End of training

all_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 30.026442307692307,
-    "eval_accuracy": 0.35570469798657717,
-    "eval_loss": 2.520230293273926,
-    "eval_runtime": 112.9248,
-    "eval_samples_per_second": 3.958,
-    "eval_steps_per_second": 0.124
+    "epoch": 59.00721153846154,
+    "eval_accuracy": 0.7319819819819819,
+    "eval_loss": 1.084416389465332,
+    "eval_runtime": 75.3126,
+    "eval_samples_per_second": 5.895,
+    "eval_steps_per_second": 0.093
 }

runs/Jan14_16-01-40_GAN-SVR/events.out.tfevents.1736851415.GAN-SVR.3870842.8

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7a5dfbf870d294a569baa230c78ce95feed24ab80b22e9a78ba14e68699ae25d
-size 411
+oid sha256:3aa347f8819808b6ed0a069ac2bd27d55402bf8858e7c9bbf607cecfab1d1288
+size 734

test_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 30.026442307692307,
-    "eval_accuracy": 0.35570469798657717,
-    "eval_loss": 2.520230293273926,
-    "eval_runtime": 112.9248,
-    "eval_samples_per_second": 3.958,
-    "eval_steps_per_second": 0.124
+    "epoch": 59.00721153846154,
+    "eval_accuracy": 0.7319819819819819,
+    "eval_loss": 1.084416389465332,
+    "eval_runtime": 75.3126,
+    "eval_samples_per_second": 5.895,
+    "eval_steps_per_second": 0.093
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.35570469798657717,
-  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-783",
-  "epoch": 30.026442307692307,
+  "best_metric": 0.7319819819819819,
+  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-812",
+  "epoch": 59.00721153846154,
   "eval_steps": 500,
   "global_step": 832,
   "is_hyper_param_search": false,
@@ -10,889 +10,1150 @@
   "log_history": [
     {
       "epoch": 0.01201923076923077,
-      "grad_norm": 9.181777954101562,
+      "grad_norm": 4.912038326263428,
       "learning_rate": 5.9523809523809525e-06,
-      "loss": 4.2532,
+      "loss": 4.238,
       "step": 10
     },
     {
-      "epoch": 0.02403846153846154,
-      "grad_norm": 6.192462921142578,
+      "epoch": 0.016826923076923076,
+      "eval_accuracy": 0.018018018018018018,
+      "eval_loss": 4.223487377166748,
+      "eval_runtime": 70.4378,
+      "eval_samples_per_second": 6.303,
+      "eval_steps_per_second": 0.099,
+      "step": 14
+    },
+    {
+      "epoch": 1.0072115384615385,
+      "grad_norm": 5.0958333015441895,
       "learning_rate": 1.1904761904761905e-05,
-      "loss": 4.2583,
+      "loss": 4.2658,
       "step": 20
     },
     {
-      "epoch": 0.03245192307692308,
-      "eval_accuracy": 0.015659955257270694,
-      "eval_loss": 4.225129127502441,
-      "eval_runtime": 113.6438,
-      "eval_samples_per_second": 3.933,
-      "eval_steps_per_second": 0.123,
-      "step": 27
+      "epoch": 1.0168269230769231,
+      "eval_accuracy": 0.02027027027027027,
+      "eval_loss": 4.207489490509033,
+      "eval_runtime": 72.9674,
+      "eval_samples_per_second": 6.085,
+      "eval_steps_per_second": 0.096,
+      "step": 28
     },
     {
-      "epoch": 1.0036057692307692,
-      "grad_norm": 6.222796440124512,
+      "epoch": 2.0024038461538463,
+      "grad_norm": 3.8229589462280273,
       "learning_rate": 1.785714285714286e-05,
-      "loss": 4.2505,
+      "loss": 4.2236,
       "step": 30
     },
     {
-      "epoch": 1.015625,
-      "grad_norm": 3.965486526489258,
+      "epoch": 2.014423076923077,
+      "grad_norm": 3.7527875900268555,
       "learning_rate": 2.380952380952381e-05,
-      "loss": 4.2489,
+      "loss": 4.219,
       "step": 40
     },
     {
-      "epoch": 1.0276442307692308,
-      "grad_norm": 5.0857744216918945,
+      "epoch": 2.016826923076923,
+      "eval_accuracy": 0.015765765765765764,
+      "eval_loss": 4.205322265625,
+      "eval_runtime": 70.576,
+      "eval_samples_per_second": 6.291,
+      "eval_steps_per_second": 0.099,
+      "step": 42
+    },
+    {
+      "epoch": 3.0096153846153846,
+      "grad_norm": 2.5568084716796875,
       "learning_rate": 2.9761904761904762e-05,
-      "loss": 4.2374,
+      "loss": 4.2146,
       "step": 50
     },
     {
-      "epoch": 1.0324519230769231,
-      "eval_accuracy": 0.013422818791946308,
-      "eval_loss": 4.226736068725586,
-      "eval_runtime": 113.905,
-      "eval_samples_per_second": 3.924,
-      "eval_steps_per_second": 0.123,
-      "step": 54
+      "epoch": 3.016826923076923,
+      "eval_accuracy": 0.02252252252252252,
+      "eval_loss": 4.209786415100098,
+      "eval_runtime": 73.3716,
+      "eval_samples_per_second": 6.051,
+      "eval_steps_per_second": 0.095,
+      "step": 56
     },
     {
-      "epoch": 2.0072115384615383,
-      "grad_norm": 3.5929408073425293,
+      "epoch": 4.0048076923076925,
+      "grad_norm": 2.623307943344116,
       "learning_rate": 3.571428571428572e-05,
-      "loss": 4.2403,
+      "loss": 4.2083,
       "step": 60
     },
     {
-      "epoch": 2.019230769230769,
-      "grad_norm": 3.6417300701141357,
+      "epoch": 4.016826923076923,
+      "grad_norm": 11.633626937866211,
       "learning_rate": 4.166666666666667e-05,
-      "loss": 4.255,
+      "loss": 4.1925,
+      "step": 70
+    },
+    {
+      "epoch": 4.016826923076923,
+      "eval_accuracy": 0.02252252252252252,
+      "eval_loss": 4.182406902313232,
+      "eval_runtime": 73.5871,
+      "eval_samples_per_second": 6.034,
+      "eval_steps_per_second": 0.095,
       "step": 70
     },
     {
-      "epoch": 2.03125,
-      "grad_norm": 3.2748305797576904,
+      "epoch": 5.012019230769231,
+      "grad_norm": 4.2888689041137695,
       "learning_rate": 4.761904761904762e-05,
-      "loss": 4.2678,
+      "loss": 4.1192,
       "step": 80
     },
     {
-      "epoch": 2.032451923076923,
-      "eval_accuracy": 0.013422818791946308,
-      "eval_loss": 4.2263360023498535,
-      "eval_runtime": 115.125,
-      "eval_samples_per_second": 3.883,
-      "eval_steps_per_second": 0.122,
-      "step": 81
+      "epoch": 5.016826923076923,
+      "eval_accuracy": 0.038288288288288286,
+      "eval_loss": 4.081945896148682,
+      "eval_runtime": 77.2884,
+      "eval_samples_per_second": 5.745,
+      "eval_steps_per_second": 0.091,
+      "step": 84
     },
     {
-      "epoch": 3.0108173076923075,
-      "grad_norm": 3.1406173706054688,
