,
+      "step": 2680
+    },
+    {
+      "epoch": 47.01,
+      "eval_accuracy": 0.576036866359447,
+      "eval_loss": 0.9799597859382629,
+      "eval_runtime": 175.2983,
+      "eval_samples_per_second": 1.238,
+      "eval_steps_per_second": 0.16,
+      "step": 2688
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 2.8628628628628628e-05,
+      "loss": 0.5597,
+      "step": 2690
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 2.852852852852853e-05,
+      "loss": 0.579,
+      "step": 2700
+    },
+    {
+      "epoch": 48.0,
+      "learning_rate": 2.8428428428428428e-05,
+      "loss": 0.5215,
+      "step": 2710
+    },
+    {
+      "epoch": 48.01,
+      "learning_rate": 2.832832832832833e-05,
+      "loss": 0.4575,
+      "step": 2720
+    },
+    {
+      "epoch": 48.01,
+      "learning_rate": 2.8228228228228232e-05,
+      "loss": 0.4567,
+      "step": 2730
+    },
+    {
+      "epoch": 48.01,
+      "learning_rate": 2.812812812812813e-05,
+      "loss": 0.5165,
+      "step": 2740
+    },
+    {
+      "epoch": 48.01,
+      "eval_accuracy": 0.5069124423963134,
+      "eval_loss": 1.3708863258361816,
+      "eval_runtime": 174.555,
+      "eval_samples_per_second": 1.243,
+      "eval_steps_per_second": 0.16,
+      "step": 2744
+    },
+    {
+      "epoch": 49.0,
+      "learning_rate": 2.8028028028028032e-05,
+      "loss": 0.4317,
+      "step": 2750
+    },
+    {
+      "epoch": 49.0,
+      "learning_rate": 2.7927927927927926e-05,
+      "loss": 0.5479,
+      "step": 2760
+    },
+    {
+      "epoch": 49.0,
+      "learning_rate": 2.782782782782783e-05,
+      "loss": 0.4649,
+      "step": 2770
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 2.7727727727727733e-05,
+      "loss": 0.5208,
+      "step": 2780
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 2.762762762762763e-05,
+      "loss": 0.5801,
+      "step": 2790
+    },
+    {
+      "epoch": 49.01,
+      "learning_rate": 2.752752752752753e-05,
+      "loss": 0.5727,
+      "step": 2800
+    },
+    {
+      "epoch": 49.01,
+      "eval_accuracy": 0.6866359447004609,
+      "eval_loss": 0.9960273504257202,
+      "eval_runtime": 169.912,
+      "eval_samples_per_second": 1.277,
+      "eval_steps_per_second": 0.165,
+      "step": 2800
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 2.7427427427427426e-05,
+      "loss": 0.5112,
+      "step": 2810
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 2.732732732732733e-05,
+      "loss": 0.37,
+      "step": 2820
+    },
+    {
+      "epoch": 50.01,
+      "learning_rate": 2.722722722722723e-05,
+      "loss": 0.6214,
+      "step": 2830
+    },
+    {
+      "epoch": 50.01,
+      "learning_rate": 2.7127127127127127e-05,
+      "loss": 0.375,
+      "step": 2840
+    },
+    {
+      "epoch": 50.01,
+      "learning_rate": 2.702702702702703e-05,
+      "loss": 0.3698,
+      "step": 2850
+    },
+    {
+      "epoch": 50.01,
+      "eval_accuracy": 0.5483870967741935,
+      "eval_loss": 1.2246321439743042,
+      "eval_runtime": 164.0861,
+      "eval_samples_per_second": 1.322,
+      "eval_steps_per_second": 0.171,
+      "step": 2856
+    },
+    {
+      "epoch": 51.0,
+      "learning_rate": 2.6926926926926927e-05,
+      "loss": 0.6204,
+      "step": 2860
+    },
+    {
+      "epoch": 51.0,
+      "learning_rate": 2.6826826826826827e-05,
+      "loss": 0.4234,
+      "step": 2870
+    },
+    {
+      "epoch": 51.0,
+      "learning_rate": 2.672672672672673e-05,
+      "loss": 0.4375,
+      "step": 2880
+    },
+    {
+      "epoch": 51.01,
+      "learning_rate": 2.6626626626626627e-05,
+      "loss": 0.3495,
+      "step": 2890
+    },
+    {
+      "epoch": 51.01,
+      "learning_rate": 2.652652652652653e-05,
+      "loss": 0.4522,
+      "step": 2900
+    },
+    {
+      "epoch": 51.01,
+      "learning_rate": 2.6426426426426428e-05,
+      "loss": 0.5836,
+      "step": 2910
+    },
+    {
+      "epoch": 51.01,
+      "eval_accuracy": 0.6866359447004609,
+      "eval_loss": 0.9892141222953796,
+      "eval_runtime": 166.0199,
+      "eval_samples_per_second": 1.307,
+      "eval_steps_per_second": 0.169,
+      "step": 2912
+    },
+    {
+      "epoch": 52.0,
+      "learning_rate": 2.6326326326326328e-05,
+      "loss": 0.4078,
+      "step": 2920
+    },
+    {
+      "epoch": 52.0,
+      "learning_rate": 2.6226226226226224e-05,
+      "loss": 0.5069,
+      "step": 2930
+    },
+    {
+      "epoch": 52.01,
+      "learning_rate": 2.6126126126126128e-05,
+      "loss": 0.4839,
+      "step": 2940
+    },
+    {
+      "epoch": 52.01,
+      "learning_rate": 2.6026026026026028e-05,
+      "loss": 0.418,
+      "step": 2950
+    },
+    {
+      "epoch": 52.01,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.6017,
+      "step": 2960
+    },
+    {
+      "epoch": 52.01,
+      "eval_accuracy": 0.6589861751152074,
+      "eval_loss": 0.9387974143028259,
+      "eval_runtime": 163.3812,
+      "eval_samples_per_second": 1.328,
+      "eval_steps_per_second": 0.171,
+      "step": 2968
+    },
+    {
+      "epoch": 53.0,
+      "learning_rate": 2.582582582582583e-05,
+      "loss": 0.5042,
+      "step": 2970
+    },
+    {
+      "epoch": 53.0,
+      "learning_rate": 2.5725725725725725e-05,
+      "loss": 0.7511,
+      "step": 2980
+    },
+    {
+      "epoch": 53.0,
+      "learning_rate": 2.5625625625625625e-05,
+      "loss": 0.5523,
+      "step": 2990
+    },
+    {
+      "epoch": 53.01,
+      "learning_rate": 2.552552552552553e-05,
+      "loss": 0.3819,
+      "step": 3000
+    },
+    {
+      "epoch": 53.01,
+      "learning_rate": 2.5425425425425426e-05,
+      "loss": 0.5585,
+      "step": 3010
+    },
+    {
+      "epoch": 53.01,
+      "learning_rate": 2.532532532532533e-05,
+      "loss": 0.4851,
+      "step": 3020
+    },
+    {
+      "epoch": 53.01,
+      "eval_accuracy": 0.6589861751152074,
+      "eval_loss": 1.1415479183197021,
+      "eval_runtime": 166.6314,
+      "eval_samples_per_second": 1.302,
+      "eval_steps_per_second": 0.168,
+      "step": 3024
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 2.5225225225225222e-05,
+      "loss": 0.564,
+      "step": 3030
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 2.5125125125125126e-05,