+      "epoch": 6.007211538461538,
+      "grad_norm": 3.5548884868621826,
       "learning_rate": 4.959893048128342e-05,
-      "loss": 4.2151,
+      "loss": 4.0297,
       "step": 90
     },
     {
-      "epoch": 3.0228365384615383,
-      "grad_norm": 2.8441436290740967,
-      "learning_rate": 4.8930481283422465e-05,
-      "loss": 4.2537,
-      "step": 100
+      "epoch": 6.016826923076923,
+      "eval_accuracy": 0.06981981981981981,
+      "eval_loss": 3.981917142868042,
+      "eval_runtime": 72.62,
+      "eval_samples_per_second": 6.114,
+      "eval_steps_per_second": 0.096,
+      "step": 98
     },
     {
-      "epoch": 3.032451923076923,
-      "eval_accuracy": 0.015659955257270694,
-      "eval_loss": 4.221240997314453,
-      "eval_runtime": 115.863,
-      "eval_samples_per_second": 3.858,
-      "eval_steps_per_second": 0.121,
-      "step": 108
+      "epoch": 7.002403846153846,
+      "grad_norm": 6.507851600646973,
+      "learning_rate": 4.8930481283422465e-05,
+      "loss": 3.9499,
+      "step": 100
     },
     {
-      "epoch": 4.002403846153846,
-      "grad_norm": 2.7364041805267334,
+      "epoch": 7.014423076923077,
+      "grad_norm": 5.1739959716796875,
       "learning_rate": 4.8262032085561496e-05,
-      "loss": 4.2568,
+      "loss": 3.7134,
       "step": 110
     },
     {
-      "epoch": 4.014423076923077,
-      "grad_norm": 2.71181058883667,
+      "epoch": 7.016826923076923,
+      "eval_accuracy": 0.1036036036036036,
+      "eval_loss": 3.7339625358581543,
+      "eval_runtime": 73.0624,
+      "eval_samples_per_second": 6.077,
+      "eval_steps_per_second": 0.096,
+      "step": 112
+    },
+    {
+      "epoch": 8.009615384615385,
+      "grad_norm": 4.7554779052734375,
       "learning_rate": 4.759358288770054e-05,
-      "loss": 4.2288,
+      "loss": 3.5289,
       "step": 120
     },
     {
-      "epoch": 4.0264423076923075,
-      "grad_norm": 2.7206692695617676,
+      "epoch": 8.016826923076923,
+      "eval_accuracy": 0.18018018018018017,
+      "eval_loss": 3.420488119125366,
+      "eval_runtime": 73.5384,
+      "eval_samples_per_second": 6.038,
+      "eval_steps_per_second": 0.095,
+      "step": 126
+    },
+    {
+      "epoch": 9.004807692307692,
+      "grad_norm": 6.714512348175049,
       "learning_rate": 4.6925133689839576e-05,
-      "loss": 4.2401,
+      "loss": 3.3709,
       "step": 130
     },
     {
-      "epoch": 4.032451923076923,
-      "eval_accuracy": 0.015659955257270694,
-      "eval_loss": 4.206515312194824,
-      "eval_runtime": 114.2862,
-      "eval_samples_per_second": 3.911,
-      "eval_steps_per_second": 0.122,
-      "step": 135
+      "epoch": 9.016826923076923,
+      "grad_norm": 18.599552154541016,
+      "learning_rate": 4.625668449197861e-05,
+      "loss": 3.0625,
+      "step": 140
     },
     {
-      "epoch": 5.006009615384615,
-      "grad_norm": 2.6117472648620605,
-      "learning_rate": 4.625668449197861e-05,
-      "loss": 4.2192,
+      "epoch": 9.016826923076923,
+      "eval_accuracy": 0.26126126126126126,
+      "eval_loss": 3.196960687637329,
+      "eval_runtime": 74.7592,
+      "eval_samples_per_second": 5.939,
+      "eval_steps_per_second": 0.094,
       "step": 140
     },
     {
-      "epoch": 5.018028846153846,
-      "grad_norm": 2.90580677986145,
+      "epoch": 10.01201923076923,
+      "grad_norm": 7.006440162658691,
       "learning_rate": 4.558823529411765e-05,
-      "loss": 4.2204,
+      "loss": 2.8776,
       "step": 150
     },
     {
-      "epoch": 5.030048076923077,
-      "grad_norm": 2.9660141468048096,
+      "epoch": 10.016826923076923,
+      "eval_accuracy": 0.30180180180180183,
+      "eval_loss": 2.9350106716156006,
+      "eval_runtime": 75.4948,
+      "eval_samples_per_second": 5.881,
+      "eval_steps_per_second": 0.093,
+      "step": 154
+    },
+    {
+      "epoch": 11.007211538461538,
+      "grad_norm": 6.219061374664307,
       "learning_rate": 4.491978609625669e-05,
-      "loss": 4.2519,
+      "loss": 2.6375,
       "step": 160
     },
     {
-      "epoch": 5.032451923076923,
-      "eval_accuracy": 0.017897091722595078,
-      "eval_loss": 4.207516193389893,
-      "eval_runtime": 117.7024,
-      "eval_samples_per_second": 3.798,
-      "eval_steps_per_second": 0.119,
-      "step": 162
+      "epoch": 11.016826923076923,
+      "eval_accuracy": 0.36036036036036034,
+      "eval_loss": 2.762986660003662,
+      "eval_runtime": 73.4162,
+      "eval_samples_per_second": 6.048,
+      "eval_steps_per_second": 0.095,
+      "step": 168
     },
     {
-      "epoch": 6.009615384615385,
-      "grad_norm": 2.8973159790039062,
+      "epoch": 12.002403846153847,
+      "grad_norm": 6.574148178100586,
       "learning_rate": 4.4251336898395724e-05,
-      "loss": 4.2112,
+      "loss": 2.4071,
       "step": 170
     },
     {
-      "epoch": 6.021634615384615,
-      "grad_norm": 2.734694480895996,
+      "epoch": 12.014423076923077,
+      "grad_norm": 8.191123008728027,
       "learning_rate": 4.358288770053476e-05,
-      "loss": 4.2198,
+      "loss": 2.2954,
       "step": 180
     },
     {
-      "epoch": 6.032451923076923,
-      "eval_accuracy": 0.013422818791946308,
-      "eval_loss": 4.205463409423828,
-      "eval_runtime": 116.7575,
-      "eval_samples_per_second": 3.828,
-      "eval_steps_per_second": 0.12,
-      "step": 189
+      "epoch": 12.016826923076923,
+      "eval_accuracy": 0.42792792792792794,
+      "eval_loss": 2.4990594387054443,
+      "eval_runtime": 80.1492,
+      "eval_samples_per_second": 5.54,
+      "eval_steps_per_second": 0.087,
+      "step": 182
     },
     {
-      "epoch": 7.001201923076923,
-      "grad_norm": 2.421506881713867,
+      "epoch": 13.009615384615385,
+      "grad_norm": 6.4414825439453125,
       "learning_rate": 4.29144385026738e-05,
-      "loss": 4.2292,
+      "loss": 2.1337,
       "step": 190
     },
     {
-      "epoch": 7.013221153846154,
-      "grad_norm": 2.82832670211792,
+      "epoch": 13.016826923076923,
+      "eval_accuracy": 0.43243243243243246,
+      "eval_loss": 2.3827102184295654,
+      "eval_runtime": 73.7035,
+      "eval_samples_per_second": 6.024,
+      "eval_steps_per_second": 0.095,
+      "step": 196
+    },
+    {
+      "epoch": 14.004807692307692,
+      "grad_norm": 8.87990951538086,
       "learning_rate": 4.224598930481284e-05,
-      "loss": 4.2019,
+      "loss": 1.9319,
       "step": 200
     },
     {
-      "epoch": 7.025240384615385,
-      "grad_norm": 3.0022928714752197,
+      "epoch": 14.016826923076923,
+      "grad_norm": 22.130002975463867,
       "learning_rate": 4.157754010695187e-05,
-      "loss": 4.2111,
+      "loss": 1.8195,
       "step": 210
     },
     {
-      "epoch": 7.032451923076923,
-      "eval_accuracy": 0.017897091722595078,
-      "eval_loss": 4.19577693939209,
-      "eval_runtime": 119.1446,
-      "eval_samples_per_second": 3.752,
-      "eval_steps_per_second": 0.118,
-      "step": 216
+      "epoch": 14.016826923076923,
+      "eval_accuracy": 0.481981981981982,
+      "eval_loss": 2.2311551570892334,
+      "eval_runtime": 75.5729,
+      "eval_samples_per_second": 5.875,
+      "eval_steps_per_second": 0.093,
+      "step": 210
     },
     {
-      "epoch": 8.004807692307692,
-      "grad_norm": 2.6406443119049072,
+      "epoch": 15.01201923076923,
+      "grad_norm": 7.134392261505127,
       "learning_rate": 4.0909090909090915e-05,
-      "loss": 4.2068,
+      "loss": 1.6436,
       "step": 220
     },
     {
-      "epoch": 8.016826923076923,
-      "grad_norm": 3.169625759124756,
+      "epoch": 15.016826923076923,
+      "eval_accuracy": 0.44594594594594594,
+      "eval_loss": 2.3628787994384766,
+      "eval_runtime": 76.2152,
+      "eval_samples_per_second": 5.826,
+      "eval_steps_per_second": 0.092,
+      "step": 224
+    },
+    {
+      "epoch": 16.00721153846154,
+      "grad_norm": 11.737651824951172,
       "learning_rate": 4.024064171122995e-05,
-      "loss": 4.1759,
+      "loss": 1.6289,
       "step": 230
     },
     {
-      "epoch": 8.028846153846153,
-      "grad_norm": 2.8210370540618896,
-      "learning_rate": 3.957219251336899e-05,
-      "loss": 4.1871,
-      "step": 240
+      "epoch": 16.016826923076923,
+      "eval_accuracy": 0.4752252252252252,
+      "eval_loss": 2.151165723800659,
+      "eval_runtime": 72.7861,
+      "eval_samples_per_second": 6.1,
+      "eval_steps_per_second": 0.096,
+      "step": 238
     },
     {
-      "epoch": 8.032451923076923,
-      "eval_accuracy": 0.04697986577181208,
-      "eval_loss": 4.147375583648682,
-      "eval_runtime": 117.849,
-      "eval_samples_per_second": 3.793,
-      "eval_steps_per_second": 0.119,
-      "step": 243
+      "epoch": 17.002403846153847,
+      "grad_norm": 12.988323211669922,
+      "learning_rate": 3.957219251336899e-05,
+      "loss": 1.423,
+      "step": 240
     },
     {
-      "epoch": 9.008413461538462,
-      "grad_norm": 3.140075206756592,
+      "epoch": 17.014423076923077,
+      "grad_norm": 5.394784450531006,
       "learning_rate": 3.8903743315508025e-05,
-      "loss": 4.1314,
+      "loss": 1.2957,
       "step": 250
     },
     {
-      "epoch": 9.020432692307692,
-      "grad_norm": 3.0806028842926025,
+      "epoch": 17.016826923076923,
+      "eval_accuracy": 0.5022522522522522,
+      "eval_loss": 2.0142312049865723,
+      "eval_runtime": 74.6687,
+      "eval_samples_per_second": 5.946,
+      "eval_steps_per_second": 0.094,
+      "step": 252
+    },
+    {
+      "epoch": 18.009615384615383,
+      "grad_norm": 6.641128063201904,
       "learning_rate": 3.8235294117647055e-05,
-      "loss": 4.0778,
+      "loss": 1.2761,
       "step": 260
     },
     {
-      "epoch": 9.032451923076923,
-      "grad_norm": 12.136305809020996,
-      "learning_rate": 3.75668449197861e-05,
-      "loss": 4.0891,
-      "step": 270
+      "epoch": 18.016826923076923,
+      "eval_accuracy": 0.527027027027027,
+      "eval_loss": 1.906096339225769,
+      "eval_runtime": 75.2853,
+      "eval_samples_per_second": 5.898,
+      "eval_steps_per_second": 0.093,
+      "step": 266
     },
     {
-      "epoch": 9.032451923076923,
-      "eval_accuracy": 0.0447427293064877,
-      "eval_loss": 4.032660007476807,
-      "eval_runtime": 116.5729,
-      "eval_samples_per_second": 3.835,
-      "eval_steps_per_second": 0.12,
+      "epoch": 19.004807692307693,
+      "grad_norm": 7.859009742736816,
+      "learning_rate": 3.75668449197861e-05,
+      "loss": 1.217,
       "step": 270
     },
     {
-      "epoch": 10.01201923076923,
-      "grad_norm": 4.812855243682861,
+      "epoch": 19.016826923076923,
+      "grad_norm": 29.06666374206543,
       "learning_rate": 3.6898395721925136e-05,
-      "loss": 3.9545,
+      "loss": 1.1118,
       "step": 280
     },
     {
-      "epoch": 10.024038461538462,
-      "grad_norm": 5.218036651611328,
+      "epoch": 19.016826923076923,
+      "eval_accuracy": 0.5495495495495496,
+      "eval_loss": 1.811281681060791,
+      "eval_runtime": 81.5227,
+      "eval_samples_per_second": 5.446,
+      "eval_steps_per_second": 0.086,
+      "step": 280
+    },
+    {
+      "epoch": 20.01201923076923,
+      "grad_norm": 8.949642181396484,
       "learning_rate": 3.622994652406417e-05,
-      "loss": 3.7963,
+      "loss": 0.9642,
       "step": 290
     },
     {
-      "epoch": 10.032451923076923,
-      "eval_accuracy": 0.08277404921700224,
-      "eval_loss": 3.82175874710083,
-      "eval_runtime": 115.9952,
-      "eval_samples_per_second": 3.854,
-      "eval_steps_per_second": 0.121,
-      "step": 297
+      "epoch": 20.016826923076923,
+      "eval_accuracy": 0.6036036036036037,
+      "eval_loss": 1.727989673614502,
+      "eval_runtime": 72.1469,
+      "eval_samples_per_second": 6.154,
+      "eval_steps_per_second": 0.097,
+      "step": 294
     },
     {
-      "epoch": 11.00360576923077,
-      "grad_norm": 4.765311241149902,
+      "epoch": 21.00721153846154,
+      "grad_norm": 6.73243522644043,
       "learning_rate": 3.556149732620321e-05,
-      "loss": 3.7306,
+      "loss": 0.894,
       "step": 300
     },
     {
-      "epoch": 11.015625,
-      "grad_norm": 6.956701755523682,
+      "epoch": 21.016826923076923,
+      "eval_accuracy": 0.5382882882882883,
+      "eval_loss": 1.8722944259643555,
+      "eval_runtime": 76.1286,
+      "eval_samples_per_second": 5.832,
+      "eval_steps_per_second": 0.092,
+      "step": 308
+    },
+    {
+      "epoch": 22.002403846153847,
+      "grad_norm": 8.856271743774414,