+      "loss": 0.4419,
+      "step": 3040
+    },
+    {
+      "epoch": 54.0,
+      "learning_rate": 2.502502502502503e-05,
+      "loss": 0.3716,
+      "step": 3050
+    },
+    {
+      "epoch": 54.01,
+      "learning_rate": 2.4924924924924926e-05,
+      "loss": 0.6022,
+      "step": 3060
+    },
+    {
+      "epoch": 54.01,
+      "learning_rate": 2.4824824824824826e-05,
+      "loss": 0.5008,
+      "step": 3070
+    },
+    {
+      "epoch": 54.01,
+      "learning_rate": 2.4724724724724727e-05,
+      "loss": 0.3038,
+      "step": 3080
+    },
+    {
+      "epoch": 54.01,
+      "eval_accuracy": 0.695852534562212,
+      "eval_loss": 0.9412721991539001,
+      "eval_runtime": 163.6724,
+      "eval_samples_per_second": 1.326,
+      "eval_steps_per_second": 0.171,
+      "step": 3080
+    },
+    {
+      "epoch": 55.0,
+      "learning_rate": 2.4624624624624627e-05,
+      "loss": 0.5041,
+      "step": 3090
+    },
+    {
+      "epoch": 55.0,
+      "learning_rate": 2.4524524524524527e-05,
+      "loss": 0.4752,
+      "step": 3100
+    },
+    {
+      "epoch": 55.01,
+      "learning_rate": 2.4424424424424424e-05,
+      "loss": 0.4335,
+      "step": 3110
+    },
+    {
+      "epoch": 55.01,
+      "learning_rate": 2.4324324324324327e-05,
+      "loss": 0.5474,
+      "step": 3120
+    },
+    {
+      "epoch": 55.01,
+      "learning_rate": 2.4224224224224227e-05,
+      "loss": 0.6075,
+      "step": 3130
+    },
+    {
+      "epoch": 55.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 1.0466532707214355,
+      "eval_runtime": 164.2606,
+      "eval_samples_per_second": 1.321,
+      "eval_steps_per_second": 0.17,
+      "step": 3136
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 2.4124124124124124e-05,
+      "loss": 0.4485,
+      "step": 3140
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 2.4024024024024024e-05,
+      "loss": 0.5252,
+      "step": 3150
+    },
+    {
+      "epoch": 56.0,
+      "learning_rate": 2.3923923923923924e-05,
+      "loss": 0.6121,
+      "step": 3160
+    },
+    {
+      "epoch": 56.01,
+      "learning_rate": 2.3823823823823824e-05,
+      "loss": 0.3327,
+      "step": 3170
+    },
+    {
+      "epoch": 56.01,
+      "learning_rate": 2.3723723723723725e-05,
+      "loss": 0.4494,
+      "step": 3180
+    },
+    {
+      "epoch": 56.01,
+      "learning_rate": 2.3623623623623625e-05,
+      "loss": 0.4474,
+      "step": 3190
+    },
+    {
+      "epoch": 56.01,
+      "eval_accuracy": 0.6866359447004609,
+      "eval_loss": 0.8436079025268555,
+      "eval_runtime": 167.7622,
+      "eval_samples_per_second": 1.293,
+      "eval_steps_per_second": 0.167,
+      "step": 3192
+    },
+    {
+      "epoch": 57.0,
+      "learning_rate": 2.3523523523523525e-05,
+      "loss": 0.3922,
+      "step": 3200
+    },
+    {
+      "epoch": 57.0,
+      "learning_rate": 2.3423423423423425e-05,
+      "loss": 0.5241,
+      "step": 3210
+    },
+    {
+      "epoch": 57.01,
+      "learning_rate": 2.3323323323323322e-05,
+      "loss": 0.3707,
+      "step": 3220
+    },
+    {
+      "epoch": 57.01,
+      "learning_rate": 2.3223223223223225e-05,
+      "loss": 0.6444,
+      "step": 3230
+    },
+    {
+      "epoch": 57.01,
+      "learning_rate": 2.3123123123123125e-05,
+      "loss": 0.3711,
+      "step": 3240
+    },
+    {
+      "epoch": 57.01,
+      "eval_accuracy": 0.6774193548387096,
+      "eval_loss": 0.8994067311286926,
+      "eval_runtime": 173.6457,
+      "eval_samples_per_second": 1.25,
+      "eval_steps_per_second": 0.161,
+      "step": 3248
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 2.3023023023023026e-05,
+      "loss": 0.4191,
+      "step": 3250
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 2.2922922922922922e-05,
+      "loss": 0.4935,
+      "step": 3260
+    },
+    {
+      "epoch": 58.0,
+      "learning_rate": 2.2822822822822822e-05,
+      "loss": 0.56,
+      "step": 3270
+    },
+    {
+      "epoch": 58.01,
+      "learning_rate": 2.2722722722722726e-05,
+      "loss": 0.283,
+      "step": 3280
+    },
+    {
+      "epoch": 58.01,
+      "learning_rate": 2.2622622622622626e-05,
+      "loss": 0.4957,
+      "step": 3290
+    },
+    {
+      "epoch": 58.01,
+      "learning_rate": 2.2522522522522523e-05,
+      "loss": 0.5279,
+      "step": 3300
+    },
+    {
+      "epoch": 58.01,
+      "eval_accuracy": 0.7188940092165899,
+      "eval_loss": 0.885903537273407,
+      "eval_runtime": 176.7148,
+      "eval_samples_per_second": 1.228,
+      "eval_steps_per_second": 0.158,
+      "step": 3304
+    },
+    {
+      "epoch": 59.0,
+      "learning_rate": 2.2422422422422423e-05,
+      "loss": 0.4729,
+      "step": 3310
+    },
+    {
+      "epoch": 59.0,
+      "learning_rate": 2.2322322322322323e-05,
+      "loss": 0.3717,
+      "step": 3320
+    },
+    {
+      "epoch": 59.0,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.538,
+      "step": 3330
+    },
+    {
+      "epoch": 59.01,
+      "learning_rate": 2.2122122122122123e-05,
+      "loss": 0.3848,
+      "step": 3340
+    },
+    {
+      "epoch": 59.01,
+      "learning_rate": 2.2022022022022024e-05,
+      "loss": 0.5991,
+      "step": 3350
+    },
+    {
+      "epoch": 59.01,
+      "learning_rate": 2.1921921921921924e-05,
+      "loss": 0.6032,
+      "step": 3360
+    },
+    {
+      "epoch": 59.01,
+      "eval_accuracy": 0.6497695852534562,
+      "eval_loss": 1.293091058731079,
+      "eval_runtime": 170.7811,
+      "eval_samples_per_second": 1.271,
+      "eval_steps_per_second": 0.164,
+      "step": 3360
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 2.1821821821821824e-05,
+      "loss": 0.419,
+      "step": 3370
+    },
+    {
+      "epoch": 60.0,
+      "learning_rate": 2.172172172172172e-05,
+      "loss": 0.3919,
+      "step": 3380
+    },
+    {
+      "epoch": 60.01,
+      "learning_rate": 2.1621621621621624e-05,
+      "loss": 0.38,
+      "step": 3390
+    },
+    {
+      "epoch": 60.01,
+      "learning_rate": 2.1521521521521524e-05,
+      "loss": 0.469,
+      "step": 3400
+    },
+    {
+      "epoch": 60.01,
+      "learning_rate": 2.142142142142142e-05,
+      "loss": 0.3282,
+      "step": 3410
+    },
+    {