       "learning_rate": 3.489304812834225e-05,
-      "loss": 3.5563,
+      "loss": 0.8454,
       "step": 310
     },
     {
-      "epoch": 11.02764423076923,
-      "grad_norm": 5.394344329833984,
+      "epoch": 22.014423076923077,
+      "grad_norm": 6.268283843994141,
       "learning_rate": 3.4224598930481284e-05,
-      "loss": 3.4787,
+      "loss": 0.7974,
       "step": 320
     },
     {
-      "epoch": 11.032451923076923,
-      "eval_accuracy": 0.11185682326621924,
-      "eval_loss": 3.7061688899993896,
-      "eval_runtime": 113.6071,
-      "eval_samples_per_second": 3.935,
-      "eval_steps_per_second": 0.123,
-      "step": 324
+      "epoch": 22.016826923076923,
+      "eval_accuracy": 0.6058558558558559,
+      "eval_loss": 1.658478021621704,
+      "eval_runtime": 69.8498,
+      "eval_samples_per_second": 6.356,
+      "eval_steps_per_second": 0.1,
+      "step": 322
     },
     {
-      "epoch": 12.007211538461538,
-      "grad_norm": 5.70036506652832,
+      "epoch": 23.009615384615383,
+      "grad_norm": 6.027498722076416,
       "learning_rate": 3.355614973262032e-05,
-      "loss": 3.4821,
+      "loss": 0.833,
       "step": 330
     },
     {
-      "epoch": 12.01923076923077,
-      "grad_norm": 6.370093822479248,
+      "epoch": 23.016826923076923,
+      "eval_accuracy": 0.6148648648648649,
+      "eval_loss": 1.599608302116394,
+      "eval_runtime": 76.7983,
+      "eval_samples_per_second": 5.781,
+      "eval_steps_per_second": 0.091,
+      "step": 336
+    },
+    {
+      "epoch": 24.004807692307693,
+      "grad_norm": 5.669070243835449,
       "learning_rate": 3.288770053475936e-05,
-      "loss": 3.3756,
+      "loss": 0.6668,
       "step": 340
     },
     {
-      "epoch": 12.03125,
-      "grad_norm": 5.8564133644104,
+      "epoch": 24.016826923076923,
+      "grad_norm": 12.669683456420898,
       "learning_rate": 3.22192513368984e-05,
-      "loss": 3.1883,
+      "loss": 0.6431,
       "step": 350
     },
     {
-      "epoch": 12.032451923076923,
-      "eval_accuracy": 0.11185682326621924,
-      "eval_loss": 3.58866810798645,
-      "eval_runtime": 112.5771,
-      "eval_samples_per_second": 3.971,
-      "eval_steps_per_second": 0.124,
-      "step": 351
+      "epoch": 24.016826923076923,
+      "eval_accuracy": 0.6148648648648649,
+      "eval_loss": 1.560927152633667,
+      "eval_runtime": 69.5019,
+      "eval_samples_per_second": 6.388,
+      "eval_steps_per_second": 0.101,
+      "step": 350
     },
     {
-      "epoch": 13.010817307692308,
-      "grad_norm": 6.884145736694336,
+      "epoch": 25.01201923076923,
+      "grad_norm": 5.4236626625061035,
       "learning_rate": 3.155080213903743e-05,
-      "loss": 3.1542,
+      "loss": 0.5873,
       "step": 360
     },
     {
-      "epoch": 13.022836538461538,
-      "grad_norm": 7.7231059074401855,
+      "epoch": 25.016826923076923,
+      "eval_accuracy": 0.6171171171171171,
+      "eval_loss": 1.6108297109603882,
+      "eval_runtime": 73.8756,
+      "eval_samples_per_second": 6.01,
+      "eval_steps_per_second": 0.095,
+      "step": 364
+    },
+    {
+      "epoch": 26.00721153846154,
+      "grad_norm": 5.605001449584961,
       "learning_rate": 3.0882352941176475e-05,
-      "loss": 3.0045,
+      "loss": 0.5554,
       "step": 370
     },
     {
-      "epoch": 13.032451923076923,
-      "eval_accuracy": 0.14317673378076062,
-      "eval_loss": 3.380981206893921,
-      "eval_runtime": 113.3509,
-      "eval_samples_per_second": 3.944,
-      "eval_steps_per_second": 0.124,
+      "epoch": 26.016826923076923,
+      "eval_accuracy": 0.6531531531531531,
+      "eval_loss": 1.4013562202453613,
+      "eval_runtime": 79.7998,
+      "eval_samples_per_second": 5.564,
+      "eval_steps_per_second": 0.088,
       "step": 378
     },
     {
-      "epoch": 14.002403846153847,
-      "grad_norm": 6.160039901733398,
+      "epoch": 27.002403846153847,
+      "grad_norm": 6.0180277824401855,
       "learning_rate": 3.0213903743315508e-05,
-      "loss": 2.9704,
+      "loss": 0.5428,
       "step": 380
     },
     {
-      "epoch": 14.014423076923077,
-      "grad_norm": 8.257513046264648,
+      "epoch": 27.014423076923077,
+      "grad_norm": 7.198999404907227,
       "learning_rate": 2.954545454545455e-05,
-      "loss": 2.7723,
+      "loss": 0.4786,
       "step": 390
     },
     {
-      "epoch": 14.026442307692308,
-      "grad_norm": 6.512996196746826,
+      "epoch": 27.016826923076923,
+      "eval_accuracy": 0.6621621621621622,
+      "eval_loss": 1.433493733406067,
+      "eval_runtime": 71.2505,
+      "eval_samples_per_second": 6.232,
+      "eval_steps_per_second": 0.098,
+      "step": 392
+    },
+    {
+      "epoch": 28.009615384615383,
+      "grad_norm": 4.281993389129639,
       "learning_rate": 2.8877005347593582e-05,
-      "loss": 2.8045,
+      "loss": 0.4252,
       "step": 400
     },
     {
-      "epoch": 14.032451923076923,
-      "eval_accuracy": 0.21923937360178972,
-      "eval_loss": 3.2211973667144775,
-      "eval_runtime": 116.514,
-      "eval_samples_per_second": 3.836,
-      "eval_steps_per_second": 0.12,
-      "step": 405
+      "epoch": 28.016826923076923,
+      "eval_accuracy": 0.6509009009009009,
+      "eval_loss": 1.444474220275879,
+      "eval_runtime": 71.5031,
+      "eval_samples_per_second": 6.21,
+      "eval_steps_per_second": 0.098,
+      "step": 406
     },
     {
-      "epoch": 15.006009615384615,
-      "grad_norm": 6.627754211425781,
+      "epoch": 29.004807692307693,
+      "grad_norm": 6.2630534172058105,
       "learning_rate": 2.8208556149732622e-05,
-      "loss": 2.7247,
+      "loss": 0.4527,
       "step": 410
     },
     {
-      "epoch": 15.018028846153847,
-      "grad_norm": 6.555057048797607,
+      "epoch": 29.016826923076923,
+      "grad_norm": 4.163498878479004,
       "learning_rate": 2.754010695187166e-05,
-      "loss": 2.5789,
+      "loss": 0.382,
+      "step": 420
+    },
+    {
+      "epoch": 29.016826923076923,
+      "eval_accuracy": 0.6621621621621622,
+      "eval_loss": 1.3914759159088135,
+      "eval_runtime": 72.0609,
+      "eval_samples_per_second": 6.161,
+      "eval_steps_per_second": 0.097,
       "step": 420
     },
     {