+      "epoch": 60.01,
+      "eval_accuracy": 0.7142857142857143,
+      "eval_loss": 0.9435374140739441,
+      "eval_runtime": 175.2331,
+      "eval_samples_per_second": 1.238,
+      "eval_steps_per_second": 0.16,
+      "step": 3416
+    },
+    {
+      "epoch": 61.0,
+      "learning_rate": 2.132132132132132e-05,
+      "loss": 0.4183,
+      "step": 3420
+    },
+    {
+      "epoch": 61.0,
+      "learning_rate": 2.122122122122122e-05,
+      "loss": 0.5195,
+      "step": 3430
+    },
+    {
+      "epoch": 61.0,
+      "learning_rate": 2.1121121121121125e-05,
+      "loss": 0.5613,
+      "step": 3440
+    },
+    {
+      "epoch": 61.01,
+      "learning_rate": 2.102102102102102e-05,
+      "loss": 0.5026,
+      "step": 3450
+    },
+    {
+      "epoch": 61.01,
+      "learning_rate": 2.0920920920920922e-05,
+      "loss": 0.4604,
+      "step": 3460
+    },
+    {
+      "epoch": 61.01,
+      "learning_rate": 2.0820820820820822e-05,
+      "loss": 0.3506,
+      "step": 3470
+    },
+    {
+      "epoch": 61.01,
+      "eval_accuracy": 0.6728110599078341,
+      "eval_loss": 1.0970582962036133,
+      "eval_runtime": 167.732,
+      "eval_samples_per_second": 1.294,
+      "eval_steps_per_second": 0.167,
+      "step": 3472
+    },
+    {
+      "epoch": 62.0,
+      "learning_rate": 2.0720720720720722e-05,
+      "loss": 0.3312,
+      "step": 3480
+    },
+    {
+      "epoch": 62.0,
+      "learning_rate": 2.062062062062062e-05,
+      "loss": 0.2965,
+      "step": 3490
+    },
+    {
+      "epoch": 62.01,
+      "learning_rate": 2.0520520520520522e-05,
+      "loss": 0.396,
+      "step": 3500
+    },
+    {
+      "epoch": 62.01,
+      "learning_rate": 2.0420420420420422e-05,
+      "loss": 0.5294,
+      "step": 3510
+    },
+    {
+      "epoch": 62.01,
+      "learning_rate": 2.0320320320320323e-05,
+      "loss": 0.3169,
+      "step": 3520
+    },
+    {
+      "epoch": 62.01,
+      "eval_accuracy": 0.7511520737327189,
+      "eval_loss": 0.910149335861206,
+      "eval_runtime": 165.7118,
+      "eval_samples_per_second": 1.31,
+      "eval_steps_per_second": 0.169,
+      "step": 3528
+    },
+    {
+      "epoch": 63.0,
+      "learning_rate": 2.022022022022022e-05,
+      "loss": 0.5156,
+      "step": 3530
+    },
+    {
+      "epoch": 63.0,
+      "learning_rate": 2.012012012012012e-05,
+      "loss": 0.4815,
+      "step": 3540
+    },
+    {
+      "epoch": 63.0,
+      "learning_rate": 2.0020020020020023e-05,
+      "loss": 0.3874,
+      "step": 3550
+    },
+    {
+      "epoch": 63.01,
+      "learning_rate": 1.9919919919919923e-05,
+      "loss": 0.5157,
+      "step": 3560
+    },
+    {
+      "epoch": 63.01,
+      "learning_rate": 1.981981981981982e-05,
+      "loss": 0.5023,
+      "step": 3570
+    },
+    {
+      "epoch": 63.01,
+      "learning_rate": 1.971971971971972e-05,
+      "loss": 0.438,
+      "step": 3580
+    },
+    {
+      "epoch": 63.01,
+      "eval_accuracy": 0.6359447004608295,
+      "eval_loss": 1.4072048664093018,
+      "eval_runtime": 163.2516,
+      "eval_samples_per_second": 1.329,
+      "eval_steps_per_second": 0.172,
+      "step": 3584
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 1.961961961961962e-05,
+      "loss": 0.4035,
+      "step": 3590
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 1.951951951951952e-05,
+      "loss": 0.2573,
+      "step": 3600
+    },
+    {
+      "epoch": 64.0,
+      "learning_rate": 1.941941941941942e-05,
+      "loss": 0.3321,
+      "step": 3610
+    },
+    {
+      "epoch": 64.01,
+      "learning_rate": 1.931931931931932e-05,
+      "loss": 0.6128,
+      "step": 3620
+    },
+    {
+      "epoch": 64.01,
+      "learning_rate": 1.921921921921922e-05,
+      "loss": 0.5059,
+      "step": 3630
+    },
+    {
+      "epoch": 64.01,
+      "learning_rate": 1.911911911911912e-05,
+      "loss": 0.5208,
+      "step": 3640
+    },
+    {
+      "epoch": 64.01,
+      "eval_accuracy": 0.6543778801843319,
+      "eval_loss": 1.2648274898529053,
+      "eval_runtime": 163.0954,
+      "eval_samples_per_second": 1.331,
+      "eval_steps_per_second": 0.172,
+      "step": 3640
+    },
+    {
+      "epoch": 65.0,
+      "learning_rate": 1.9019019019019018e-05,
+      "loss": 0.3918,
+      "step": 3650
+    },
+    {
+      "epoch": 65.0,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.407,
+      "step": 3660
+    },
+    {
+      "epoch": 65.01,
+      "learning_rate": 1.881881881881882e-05,
+      "loss": 0.3251,
+      "step": 3670
+    },
+    {
+      "epoch": 65.01,
+      "learning_rate": 1.8718718718718718e-05,
+      "loss": 0.4584,
+      "step": 3680
+    },
+    {
+      "epoch": 65.01,
+      "learning_rate": 1.8618618618618618e-05,
+      "loss": 0.4563,
+      "step": 3690
+    },
+    {
+      "epoch": 65.01,
+      "eval_accuracy": 0.6497695852534562,
+      "eval_loss": 1.1162357330322266,
+      "eval_runtime": 163.7463,
+      "eval_samples_per_second": 1.325,
+      "eval_steps_per_second": 0.171,
+      "step": 3696
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.2968,
+      "step": 3700
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 1.8418418418418422e-05,
+      "loss": 0.6683,
+      "step": 3710
+    },
+    {
+      "epoch": 66.0,
+      "learning_rate": 1.831831831831832e-05,
+      "loss": 0.4281,
+      "step": 3720
+    },
+    {
+      "epoch": 66.01,
+      "learning_rate": 1.821821821821822e-05,
+      "loss": 0.2642,
+      "step": 3730
+    },
+    {
+      "epoch": 66.01,
+      "learning_rate": 1.811811811811812e-05,
+      "loss": 0.4064,
+      "step": 3740
+    },
+    {
+      "epoch": 66.01,
+      "learning_rate": 1.801801801801802e-05,
+      "loss": 0.6693,
+      "step": 3750
+    },
+    {
+      "epoch": 66.01,
+      "eval_accuracy": 0.5576036866359447,
+      "eval_loss": 1.8557851314544678,
+      "eval_runtime": 162.9605,
+      "eval_samples_per_second": 1.332,
+      "eval_steps_per_second": 0.172,
+      "step": 3752
+    },
+    {
+      "epoch": 67.0,
+      "learning_rate": 1.7917917917917916e-05,
+      "loss": 0.3636,
+      "step": 3760
+    },
+    {
+      "epoch": 67.0,
+      "learning_rate": 1.781781781781782e-05,
+      "loss": 0.3325,
+      "step": 3770
+    },
+    {
+      "epoch": 67.01,
+      "learning_rate": 1.771771771771772e-05,
+      "loss": 0.5104,
+      "step": 3780
+    },
+    {
+      "epoch": 67.01,
+      "learning_rate": 1.761761761761762e-05,
+      "loss": 0.3475,
+      "step": 3790
+    },
+    {
+      "epoch": 67.01,
+      "learning_rate": 1.7517517517517516e-05,
+      "loss": 0.5599,
+      "step": 3800
+    },
+    {
+      "epoch": 67.01,
+      "eval_accuracy": 0.5391705069124424,
+      "eval_loss": 1.6573548316955566,
+      "eval_runtime": 163.7298,
+      "eval_samples_per_second": 1.325,
+      "eval_steps_per_second": 0.171,
+      "step": 3808
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 1.7417417417417416e-05,
+      "loss": 0.385,
+      "step": 3810
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 1.731731731731732e-05,
+      "loss": 0.3942,
+      "step": 3820
+    },
+    {
+      "epoch": 68.0,
+      "learning_rate": 1.721721721721722e-05,
+      "loss": 0.4411,
+      "step": 3830
+    },
+    {
+      "epoch": 68.01,
+      "learning_rate": 1.7117117117117117e-05,
+      "loss": 0.5076,
+      "step": 3840
+    },
+    {
+      "epoch": 68.01,
+      "learning_rate": 1.7017017017017017e-05,
+      "loss": 0.49,
+      "step": 3850
+    },
+    {
+      "epoch": 68.01,
+      "learning_rate": 1.6916916916916917e-05,
+      "loss": 0.4751,
+      "step": 3860
+    },
+    {
+      "epoch": 68.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 1.188300609588623,
+      "eval_runtime": 164.9656,
+      "eval_samples_per_second": 1.315,
+      "eval_steps_per_second": 0.17,
+      "step": 3864
+    },
+    {
+      "epoch": 69.0,
+      "learning_rate": 1.6816816816816817e-05,
+      "loss": 0.4068,
+      "step": 3870
+    },
+    {
+      "epoch": 69.0,
+      "learning_rate": 1.6716716716716717e-05,
+      "loss": 0.3716,
+      "step": 3880
+    },
+    {
+      "epoch": 69.0,
+      "learning_rate": 1.6616616616616618e-05,
+      "loss": 0.5685,
+      "step": 3890
+    },
+    {
+      "epoch": 69.01,
+      "learning_rate": 1.6516516516516518e-05,
+      "loss": 0.4182,
+      "step": 3900
+    },
+    {
+      "epoch": 69.01,
+      "learning_rate": 1.6416416416416418e-05,
+      "loss": 0.542,
+      "step": 3910
+    },
+    {
+      "epoch": 69.01,
+      "learning_rate": 1.6316316316316315e-05,
+      "loss": 0.6489,
+      "step": 3920
+    },
+    {
+      "epoch": 69.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 1.2733248472213745,
+      "eval_runtime": 172.0622,
+      "eval_samples_per_second": 1.261,
+      "eval_steps_per_second": 0.163,
+      "step": 3920
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.3482,
+      "step": 3930
+    },
+    {
+      "epoch": 70.0,
+      "learning_rate": 1.6116116116116118e-05,
+      "loss": 0.4185,
+      "step": 3940
+    },
+    {
+      "epoch": 70.01,
+      "learning_rate": 1.6016016016016015e-05,
+      "loss": 0.3264,
+      "step": 3950
+    },
+    {
+      "epoch": 70.01,
+      "learning_rate": 1.5915915915915915e-05,
+      "loss": 0.3664,
+      "step": 3960
+    },
+    {
+      "epoch": 70.01,
+      "learning_rate": 1.5815815815815815e-05,
+      "loss": 0.4229,
+      "step": 3970
+    },
+    {
+      "epoch": 70.01,
+      "eval_accuracy": 0.6682027649769585,
+      "eval_loss": 1.0993900299072266,
+      "eval_runtime": 169.5904,
+      "eval_samples_per_second": 1.28,
+      "eval_steps_per_second": 0.165,
+      "step": 3976
+    },
+    {
+      "epoch": 71.0,
+      "learning_rate": 1.571571571571572e-05,
+      "loss": 0.2035,
+      "step": 3980
+    },
+    {
+      "epoch": 71.0,
+      "learning_rate": 1.5615615615615616e-05,
+      "loss": 0.4886,
+      "step": 3990
+    },
+    {
+      "epoch": 71.0,
+      "learning_rate": 1.5515515515515516e-05,
+      "loss": 0.4031,
+      "step": 4000
+    },
+    {
+      "epoch": 71.01,
+      "learning_rate": 1.5415415415415416e-05,
+      "loss": 0.3078,
+      "step": 4010
+    },
+    {
+      "epoch": 71.01,
+      "learning_rate": 1.5315315315315316e-05,
+      "loss": 0.5506,
+      "step": 4020
+    },
+    {
+      "epoch": 71.01,
+      "learning_rate": 1.5215215215215218e-05,
+      "loss": 0.4194,
+      "step": 4030
+    },
+    {
+      "epoch": 71.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.1464142799377441,
+      "eval_runtime": 176.0399,
+      "eval_samples_per_second": 1.233,
+      "eval_steps_per_second": 0.159,
+      "step": 4032
+    },
+    {
+      "epoch": 72.0,
+      "learning_rate": 1.5115115115115116e-05,
+      "loss": 0.3864,
+      "step": 4040
+    },
+    {
+      "epoch": 72.0,
+      "learning_rate": 1.5015015015015016e-05,
+      "loss": 0.5229,
+      "step": 4050
+    },
+    {
+      "epoch": 72.01,
+      "learning_rate": 1.4914914914914915e-05,
+      "loss": 0.4735,
+      "step": 4060
+    },
+    {
+      "epoch": 72.01,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.2941,
+      "step": 4070
+    },
+    {
+      "epoch": 72.01,
+      "learning_rate": 1.4714714714714713e-05,
+      "loss": 0.2121,
+      "step": 4080
+    },
+    {
+      "epoch": 72.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.179811716079712,
+      "eval_runtime": 174.9502,
+      "eval_samples_per_second": 1.24,
+      "eval_steps_per_second": 0.16,
+      "step": 4088
+    },
+    {
+      "epoch": 73.0,
+      "learning_rate": 1.4614614614614617e-05,
+      "loss": 0.3443,
+      "step": 4090
+    },
+    {
+      "epoch": 73.0,
+      "learning_rate": 1.4514514514514515e-05,
+      "loss": 0.4008,
+      "step": 4100
+    },
+    {
+      "epoch": 73.0,
+      "learning_rate": 1.4414414414414416e-05,
+      "loss": 0.4189,
+      "step": 4110
+    },
+    {
+      "epoch": 73.01,
+      "learning_rate": 1.4314314314314314e-05,
+      "loss": 0.5514,
+      "step": 4120
+    },
+    {
+      "epoch": 73.01,
+      "learning_rate": 1.4214214214214214e-05,
+      "loss": 0.3123,
+      "step": 4130
+    },
+    {
+      "epoch": 73.01,
+      "learning_rate": 1.4114114114114116e-05,
+      "loss": 0.4106,
+      "step": 4140
+    },
+    {
+      "epoch": 73.01,
+      "eval_accuracy": 0.5806451612903226,
+      "eval_loss": 1.329359531402588,
+      "eval_runtime": 169.6509,