-      "epoch": 15.030048076923077,
-      "grad_norm": 8.675689697265625,
+      "epoch": 30.01201923076923,
+      "grad_norm": 5.012676239013672,
       "learning_rate": 2.68716577540107e-05,
-      "loss": 2.5344,
+      "loss": 0.365,
       "step": 430
     },
     {
-      "epoch": 15.032451923076923,
-      "eval_accuracy": 0.15883668903803133,
-      "eval_loss": 3.270237922668457,
-      "eval_runtime": 114.1519,
-      "eval_samples_per_second": 3.916,
-      "eval_steps_per_second": 0.123,
-      "step": 432
+      "epoch": 30.016826923076923,
+      "eval_accuracy": 0.6846846846846847,
+      "eval_loss": 1.297812819480896,
+      "eval_runtime": 73.674,
+      "eval_samples_per_second": 6.027,
+      "eval_steps_per_second": 0.095,
+      "step": 434
     },
     {
-      "epoch": 16.009615384615383,
-      "grad_norm": 7.248364448547363,
+      "epoch": 31.00721153846154,
+      "grad_norm": 4.618253707885742,
       "learning_rate": 2.6203208556149733e-05,
-      "loss": 2.3791,
+      "loss": 0.319,
       "step": 440
     },
     {
-      "epoch": 16.021634615384617,
-      "grad_norm": 9.409223556518555,
-      "learning_rate": 2.5534759358288773e-05,
-      "loss": 2.3725,
-      "step": 450
+      "epoch": 31.016826923076923,
+      "eval_accuracy": 0.6824324324324325,
+      "eval_loss": 1.3218427896499634,
+      "eval_runtime": 74.4745,
+      "eval_samples_per_second": 5.962,
+      "eval_steps_per_second": 0.094,
+      "step": 448
     },
     {
-      "epoch": 16.032451923076923,
-      "eval_accuracy": 0.14093959731543623,
-      "eval_loss": 3.35996413230896,
-      "eval_runtime": 113.8728,
-      "eval_samples_per_second": 3.925,
-      "eval_steps_per_second": 0.123,
-      "step": 459
+      "epoch": 32.00240384615385,
+      "grad_norm": 6.823300361633301,
+      "learning_rate": 2.5534759358288773e-05,
+      "loss": 0.3829,
+      "step": 450
     },
     {
-      "epoch": 17.001201923076923,
-      "grad_norm": 7.834836006164551,
+      "epoch": 32.01442307692308,
+      "grad_norm": 5.096348285675049,
       "learning_rate": 2.4866310160427807e-05,
-      "loss": 2.2722,
+      "loss": 0.3167,
       "step": 460
     },
     {
-      "epoch": 17.013221153846153,
-      "grad_norm": 7.643532752990723,
+      "epoch": 32.01682692307692,
+      "eval_accuracy": 0.6644144144144144,
+      "eval_loss": 1.3495796918869019,
+      "eval_runtime": 73.0271,
+      "eval_samples_per_second": 6.08,
+      "eval_steps_per_second": 0.096,
+      "step": 462
+    },
+    {
+      "epoch": 33.00961538461539,
+      "grad_norm": 6.042089462280273,
       "learning_rate": 2.4197860962566847e-05,
-      "loss": 2.1434,
+      "loss": 0.2797,
       "step": 470
     },
     {
-      "epoch": 17.025240384615383,
-      "grad_norm": 8.425023078918457,
+      "epoch": 33.01682692307692,
+      "eval_accuracy": 0.6801801801801802,
+      "eval_loss": 1.2806001901626587,
+      "eval_runtime": 72.9489,
+      "eval_samples_per_second": 6.086,
+      "eval_steps_per_second": 0.096,
+      "step": 476
+    },
+    {
+      "epoch": 34.00480769230769,
+      "grad_norm": 4.125849723815918,
       "learning_rate": 2.3529411764705884e-05,
-      "loss": 2.2074,
+      "loss": 0.281,
       "step": 480
     },
     {
-      "epoch": 17.032451923076923,
-      "eval_accuracy": 0.2371364653243848,
-      "eval_loss": 2.9730582237243652,
-      "eval_runtime": 113.5853,
-      "eval_samples_per_second": 3.935,
-      "eval_steps_per_second": 0.123,
-      "step": 486
+      "epoch": 34.01682692307692,
+      "grad_norm": 7.1154704093933105,
+      "learning_rate": 2.286096256684492e-05,
+      "loss": 0.2864,
+      "step": 490
     },
     {
-      "epoch": 18.004807692307693,
-      "grad_norm": 7.840888500213623,
-      "learning_rate": 2.286096256684492e-05,
-      "loss": 2.0823,
+      "epoch": 34.01682692307692,
+      "eval_accuracy": 0.7072072072072072,
+      "eval_loss": 1.219117283821106,
+      "eval_runtime": 74.6962,
+      "eval_samples_per_second": 5.944,
+      "eval_steps_per_second": 0.094,
       "step": 490
     },
     {
-      "epoch": 18.016826923076923,
-      "grad_norm": 9.385211944580078,
+      "epoch": 35.01201923076923,
+      "grad_norm": 6.536057949066162,
       "learning_rate": 2.2192513368983957e-05,
-      "loss": 1.9511,
+      "loss": 0.2927,
       "step": 500
     },
     {
-      "epoch": 18.028846153846153,
-      "grad_norm": 8.72028636932373,
+      "epoch": 35.01682692307692,
+      "eval_accuracy": 0.7207207207207207,
+      "eval_loss": 1.2134795188903809,
+      "eval_runtime": 72.1404,
+      "eval_samples_per_second": 6.155,
+      "eval_steps_per_second": 0.097,
+      "step": 504
+    },
+    {
+      "epoch": 36.00721153846154,
+      "grad_norm": 4.703156471252441,
       "learning_rate": 2.1524064171122994e-05,
-      "loss": 2.1094,
+      "loss": 0.2698,
       "step": 510
     },
     {
-      "epoch": 18.032451923076923,
-      "eval_accuracy": 0.26174496644295303,
-      "eval_loss": 2.8679935932159424,
-      "eval_runtime": 113.6275,
-      "eval_samples_per_second": 3.934,
-      "eval_steps_per_second": 0.123,
-      "step": 513
+      "epoch": 36.01682692307692,
+      "eval_accuracy": 0.6914414414414415,
+      "eval_loss": 1.250654697418213,
+      "eval_runtime": 73.2059,
+      "eval_samples_per_second": 6.065,
+      "eval_steps_per_second": 0.096,
+      "step": 518
     },
     {
-      "epoch": 19.00841346153846,
-      "grad_norm": 9.284242630004883,
+      "epoch": 37.00240384615385,
+      "grad_norm": 4.395312786102295,
       "learning_rate": 2.0855614973262035e-05,
-      "loss": 2.0278,
+      "loss": 0.256,
       "step": 520
     },
     {
-      "epoch": 19.020432692307693,
-      "grad_norm": 6.913205623626709,
+      "epoch": 37.01442307692308,
+      "grad_norm": 3.43831729888916,
       "learning_rate": 2.018716577540107e-05,
-      "loss": 1.8727,
+      "loss": 0.2333,
       "step": 530
     },
     {
-      "epoch": 19.032451923076923,
-      "grad_norm": 21.923686981201172,
+      "epoch": 37.01682692307692,
+      "eval_accuracy": 0.7094594594594594,
+      "eval_loss": 1.2037817239761353,
+      "eval_runtime": 74.757,
+      "eval_samples_per_second": 5.939,
+      "eval_steps_per_second": 0.094,