+      "eval_samples_per_second": 1.279,
+      "eval_steps_per_second": 0.165,
+      "step": 4144
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 1.4014014014014016e-05,
+      "loss": 0.2142,
+      "step": 4150
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 1.3913913913913915e-05,
+      "loss": 0.3547,
+      "step": 4160
+    },
+    {
+      "epoch": 74.0,
+      "learning_rate": 1.3813813813813815e-05,
+      "loss": 0.2571,
+      "step": 4170
+    },
+    {
+      "epoch": 74.01,
+      "learning_rate": 1.3713713713713713e-05,
+      "loss": 0.4924,
+      "step": 4180
+    },
+    {
+      "epoch": 74.01,
+      "learning_rate": 1.3613613613613615e-05,
+      "loss": 0.2901,
+      "step": 4190
+    },
+    {
+      "epoch": 74.01,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.3962,
+      "step": 4200
+    },
+    {
+      "epoch": 74.01,
+      "eval_accuracy": 0.6359447004608295,
+      "eval_loss": 1.4209370613098145,
+      "eval_runtime": 169.3625,
+      "eval_samples_per_second": 1.281,
+      "eval_steps_per_second": 0.165,
+      "step": 4200
+    },
+    {
+      "epoch": 75.0,
+      "learning_rate": 1.3413413413413414e-05,
+      "loss": 0.3685,
+      "step": 4210
+    },
+    {
+      "epoch": 75.0,
+      "learning_rate": 1.3313313313313314e-05,
+      "loss": 0.2953,
+      "step": 4220
+    },
+    {
+      "epoch": 75.01,
+      "learning_rate": 1.3213213213213214e-05,
+      "loss": 0.3889,
+      "step": 4230
+    },
+    {
+      "epoch": 75.01,
+      "learning_rate": 1.3113113113113112e-05,
+      "loss": 0.4059,
+      "step": 4240
+    },
+    {
+      "epoch": 75.01,
+      "learning_rate": 1.3013013013013014e-05,
+      "loss": 0.2963,
+      "step": 4250
+    },
+    {
+      "epoch": 75.01,
+      "eval_accuracy": 0.5944700460829493,
+      "eval_loss": 1.5015925168991089,
+      "eval_runtime": 162.4547,
+      "eval_samples_per_second": 1.336,
+      "eval_steps_per_second": 0.172,
+      "step": 4256
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 1.2912912912912914e-05,
+      "loss": 0.2218,
+      "step": 4260
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 1.2812812812812813e-05,
+      "loss": 0.4315,
+      "step": 4270
+    },
+    {
+      "epoch": 76.0,
+      "learning_rate": 1.2712712712712713e-05,
+      "loss": 0.4317,
+      "step": 4280
+    },
+    {
+      "epoch": 76.01,
+      "learning_rate": 1.2612612612612611e-05,
+      "loss": 0.4733,
+      "step": 4290
+    },
+    {
+      "epoch": 76.01,
+      "learning_rate": 1.2512512512512515e-05,
+      "loss": 0.4233,
+      "step": 4300
+    },
+    {
+      "epoch": 76.01,
+      "learning_rate": 1.2412412412412413e-05,
+      "loss": 0.5436,
+      "step": 4310
+    },
+    {
+      "epoch": 76.01,
+      "eval_accuracy": 0.5483870967741935,
+      "eval_loss": 1.5647104978561401,
+      "eval_runtime": 164.1354,
+      "eval_samples_per_second": 1.322,
+      "eval_steps_per_second": 0.171,
+      "step": 4312
+    },
+    {
+      "epoch": 77.0,
+      "learning_rate": 1.2312312312312313e-05,
+      "loss": 0.3,
+      "step": 4320
+    },
+    {
+      "epoch": 77.0,
+      "learning_rate": 1.2212212212212212e-05,
+      "loss": 0.3433,
+      "step": 4330
+    },
+    {
+      "epoch": 77.01,
+      "learning_rate": 1.2112112112112114e-05,
+      "loss": 0.2487,
+      "step": 4340
+    },
+    {
+      "epoch": 77.01,
+      "learning_rate": 1.2012012012012012e-05,
+      "loss": 0.4711,
+      "step": 4350
+    },
+    {
+      "epoch": 77.01,
+      "learning_rate": 1.1911911911911912e-05,
+      "loss": 0.4115,
+      "step": 4360
+    },
+    {
+      "epoch": 77.01,
+      "eval_accuracy": 0.6036866359447005,
+      "eval_loss": 1.4308525323867798,
+      "eval_runtime": 160.3268,
+      "eval_samples_per_second": 1.353,
+      "eval_steps_per_second": 0.175,
+      "step": 4368
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 1.1811811811811812e-05,
+      "loss": 0.278,
+      "step": 4370
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 1.1711711711711713e-05,
+      "loss": 0.3355,
+      "step": 4380
+    },
+    {
+      "epoch": 78.0,
+      "learning_rate": 1.1611611611611613e-05,
+      "loss": 0.3502,
+      "step": 4390
+    },
+    {
+      "epoch": 78.01,
+      "learning_rate": 1.1511511511511513e-05,
+      "loss": 0.3967,
+      "step": 4400
+    },
+    {
+      "epoch": 78.01,
+      "learning_rate": 1.1411411411411411e-05,
+      "loss": 0.3376,
+      "step": 4410
+    },
+    {
+      "epoch": 78.01,
+      "learning_rate": 1.1311311311311313e-05,
+      "loss": 0.1635,
+      "step": 4420
+    },
+    {
+      "epoch": 78.01,
+      "eval_accuracy": 0.6451612903225806,
+      "eval_loss": 1.3660060167312622,
+      "eval_runtime": 163.3833,
+      "eval_samples_per_second": 1.328,
+      "eval_steps_per_second": 0.171,
+      "step": 4424
+    },
+    {
+      "epoch": 79.0,
+      "learning_rate": 1.1211211211211212e-05,
+      "loss": 0.3182,
+      "step": 4430
+    },
+    {
+      "epoch": 79.0,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.3935,
+      "step": 4440
+    },
+    {
+      "epoch": 79.0,
+      "learning_rate": 1.1011011011011012e-05,
+      "loss": 0.3312,
+      "step": 4450
+    },
+    {
+      "epoch": 79.01,
+      "learning_rate": 1.0910910910910912e-05,
+      "loss": 0.4817,
+      "step": 4460
+    },
+    {
+      "epoch": 79.01,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.3807,
+      "step": 4470
+    },
+    {
+      "epoch": 79.01,
+      "learning_rate": 1.071071071071071e-05,
+      "loss": 0.2931,
+      "step": 4480
+    },
+    {
+      "epoch": 79.01,
+      "eval_accuracy": 0.6497695852534562,
+      "eval_loss": 1.3298723697662354,
+      "eval_runtime": 163.0047,
+      "eval_samples_per_second": 1.331,
+      "eval_steps_per_second": 0.172,
+      "step": 4480
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 1.061061061061061e-05,
+      "loss": 0.2781,
+      "step": 4490
+    },
+    {
+      "epoch": 80.0,
+      "learning_rate": 1.051051051051051e-05,
+      "loss": 0.258,
+      "step": 4500
+    },
+    {
+      "epoch": 80.01,
+      "learning_rate": 1.0410410410410411e-05,
+      "loss": 0.4798,
+      "step": 4510
+    },
+    {
+      "epoch": 80.01,
+      "learning_rate": 1.031031031031031e-05,
+      "loss": 0.4413,
+      "step": 4520
+    },
+    {
+      "epoch": 80.01,
+      "learning_rate": 1.0210210210210211e-05,
+      "loss": 0.5154,
+      "step": 4530
+    },
+    {
+      "epoch": 80.01,
+      "eval_accuracy": 0.5806451612903226,
+      "eval_loss": 1.6550365686416626,
+      "eval_runtime": 166.6749,
+      "eval_samples_per_second": 1.302,
+      "eval_steps_per_second": 0.168,
+      "step": 4536
+    },
+    {
+      "epoch": 81.0,
+      "learning_rate": 1.011011011011011e-05,
+      "loss": 0.5175,
+      "step": 4540
+    },
+    {
+      "epoch": 81.0,
+      "learning_rate": 1.0010010010010011e-05,
+      "loss": 0.3942,
+      "step": 4550
+    },
+    {
+      "epoch": 81.0,
+      "learning_rate": 9.90990990990991e-06,
+      "loss": 0.2373,
+      "step": 4560
+    },
+    {
+      "epoch": 81.01,
+      "learning_rate": 9.80980980980981e-06,
+      "loss": 0.3044,
+      "step": 4570
+    },
+    {
+      "epoch": 81.01,
+      "learning_rate": 9.70970970970971e-06,
+      "loss": 0.229,
+      "step": 4580
+    },
+    {
+      "epoch": 81.01,
+      "learning_rate": 9.60960960960961e-06,
+      "loss": 0.2993,
+      "step": 4590
+    },
+    {
+      "epoch": 81.01,
+      "eval_accuracy": 0.5990783410138248,
+      "eval_loss": 1.6520466804504395,
+      "eval_runtime": 169.6637,
+      "eval_samples_per_second": 1.279,
+      "eval_steps_per_second": 0.165,
+      "step": 4592
+    },
+    {
+      "epoch": 82.0,
+      "learning_rate": 9.509509509509509e-06,
+      "loss": 0.4553,
+      "step": 4600
+    },
+    {
+      "epoch": 82.0,
+      "learning_rate": 9.40940940940941e-06,
+      "loss": 0.338,
+      "step": 4610
+    },
+    {
+      "epoch": 82.01,
+      "learning_rate": 9.309309309309309e-06,
+      "loss": 0.2887,
+      "step": 4620
+    },
+    {
+      "epoch": 82.01,
+      "learning_rate": 9.209209209209211e-06,
+      "loss": 0.2972,
+      "step": 4630
+    },
+    {
+      "epoch": 82.01,
+      "learning_rate": 9.10910910910911e-06,
+      "loss": 0.4391,
+      "step": 4640
+    },
+    {
+      "epoch": 82.01,
+      "eval_accuracy": 0.6405529953917051,
+      "eval_loss": 1.3823057413101196,
+      "eval_runtime": 174.0025,
+      "eval_samples_per_second": 1.247,
+      "eval_steps_per_second": 0.161,
+      "step": 4648
+    },
+    {
+      "epoch": 83.0,
+      "learning_rate": 9.00900900900901e-06,
+      "loss": 0.5297,
+      "step": 4650
+    },
+    {
+      "epoch": 83.0,
+      "learning_rate": 8.90890890890891e-06,
+      "loss": 0.4287,
+      "step": 4660
+    },
+    {
+      "epoch": 83.0,
+      "learning_rate": 8.80880880880881e-06,
+      "loss": 0.3487,
+      "step": 4670
+    },
+    {
+      "epoch": 83.01,
+      "learning_rate": 8.708708708708708e-06,
+      "loss": 0.3783,
+      "step": 4680
+    },
+    {
+      "epoch": 83.01,
+      "learning_rate": 8.60860860860861e-06,
+      "loss": 0.4584,
+      "step": 4690
+    },
+    {
+      "epoch": 83.01,
+      "learning_rate": 8.508508508508508e-06,
+      "loss": 0.485,
+      "step": 4700
+    },
+    {
+      "epoch": 83.01,
+      "eval_accuracy": 0.6036866359447005,
+      "eval_loss": 1.4859918355941772,
+      "eval_runtime": 169.7438,
+      "eval_samples_per_second": 1.278,
+      "eval_steps_per_second": 0.165,
+      "step": 4704
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 8.408408408408409e-06,
+      "loss": 0.3618,
+      "step": 4710
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 8.308308308308309e-06,
+      "loss": 0.3709,
+      "step": 4720
+    },
+    {
+      "epoch": 84.0,
+      "learning_rate": 8.208208208208209e-06,
+      "loss": 0.3175,
+      "step": 4730
+    },
+    {
+      "epoch": 84.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 0.305,
+      "step": 4740
+    },
+    {
+      "epoch": 84.01,
+      "learning_rate": 8.008008008008007e-06,
+      "loss": 0.4123,
+      "step": 4750
+    },
+    {
+      "epoch": 84.01,
+      "learning_rate": 7.907907907907908e-06,
+      "loss": 0.3313,
+      "step": 4760
+    },
+    {
+      "epoch": 84.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.3875089883804321,
+      "eval_runtime": 173.3931,
+      "eval_samples_per_second": 1.251,
+      "eval_steps_per_second": 0.161,
+      "step": 4760
+    },
+    {
+      "epoch": 85.0,
+      "learning_rate": 7.807807807807808e-06,
+      "loss": 0.3653,
+      "step": 4770
+    },
+    {
+      "epoch": 85.0,
+      "learning_rate": 7.707707707707708e-06,
+      "loss": 0.3763,
+      "step": 4780
+    },
+    {
+      "epoch": 85.01,
+      "learning_rate": 7.607607607607609e-06,
+      "loss": 0.5182,
+      "step": 4790
+    },
+    {
+      "epoch": 85.01,
+      "learning_rate": 7.507507507507508e-06,
+      "loss": 0.552,
+      "step": 4800
+    },
+    {
+      "epoch": 85.01,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.4194,
+      "step": 4810
+    },
+    {
+      "epoch": 85.01,
+      "eval_accuracy": 0.5898617511520737,
+      "eval_loss": 1.4334131479263306,
+      "eval_runtime": 172.5224,
+      "eval_samples_per_second": 1.258,
+      "eval_steps_per_second": 0.162,
+      "step": 4816
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 7.3073073073073085e-06,
+      "loss": 0.3314,
+      "step": 4820
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 7.207207207207208e-06,
+      "loss": 0.288,
+      "step": 4830
+    },
+    {
+      "epoch": 86.0,
+      "learning_rate": 7.107107107107107e-06,
+      "loss": 0.3408,
+      "step": 4840
+    },
+    {
+      "epoch": 86.01,
+      "learning_rate": 7.007007007007008e-06,
+      "loss": 0.3747,
+      "step": 4850
+    },
+    {
+      "epoch": 86.01,
+      "learning_rate": 6.906906906906907e-06,
+      "loss": 0.2595,
+      "step": 4860
+    },
+    {
+      "epoch": 86.01,
+      "learning_rate": 6.8068068068068075e-06,
+      "loss": 0.4515,
+      "step": 4870
+    },
+    {
+      "epoch": 86.01,
+      "eval_accuracy": 0.5990783410138248,
+      "eval_loss": 1.6488864421844482,
+      "eval_runtime": 168.4269,
+      "eval_samples_per_second": 1.288,
+      "eval_steps_per_second": 0.166,