+      "step": 532
+    },
+    {
+      "epoch": 38.00961538461539,
+      "grad_norm": 4.047479152679443,
       "learning_rate": 1.951871657754011e-05,
-      "loss": 1.9839,
+      "loss": 0.2366,
       "step": 540
     },
     {
-      "epoch": 19.032451923076923,
-      "eval_accuracy": 0.27069351230425054,
-      "eval_loss": 2.8359620571136475,
-      "eval_runtime": 115.0445,
-      "eval_samples_per_second": 3.885,
-      "eval_steps_per_second": 0.122,
-      "step": 540
+      "epoch": 38.01682692307692,
+      "eval_accuracy": 0.7207207207207207,
+      "eval_loss": 1.1517395973205566,
+      "eval_runtime": 73.764,
+      "eval_samples_per_second": 6.019,
+      "eval_steps_per_second": 0.095,
+      "step": 546
     },
     {
-      "epoch": 20.01201923076923,
-      "grad_norm": 8.244462013244629,
+      "epoch": 39.00480769230769,
+      "grad_norm": 5.707859516143799,
       "learning_rate": 1.8850267379679145e-05,
-      "loss": 1.7584,
+      "loss": 0.1938,
       "step": 550
     },
     {
-      "epoch": 20.02403846153846,
-      "grad_norm": 8.496162414550781,
+      "epoch": 39.01682692307692,
+      "grad_norm": 3.7590107917785645,
       "learning_rate": 1.8181818181818182e-05,
-      "loss": 1.7354,
+      "loss": 0.1886,
       "step": 560
     },
     {
-      "epoch": 20.032451923076923,
-      "eval_accuracy": 0.28187919463087246,
-      "eval_loss": 2.7890186309814453,
-      "eval_runtime": 111.5479,
-      "eval_samples_per_second": 4.007,
-      "eval_steps_per_second": 0.126,
-      "step": 567
+      "epoch": 39.01682692307692,
+      "eval_accuracy": 0.7094594594594594,
+      "eval_loss": 1.2073884010314941,
+      "eval_runtime": 72.0607,
+      "eval_samples_per_second": 6.161,
+      "eval_steps_per_second": 0.097,
+      "step": 560
     },
     {
-      "epoch": 21.00360576923077,
-      "grad_norm": 7.818769931793213,
+      "epoch": 40.01201923076923,
+      "grad_norm": 2.7063074111938477,
       "learning_rate": 1.7513368983957222e-05,
-      "loss": 1.701,
+      "loss": 0.1804,
       "step": 570
     },
     {
-      "epoch": 21.015625,
-      "grad_norm": 9.921280860900879,
+      "epoch": 40.01682692307692,
+      "eval_accuracy": 0.7027027027027027,
+      "eval_loss": 1.1658011674880981,
+      "eval_runtime": 77.6928,
+      "eval_samples_per_second": 5.715,
+      "eval_steps_per_second": 0.09,
+      "step": 574
+    },
+    {
+      "epoch": 41.00721153846154,
+      "grad_norm": 6.454692363739014,
       "learning_rate": 1.684491978609626e-05,
-      "loss": 1.648,
+      "loss": 0.1778,
       "step": 580
     },
     {
-      "epoch": 21.02764423076923,
-      "grad_norm": 8.053401947021484,
-      "learning_rate": 1.6176470588235296e-05,
-      "loss": 1.6843,
-      "step": 590
+      "epoch": 41.01682692307692,
+      "eval_accuracy": 0.6824324324324325,
+      "eval_loss": 1.2350265979766846,
+      "eval_runtime": 73.7916,
+      "eval_samples_per_second": 6.017,
+      "eval_steps_per_second": 0.095,
+      "step": 588
     },
     {
-      "epoch": 21.032451923076923,
-      "eval_accuracy": 0.29977628635346754,
-      "eval_loss": 2.728635787963867,
-      "eval_runtime": 114.938,
-      "eval_samples_per_second": 3.889,
-      "eval_steps_per_second": 0.122,
-      "step": 594
+      "epoch": 42.00240384615385,
+      "grad_norm": 3.444183349609375,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.2187,
+      "step": 590
     },
     {
-      "epoch": 22.00721153846154,
-      "grad_norm": 10.640948295593262,
+      "epoch": 42.01442307692308,
+      "grad_norm": 3.9001522064208984,
       "learning_rate": 1.5508021390374333e-05,
-      "loss": 1.4562,
+      "loss": 0.1728,
       "step": 600
     },
     {
-      "epoch": 22.01923076923077,
-      "grad_norm": 9.333779335021973,
+      "epoch": 42.01682692307692,
+      "eval_accuracy": 0.7162162162162162,
+      "eval_loss": 1.1637804508209229,
+      "eval_runtime": 77.911,
+      "eval_samples_per_second": 5.699,
+      "eval_steps_per_second": 0.09,
+      "step": 602
+    },
+    {
+      "epoch": 43.00961538461539,
+      "grad_norm": 7.290647029876709,
       "learning_rate": 1.4839572192513372e-05,
-      "loss": 1.5091,
+      "loss": 0.1998,
       "step": 610
     },
     {
-      "epoch": 22.03125,
-      "grad_norm": 9.433292388916016,
+      "epoch": 43.01682692307692,
+      "eval_accuracy": 0.6959459459459459,
+      "eval_loss": 1.2359126806259155,
+      "eval_runtime": 72.9535,
+      "eval_samples_per_second": 6.086,
+      "eval_steps_per_second": 0.096,
+      "step": 616
+    },
+    {
+      "epoch": 44.00480769230769,
+      "grad_norm": 5.636298656463623,
       "learning_rate": 1.4171122994652408e-05,
-      "loss": 1.6266,
+      "loss": 0.1639,
       "step": 620
     },
     {
-      "epoch": 22.032451923076923,
-      "eval_accuracy": 0.2841163310961969,
-      "eval_loss": 2.806154489517212,
-      "eval_runtime": 115.4195,
-      "eval_samples_per_second": 3.873,
-      "eval_steps_per_second": 0.121,
-      "step": 621
+      "epoch": 44.01682692307692,
+      "grad_norm": 3.6947968006134033,
+      "learning_rate": 1.3502673796791445e-05,
+      "loss": 0.1727,
+      "step": 630
     },
     {
-      "epoch": 23.010817307692307,
-      "grad_norm": 8.609794616699219,
-      "learning_rate": 1.3502673796791445e-05,
-      "loss": 1.6214,
+      "epoch": 44.01682692307692,
+      "eval_accuracy": 0.6936936936936937,
+      "eval_loss": 1.232095718383789,
+      "eval_runtime": 81.4582,
+      "eval_samples_per_second": 5.451,
+      "eval_steps_per_second": 0.086,
       "step": 630
     },
     {
-      "epoch": 23.02283653846154,
-      "grad_norm": 6.943145275115967,
+      "epoch": 45.01201923076923,
+      "grad_norm": 3.2019245624542236,
       "learning_rate": 1.2834224598930484e-05,
-      "loss": 1.4083,
+      "loss": 0.1564,
       "step": 640
     },
     {
-      "epoch": 23.032451923076923,
-      "eval_accuracy": 0.2595078299776286,
-      "eval_loss": 2.8204569816589355,
-      "eval_runtime": 115.0186,
-      "eval_samples_per_second": 3.886,
-      "eval_steps_per_second": 0.122,
-      "step": 648
+      "epoch": 45.01682692307692,
+      "eval_accuracy": 0.713963963963964,