+      "step": 4872
+    },
+    {
+      "epoch": 87.0,
+      "learning_rate": 6.706706706706707e-06,
+      "loss": 0.4239,
+      "step": 4880
+    },
+    {
+      "epoch": 87.0,
+      "learning_rate": 6.606606606606607e-06,
+      "loss": 0.2477,
+      "step": 4890
+    },
+    {
+      "epoch": 87.01,
+      "learning_rate": 6.506506506506507e-06,
+      "loss": 0.3937,
+      "step": 4900
+    },
+    {
+      "epoch": 87.01,
+      "learning_rate": 6.406406406406406e-06,
+      "loss": 0.4013,
+      "step": 4910
+    },
+    {
+      "epoch": 87.01,
+      "learning_rate": 6.306306306306306e-06,
+      "loss": 0.3283,
+      "step": 4920
+    },
+    {
+      "epoch": 87.01,
+      "eval_accuracy": 0.6082949308755761,
+      "eval_loss": 1.4548600912094116,
+      "eval_runtime": 165.7438,
+      "eval_samples_per_second": 1.309,
+      "eval_steps_per_second": 0.169,
+      "step": 4928
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 6.206206206206207e-06,
+      "loss": 0.3104,
+      "step": 4930
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 6.106106106106106e-06,
+      "loss": 0.3391,
+      "step": 4940
+    },
+    {
+      "epoch": 88.0,
+      "learning_rate": 6.006006006006006e-06,
+      "loss": 0.3999,
+      "step": 4950
+    },
+    {
+      "epoch": 88.01,
+      "learning_rate": 5.905905905905906e-06,
+      "loss": 0.3572,
+      "step": 4960
+    },
+    {
+      "epoch": 88.01,
+      "learning_rate": 5.805805805805806e-06,
+      "loss": 0.4948,
+      "step": 4970
+    },
+    {
+      "epoch": 88.01,
+      "learning_rate": 5.705705705705706e-06,
+      "loss": 0.1914,
+      "step": 4980
+    },
+    {
+      "epoch": 88.01,
+      "eval_accuracy": 0.6267281105990783,
+      "eval_loss": 1.3415180444717407,
+      "eval_runtime": 160.989,
+      "eval_samples_per_second": 1.348,
+      "eval_steps_per_second": 0.174,
+      "step": 4984
+    },
+    {
+      "epoch": 89.0,
+      "learning_rate": 5.605605605605606e-06,
+      "loss": 0.5032,
+      "step": 4990
+    },
+    {
+      "epoch": 89.0,
+      "learning_rate": 5.505505505505506e-06,
+      "loss": 0.2588,
+      "step": 5000
+    },
+    {
+      "epoch": 89.0,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 0.3718,
+      "step": 5010
+    },
+    {
+      "epoch": 89.01,
+      "learning_rate": 5.305305305305305e-06,
+      "loss": 0.2464,
+      "step": 5020
+    },
+    {
+      "epoch": 89.01,
+      "learning_rate": 5.2052052052052055e-06,
+      "loss": 0.1357,
+      "step": 5030
+    },
+    {
+      "epoch": 89.01,
+      "learning_rate": 5.105105105105106e-06,
+      "loss": 0.2142,
+      "step": 5040
+    },
+    {
+      "epoch": 89.01,
+      "eval_accuracy": 0.6267281105990783,
+      "eval_loss": 1.642616629600525,
+      "eval_runtime": 164.3339,
+      "eval_samples_per_second": 1.32,
+      "eval_steps_per_second": 0.17,
+      "step": 5040
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 5.005005005005006e-06,
+      "loss": 0.3905,
+      "step": 5050
+    },
+    {
+      "epoch": 90.0,
+      "learning_rate": 4.904904904904905e-06,
+      "loss": 0.2482,
+      "step": 5060
+    },
+    {
+      "epoch": 90.01,
+      "learning_rate": 4.804804804804805e-06,
+      "loss": 0.3922,
+      "step": 5070
+    },
+    {
+      "epoch": 90.01,
+      "learning_rate": 4.704704704704705e-06,
+      "loss": 0.2728,
+      "step": 5080
+    },
+    {
+      "epoch": 90.01,
+      "learning_rate": 4.6046046046046055e-06,
+      "loss": 0.3121,
+      "step": 5090
+    },
+    {
+      "epoch": 90.01,
+      "eval_accuracy": 0.6036866359447005,
+      "eval_loss": 1.699904441833496,
+      "eval_runtime": 160.7204,
+      "eval_samples_per_second": 1.35,
+      "eval_steps_per_second": 0.174,
+      "step": 5096
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 4.504504504504505e-06,
+      "loss": 0.4005,
+      "step": 5100
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 4.404404404404405e-06,
+      "loss": 0.3077,
+      "step": 5110
+    },
+    {
+      "epoch": 91.0,
+      "learning_rate": 4.304304304304305e-06,
+      "loss": 0.3005,
+      "step": 5120
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.204204204204204e-06,
+      "loss": 0.357,
+      "step": 5130
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.1041041041041045e-06,
+      "loss": 0.2725,
+      "step": 5140
+    },
+    {
+      "epoch": 91.01,
+      "learning_rate": 4.004004004004004e-06,
+      "loss": 0.367,
+      "step": 5150
+    },
+    {
+      "epoch": 91.01,
+      "eval_accuracy": 0.6082949308755761,
+      "eval_loss": 1.4683284759521484,
+      "eval_runtime": 162.3117,
+      "eval_samples_per_second": 1.337,
+      "eval_steps_per_second": 0.173,
+      "step": 5152
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 3.903903903903904e-06,
+      "loss": 0.2292,
+      "step": 5160
+    },
+    {
+      "epoch": 92.0,
+      "learning_rate": 3.8038038038038044e-06,
+      "loss": 0.4217,
+      "step": 5170
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.473,
+      "step": 5180
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 3.603603603603604e-06,
+      "loss": 0.2874,
+      "step": 5190
+    },
+    {
+      "epoch": 92.01,
+      "learning_rate": 3.503503503503504e-06,
+      "loss": 0.178,
+      "step": 5200
+    },
+    {
+      "epoch": 92.01,
+      "eval_accuracy": 0.6267281105990783,
+      "eval_loss": 1.4664562940597534,
+      "eval_runtime": 166.2787,
+      "eval_samples_per_second": 1.305,
+      "eval_steps_per_second": 0.168,
+      "step": 5208
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.4034034034034037e-06,
+      "loss": 0.2857,
+      "step": 5210
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.3033033033033035e-06,
+      "loss": 0.4529,
+      "step": 5220
+    },
+    {
+      "epoch": 93.0,
+      "learning_rate": 3.203203203203203e-06,
+      "loss": 0.3539,
+      "step": 5230
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.1031031031031033e-06,
+      "loss": 0.508,
+      "step": 5240
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 3.003003003003003e-06,
+      "loss": 0.3495,
+      "step": 5250
+    },
+    {
+      "epoch": 93.01,