+      "eval_loss": 1.1604844331741333,
+      "eval_runtime": 72.1149,
+      "eval_samples_per_second": 6.157,
+      "eval_steps_per_second": 0.097,
+      "step": 644
     },
     {
-      "epoch": 24.002403846153847,
-      "grad_norm": 10.106169700622559,
+      "epoch": 46.00721153846154,
+      "grad_norm": 1.9841790199279785,
       "learning_rate": 1.2165775401069519e-05,
-      "loss": 1.3264,
+      "loss": 0.1888,
       "step": 650
     },
     {
-      "epoch": 24.014423076923077,
-      "grad_norm": 10.27568244934082,
+      "epoch": 46.01682692307692,
+      "eval_accuracy": 0.7094594594594594,
+      "eval_loss": 1.1609092950820923,
+      "eval_runtime": 71.8777,
+      "eval_samples_per_second": 6.177,
+      "eval_steps_per_second": 0.097,
+      "step": 658
+    },
+    {
+      "epoch": 47.00240384615385,
+      "grad_norm": 7.085150241851807,
       "learning_rate": 1.1497326203208558e-05,
-      "loss": 1.3662,
+      "loss": 0.1618,
       "step": 660
     },
     {
-      "epoch": 24.026442307692307,
-      "grad_norm": 10.36124038696289,
+      "epoch": 47.01442307692308,
+      "grad_norm": 5.41819429397583,
       "learning_rate": 1.0828877005347594e-05,
-      "loss": 1.4422,
+      "loss": 0.1227,
       "step": 670
     },
     {
-      "epoch": 24.032451923076923,
-      "eval_accuracy": 0.30648769574944074,
-      "eval_loss": 2.6406848430633545,
-      "eval_runtime": 116.1998,
-      "eval_samples_per_second": 3.847,
-      "eval_steps_per_second": 0.12,
-      "step": 675
+      "epoch": 47.01682692307692,
+      "eval_accuracy": 0.7117117117117117,
+      "eval_loss": 1.1588457822799683,
+      "eval_runtime": 82.846,
+      "eval_samples_per_second": 5.359,
+      "eval_steps_per_second": 0.084,
+      "step": 672
     },
     {
-      "epoch": 25.006009615384617,
-      "grad_norm": 8.807711601257324,
+      "epoch": 48.00961538461539,
+      "grad_norm": 2.3894031047821045,
       "learning_rate": 1.0160427807486631e-05,
-      "loss": 1.3912,
+      "loss": 0.134,
       "step": 680
     },
     {
-      "epoch": 25.018028846153847,
-      "grad_norm": 8.629064559936523,
+      "epoch": 48.01682692307692,
+      "eval_accuracy": 0.7072072072072072,
+      "eval_loss": 1.1698147058486938,
+      "eval_runtime": 81.0499,
+      "eval_samples_per_second": 5.478,
+      "eval_steps_per_second": 0.086,
+      "step": 686
+    },
+    {
+      "epoch": 49.00480769230769,
+      "grad_norm": 4.01511287689209,
       "learning_rate": 9.49197860962567e-06,
-      "loss": 1.2904,
+      "loss": 0.1234,
       "step": 690
     },
     {
-      "epoch": 25.030048076923077,
-      "grad_norm": 9.190649032592773,
+      "epoch": 49.01682692307692,
+      "grad_norm": 7.249420642852783,
       "learning_rate": 8.823529411764707e-06,
-      "loss": 1.3897,
+      "loss": 0.1622,
       "step": 700
     },
     {
-      "epoch": 25.032451923076923,
-      "eval_accuracy": 0.34675615212527966,
-      "eval_loss": 2.5948002338409424,
-      "eval_runtime": 118.4696,
-      "eval_samples_per_second": 3.773,
-      "eval_steps_per_second": 0.118,
-      "step": 702
+      "epoch": 49.01682692307692,
+      "eval_accuracy": 0.6981981981981982,
+      "eval_loss": 1.2014145851135254,
+      "eval_runtime": 73.38,
+      "eval_samples_per_second": 6.051,
+      "eval_steps_per_second": 0.095,
+      "step": 700
     },
     {
-      "epoch": 26.009615384615383,
-      "grad_norm": 10.941615104675293,
+      "epoch": 50.01201923076923,
+      "grad_norm": 4.699652194976807,
       "learning_rate": 8.155080213903744e-06,
-      "loss": 1.2647,
+      "loss": 0.1391,
       "step": 710
     },
     {
-      "epoch": 26.021634615384617,
-      "grad_norm": 9.282670021057129,
+      "epoch": 50.01682692307692,
+      "eval_accuracy": 0.7162162162162162,
+      "eval_loss": 1.1005959510803223,
+      "eval_runtime": 73.5034,
+      "eval_samples_per_second": 6.041,
+      "eval_steps_per_second": 0.095,
+      "step": 714
+    },
+    {
+      "epoch": 51.00721153846154,
+      "grad_norm": 3.143523931503296,
       "learning_rate": 7.4866310160427806e-06,
-      "loss": 1.3906,
+      "loss": 0.1276,
       "step": 720
     },
     {
-      "epoch": 26.032451923076923,
-      "eval_accuracy": 0.31543624161073824,
-      "eval_loss": 2.629518985748291,
-      "eval_runtime": 120.316,
-      "eval_samples_per_second": 3.715,
-      "eval_steps_per_second": 0.116,
-      "step": 729
+      "epoch": 51.01682692307692,
+      "eval_accuracy": 0.6891891891891891,
+      "eval_loss": 1.1485284566879272,
+      "eval_runtime": 77.8096,
+      "eval_samples_per_second": 5.706,
+      "eval_steps_per_second": 0.09,
+      "step": 728
     },
     {
-      "epoch": 27.001201923076923,
-      "grad_norm": 9.254827499389648,
+      "epoch": 52.00240384615385,
+      "grad_norm": 1.1826622486114502,
       "learning_rate": 6.818181818181818e-06,
-      "loss": 1.3257,
+      "loss": 0.0979,
       "step": 730
     },
     {
-      "epoch": 27.013221153846153,
-      "grad_norm": 7.69126558303833,
+      "epoch": 52.01442307692308,
+      "grad_norm": 3.218552589416504,
       "learning_rate": 6.149732620320856e-06,
-      "loss": 1.2581,
+      "loss": 0.1222,
       "step": 740
     },
     {
-      "epoch": 27.025240384615383,
-      "grad_norm": 10.26952075958252,
+      "epoch": 52.01682692307692,
+      "eval_accuracy": 0.713963963963964,
+      "eval_loss": 1.0801595449447632,
+      "eval_runtime": 73.9346,
+      "eval_samples_per_second": 6.005,
+      "eval_steps_per_second": 0.095,
+      "step": 742
+    },
+    {
+      "epoch": 53.00961538461539,
+      "grad_norm": 4.1740312576293945,
       "learning_rate": 5.481283422459893e-06,
-      "loss": 1.2291,
+      "loss": 0.1024,
       "step": 750
     },
     {
-      "epoch": 27.032451923076923,
-      "eval_accuracy": 0.3378076062639821,
-      "eval_loss": 2.5539445877075195,
-      "eval_runtime": 116.6007,
-      "eval_samples_per_second": 3.834,
-      "eval_steps_per_second": 0.12,
+      "epoch": 53.01682692307692,
+      "eval_accuracy": 0.704954954954955,
+      "eval_loss": 1.1242848634719849,
+      "eval_runtime": 72.7853,
+      "eval_samples_per_second": 6.1,
+      "eval_steps_per_second": 0.096,
       "step": 756
     },
     {
-      "epoch": 28.004807692307693,
-      "grad_norm": 11.07676887512207,
+      "epoch": 54.00480769230769,
+      "grad_norm": 3.4602904319763184,
       "learning_rate": 4.812834224598931e-06,
-      "loss": 1.29,
+      "loss": 0.1014,
       "step": 760
     },
     {
-      "epoch": 28.016826923076923,
-      "grad_norm": 8.940054893493652,
+      "epoch": 54.01682692307692,
+      "grad_norm": 24.67173957824707,
       "learning_rate": 4.144385026737968e-06,
-      "loss": 1.1383,
+      "loss": 0.1186,
+      "step": 770
+    },
+    {
+      "epoch": 54.01682692307692,
+      "eval_accuracy": 0.7297297297297297,
+      "eval_loss": 1.0909080505371094,
+      "eval_runtime": 74.1097,
+      "eval_samples_per_second": 5.991,
+      "eval_steps_per_second": 0.094,
       "step": 770
     },
     {
-      "epoch": 28.028846153846153,
-      "grad_norm": 9.17213249206543,
+      "epoch": 55.01201923076923,
+      "grad_norm": 2.6615827083587646,
       "learning_rate": 3.4759358288770056e-06,
-      "loss": 1.3166,
+      "loss": 0.1121,
       "step": 780
     },
     {
-      "epoch": 28.032451923076923,
-      "eval_accuracy": 0.35570469798657717,
-      "eval_loss": 2.519993305206299,
-      "eval_runtime": 116.4901,
-      "eval_samples_per_second": 3.837,
-      "eval_steps_per_second": 0.12,
-      "step": 783
+      "epoch": 55.01682692307692,
+      "eval_accuracy": 0.7094594594594594,
+      "eval_loss": 1.1351189613342285,
+      "eval_runtime": 73.5115,
+      "eval_samples_per_second": 6.04,
+      "eval_steps_per_second": 0.095,
+      "step": 784
     },
     {
-      "epoch": 29.00841346153846,
-      "grad_norm": 11.884153366088867,
+      "epoch": 56.00721153846154,
+      "grad_norm": 3.412928581237793,
       "learning_rate": 2.807486631016043e-06,
-      "loss": 1.232,
+      "loss": 0.1284,
       "step": 790
     },
     {
-      "epoch": 29.020432692307693,
-      "grad_norm": 9.171028137207031,
+      "epoch": 56.01682692307692,
+      "eval_accuracy": 0.7274774774774775,
+      "eval_loss": 1.1095025539398193,
+      "eval_runtime": 73.055,
+      "eval_samples_per_second": 6.078,
+      "eval_steps_per_second": 0.096,
+      "step": 798
+    },
+    {
+      "epoch": 57.00240384615385,
+      "grad_norm": 4.214470386505127,
       "learning_rate": 2.1390374331550802e-06,
-      "loss": 1.1733,
+      "loss": 0.1418,
       "step": 800
     },
     {
-      "epoch": 29.032451923076923,
-      "grad_norm": 24.7429256439209,
+      "epoch": 57.01442307692308,
+      "grad_norm": 2.0941081047058105,
       "learning_rate": 1.4705882352941177e-06,
-      "loss": 1.2619,
+      "loss": 0.0893,
       "step": 810
     },
     {
-      "epoch": 29.032451923076923,
-      "eval_accuracy": 0.35570469798657717,
-      "eval_loss": 2.530792236328125,
-      "eval_runtime": 118.4418,
-      "eval_samples_per_second": 3.774,
-      "eval_steps_per_second": 0.118,
-      "step": 810
+      "epoch": 57.01682692307692,
+      "eval_accuracy": 0.7319819819819819,
+      "eval_loss": 1.084416389465332,
+      "eval_runtime": 71.5145,
+      "eval_samples_per_second": 6.209,
+      "eval_steps_per_second": 0.098,
+      "step": 812
     },
     {
-      "epoch": 30.01201923076923,
-      "grad_norm": 8.288737297058105,
+      "epoch": 58.00961538461539,
+      "grad_norm": 4.0398101806640625,
       "learning_rate": 8.021390374331552e-07,
-      "loss": 1.1824,
+      "loss": 0.0878,
       "step": 820
     },
     {
-      "epoch": 30.02403846153846,
-      "grad_norm": 8.749896049499512,
+      "epoch": 58.01682692307692,
+      "eval_accuracy": 0.7297297297297297,
+      "eval_loss": 1.0803806781768799,
+      "eval_runtime": 71.6842,
+      "eval_samples_per_second": 6.194,
+      "eval_steps_per_second": 0.098,
+      "step": 826
+    },
+    {
+      "epoch": 59.00480769230769,
+      "grad_norm": 1.0705435276031494,
       "learning_rate": 1.3368983957219251e-07,
-      "loss": 1.1393,
+      "loss": 0.0887,
       "step": 830
     },
     {
-      "epoch": 30.026442307692307,
-      "eval_accuracy": 0.3243847874720358,
-      "eval_loss": 2.5855681896209717,
-      "eval_runtime": 118.0615,
-      "eval_samples_per_second": 3.786,
-      "eval_steps_per_second": 0.119,
+      "epoch": 59.00721153846154,
+      "eval_accuracy": 0.7297297297297297,
+      "eval_loss": 1.0809234380722046,
+      "eval_runtime": 72.7582,
+      "eval_samples_per_second": 6.102,
+      "eval_steps_per_second": 0.096,
       "step": 832
     },
     {
-      "epoch": 30.026442307692307,
+      "epoch": 59.00721153846154,
       "step": 832,
-      "total_flos": 3.2147153612960956e+19,
-      "train_loss": 2.754230235058528,
-      "train_runtime": 10713.0029,
-      "train_samples_per_second": 2.485,
-      "train_steps_per_second": 0.078
-    },
-    {
-      "epoch": 30.026442307692307,
-      "eval_accuracy": 0.35570469798657717,
-      "eval_loss": 2.5194525718688965,
-      "eval_runtime": 114.2775,
-      "eval_samples_per_second": 3.912,
-      "eval_steps_per_second": 0.123,
+      "total_flos": 6.1975283578694664e+19,
+      "train_loss": 1.175738023289551,
+      "train_runtime": 12896.1846,
+      "train_samples_per_second": 4.129,
+      "train_steps_per_second": 0.065
+    },
+    {
+      "epoch": 59.00721153846154,
+      "eval_accuracy": 0.7319819819819819,
+      "eval_loss": 1.084416389465332,
+      "eval_runtime": 73.94,
+      "eval_samples_per_second": 6.005,
+      "eval_steps_per_second": 0.095,
       "step": 832
     },
     {
-      "epoch": 30.026442307692307,
-      "eval_accuracy": 0.35570469798657717,
-      "eval_loss": 2.520230293273926,
-      "eval_runtime": 112.9248,
-      "eval_samples_per_second": 3.958,
-      "eval_steps_per_second": 0.124,
+      "epoch": 59.00721153846154,
+      "eval_accuracy": 0.7319819819819819,
+      "eval_loss": 1.084416389465332,
+      "eval_runtime": 75.3126,
+      "eval_samples_per_second": 5.895,
+      "eval_steps_per_second": 0.093,
       "step": 832
     }
   ],
@@ -913,8 +1174,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 3.2147153612960956e+19,
-  "train_batch_size": 32,
+  "total_flos": 6.1975283578694664e+19,
+  "train_batch_size": 64,
   "trial_name": null,
   "trial_params": null
 }