+      "learning_rate": 2.902902902902903e-06,
+      "loss": 0.3972,
+      "step": 5260
+    },
+    {
+      "epoch": 93.01,
+      "eval_accuracy": 0.6451612903225806,
+      "eval_loss": 1.3464295864105225,
+      "eval_runtime": 177.3754,
+      "eval_samples_per_second": 1.223,
+      "eval_steps_per_second": 0.158,
+      "step": 5264
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 2.802802802802803e-06,
+      "loss": 0.239,
+      "step": 5270
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 0.3102,
+      "step": 5280
+    },
+    {
+      "epoch": 94.0,
+      "learning_rate": 2.6026026026026027e-06,
+      "loss": 0.2617,
+      "step": 5290
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 2.502502502502503e-06,
+      "loss": 0.2582,
+      "step": 5300
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 2.4024024024024026e-06,
+      "loss": 0.3483,
+      "step": 5310
+    },
+    {
+      "epoch": 94.01,
+      "learning_rate": 2.3023023023023027e-06,
+      "loss": 0.224,
+      "step": 5320
+    },
+    {
+      "epoch": 94.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.5009006261825562,
+      "eval_runtime": 173.0007,
+      "eval_samples_per_second": 1.254,
+      "eval_steps_per_second": 0.162,
+      "step": 5320
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.2022022022022024e-06,
+      "loss": 0.2503,
+      "step": 5330
+    },
+    {
+      "epoch": 95.0,
+      "learning_rate": 2.102102102102102e-06,
+      "loss": 0.4842,
+      "step": 5340
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 2.002002002002002e-06,
+      "loss": 0.2556,
+      "step": 5350
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 1.9019019019019022e-06,
+      "loss": 0.3167,
+      "step": 5360
+    },
+    {
+      "epoch": 95.01,
+      "learning_rate": 1.801801801801802e-06,
+      "loss": 0.1848,
+      "step": 5370
+    },
+    {
+      "epoch": 95.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 1.5068354606628418,
+      "eval_runtime": 170.4309,
+      "eval_samples_per_second": 1.273,
+      "eval_steps_per_second": 0.164,
+      "step": 5376
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 1.7017017017017019e-06,
+      "loss": 0.3195,
+      "step": 5380
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 1.6016016016016016e-06,
+      "loss": 0.3012,
+      "step": 5390
+    },
+    {
+      "epoch": 96.0,
+      "learning_rate": 1.5015015015015015e-06,
+      "loss": 0.3382,
+      "step": 5400
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 1.4014014014014014e-06,
+      "loss": 0.232,
+      "step": 5410
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 1.3013013013013014e-06,
+      "loss": 0.29,
+      "step": 5420
+    },
+    {
+      "epoch": 96.01,
+      "learning_rate": 1.2012012012012013e-06,
+      "loss": 0.2776,
+      "step": 5430
+    },
+    {
+      "epoch": 96.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.538283348083496,
+      "eval_runtime": 170.4441,
+      "eval_samples_per_second": 1.273,
+      "eval_steps_per_second": 0.164,
+      "step": 5432
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.1011011011011012e-06,
+      "loss": 0.31,
+      "step": 5440
+    },
+    {
+      "epoch": 97.0,
+      "learning_rate": 1.001001001001001e-06,
+      "loss": 0.3648,
+      "step": 5450
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 9.00900900900901e-07,
+      "loss": 0.3484,
+      "step": 5460
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 8.008008008008008e-07,
+      "loss": 0.2007,
+      "step": 5470
+    },
+    {
+      "epoch": 97.01,
+      "learning_rate": 7.007007007007007e-07,
+      "loss": 0.3506,
+      "step": 5480
+    },
+    {
+      "epoch": 97.01,
+      "eval_accuracy": 0.6129032258064516,
+      "eval_loss": 1.5355850458145142,
+      "eval_runtime": 174.6844,
+      "eval_samples_per_second": 1.242,
+      "eval_steps_per_second": 0.16,
+      "step": 5488
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 6.006006006006006e-07,
+      "loss": 0.4143,
+      "step": 5490
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 5.005005005005005e-07,
+      "loss": 0.3519,
+      "step": 5500
+    },
+    {
+      "epoch": 98.0,
+      "learning_rate": 4.004004004004004e-07,
+      "loss": 0.2268,
+      "step": 5510
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 3.003003003003003e-07,
+      "loss": 0.1346,
+      "step": 5520
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 2.002002002002002e-07,
+      "loss": 0.2987,
+      "step": 5530
+    },
+    {
+      "epoch": 98.01,
+      "learning_rate": 1.001001001001001e-07,
+      "loss": 0.401,
+      "step": 5540
+    },
+    {
+      "epoch": 98.01,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.5504214763641357,
+      "eval_runtime": 168.626,
+      "eval_samples_per_second": 1.287,
+      "eval_steps_per_second": 0.166,
+      "step": 5544
+    },
+    {
+      "epoch": 99.0,
+      "learning_rate": 0.0,
+      "loss": 0.3466,
+      "step": 5550
+    },
+    {
+      "epoch": 99.0,
+      "eval_accuracy": 0.6175115207373272,
+      "eval_loss": 1.5504581928253174,
+      "eval_runtime": 169.0024,
+      "eval_samples_per_second": 1.284,
+      "eval_steps_per_second": 0.166,
+      "step": 5550
+    },
+    {
+      "epoch": 99.0,
+      "step": 5550,
+      "total_flos": 5.520338427328414e+19,
+      "train_loss": 0.6062898898339486,
+      "train_runtime": 58981.2354,
+      "train_samples_per_second": 0.753,
+      "train_steps_per_second": 0.094
+    },
+    {
+      "epoch": 99.0,
+      "eval_accuracy": 0.7685185185185185,
+      "eval_loss": 0.7077057361602783,
+      "eval_runtime": 174.9386,
+      "eval_samples_per_second": 1.235,
+      "eval_steps_per_second": 0.154,
+      "step": 5550
+    },
+    {
+      "epoch": 99.0,
+      "eval_accuracy": 0.7685185185185185,
+      "eval_loss": 0.7077056765556335,
+      "eval_runtime": 166.0464,
+      "eval_samples_per_second": 1.301,
+      "eval_steps_per_second": 0.163,
+      "step": 5550
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5550,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 500,
+  "total_flos": 5.520338427328414e+19,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}