End of training

all_results.json

@@ -1,14 +1,14 @@
 {
     "epoch": 15.0,
     "eval_accuracy": 0.9067641496658325,
-    "eval_loss": 0.4963526129722595,
-    "eval_runtime": 2.1572,
+    "eval_loss": 0.7214756608009338,
+    "eval_runtime": 2.2644,
     "eval_samples": 1094,
-    "eval_samples_per_second": 507.137,
-    "eval_steps_per_second": 63.508,
-    "train_loss": 0.7935866661746093,
-    "train_runtime": 2196.4523,
+    "eval_samples_per_second": 483.122,
+    "eval_steps_per_second": 60.501,
+    "train_loss": 0.27693930435361286,
+    "train_runtime": 2167.3602,
     "train_samples": 10556,
-    "train_samples_per_second": 72.089,
-    "train_steps_per_second": 9.015
+    "train_samples_per_second": 73.057,
+    "train_steps_per_second": 9.136
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 15.0,
     "eval_accuracy": 0.9067641496658325,
-    "eval_loss": 0.4963526129722595,
-    "eval_runtime": 2.1572,
+    "eval_loss": 0.7214756608009338,
+    "eval_runtime": 2.2644,
     "eval_samples": 1094,
-    "eval_samples_per_second": 507.137,
-    "eval_steps_per_second": 63.508
+    "eval_samples_per_second": 483.122,
+    "eval_steps_per_second": 60.501
 }

runs/Dec15_14-01-16_jbuhmann/events.out.tfevents.1639576890.jbuhmann.8432.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f7df785663d705213bc744ff5983ed2c9022695b1f4e503dc22bb24ce438eae9
-size 27441
+oid sha256:92d25c72fc47aaef4b443c760e64e4303f9178d0da44308159abac76d4720070
+size 27801

runs/Dec15_14-01-16_jbuhmann/events.out.tfevents.1639579061.jbuhmann.8432.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:455e4b3d9a8f6e4586a8f45e1a8e709443355b2036a315b24582c19a12277aed
+size 369

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "train_loss": 0.7935866661746093,
-    "train_runtime": 2196.4523,
+    "train_loss": 0.27693930435361286,
+    "train_runtime": 2167.3602,
     "train_samples": 10556,
-    "train_samples_per_second": 72.089,
-    "train_steps_per_second": 9.015
+    "train_samples_per_second": 73.057,
+    "train_steps_per_second": 9.136
 }

trainer_state.json

@@ -9,381 +9,381 @@
   "log_history": [
     {
       "epoch": 0.38,
-      "learning_rate": 9.747474747474748e-06,
-      "loss": 4.718,
+      "learning_rate": 4.8737373737373736e-05,
+      "loss": 3.8927,
       "step": 500
     },
     {
       "epoch": 0.76,
-      "learning_rate": 9.494949494949497e-06,
-      "loss": 4.0196,
+      "learning_rate": 4.7474747474747476e-05,
+      "loss": 2.3719,
       "step": 1000
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.4232175648212433,
-      "eval_loss": 3.169081211090088,
-      "eval_runtime": 2.1415,
-      "eval_samples_per_second": 510.858,
-      "eval_steps_per_second": 63.974,
+      "eval_accuracy": 0.7787933945655823,
+      "eval_loss": 1.1852809190750122,
+      "eval_runtime": 2.0844,
+      "eval_samples_per_second": 524.845,
+      "eval_steps_per_second": 65.726,
       "step": 1320
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.242424242424244e-06,
-      "loss": 3.4554,
+      "learning_rate": 4.621212121212121e-05,
+      "loss": 1.4046,
       "step": 1500
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.98989898989899e-06,
-      "loss": 2.8904,
+      "learning_rate": 4.494949494949495e-05,
+      "loss": 0.7837,
       "step": 2000
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.737373737373738e-06,
-      "loss": 2.5409,
+      "learning_rate": 4.368686868686869e-05,
+      "loss": 0.5941,
       "step": 2500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.6435100436210632,
-      "eval_loss": 2.04946231842041,
-      "eval_runtime": 2.1238,
-      "eval_samples_per_second": 515.114,
-      "eval_steps_per_second": 64.507,
+      "eval_accuracy": 0.880255937576294,
+      "eval_loss": 0.5889360308647156,
+      "eval_runtime": 2.0716,
+      "eval_samples_per_second": 528.095,
+      "eval_steps_per_second": 66.133,
       "step": 2640
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.484848484848486e-06,
-      "loss": 2.1224,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.3401,
       "step": 3000
     },
     {
       "epoch": 2.65,
-      "learning_rate": 8.232323232323233e-06,
-      "loss": 1.8396,
+      "learning_rate": 4.116161616161616e-05,
+      "loss": 0.2557,
       "step": 3500
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.7861060500144958,
-      "eval_loss": 1.3566279411315918,
-      "eval_runtime": 2.0996,
-      "eval_samples_per_second": 521.061,
-      "eval_steps_per_second": 65.252,
+      "eval_accuracy": 0.8866544961929321,
+      "eval_loss": 0.5768323540687561,
+      "eval_runtime": 2.1209,
+      "eval_samples_per_second": 515.813,
+      "eval_steps_per_second": 64.595,
       "step": 3960
     },
     {
       "epoch": 3.03,
-      "learning_rate": 7.97979797979798e-06,
-      "loss": 1.5406,
+      "learning_rate": 3.98989898989899e-05,
+      "loss": 0.2031,
       "step": 4000
     },
     {
       "epoch": 3.41,
-      "learning_rate": 7.727272727272727e-06,
-      "loss": 1.2519,
+      "learning_rate": 3.8636363636363636e-05,
+      "loss": 0.1193,
       "step": 4500
     },
     {
       "epoch": 3.79,
-      "learning_rate": 7.474747474747476e-06,
-      "loss": 1.0986,
+      "learning_rate": 3.7373737373737376e-05,
+      "loss": 0.1178,
       "step": 5000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8555758595466614,
-      "eval_loss": 0.9564884305000305,
-      "eval_runtime": 2.1078,
-      "eval_samples_per_second": 519.033,
-      "eval_steps_per_second": 64.998,
+      "eval_accuracy": 0.8811700344085693,
+      "eval_loss": 0.6243242621421814,
+      "eval_runtime": 2.1461,
+      "eval_samples_per_second": 509.767,
+      "eval_steps_per_second": 63.837,
       "step": 5280
     },
     {
       "epoch": 4.17,
-      "learning_rate": 7.222222222222223e-06,
-      "loss": 0.9206,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.1101,
       "step": 5500
     },
     {
       "epoch": 4.55,
-      "learning_rate": 6.969696969696971e-06,
-      "loss": 0.7389,
+      "learning_rate": 3.484848484848485e-05,
+      "loss": 0.078,
       "step": 6000
     },
     {
       "epoch": 4.92,
-      "learning_rate": 6.717171717171718e-06,
-      "loss": 0.6711,
+      "learning_rate": 3.358585858585859e-05,
+      "loss": 0.0948,
       "step": 6500
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.880255937576294,
-      "eval_loss": 0.7118942141532898,
-      "eval_runtime": 2.0744,
-      "eval_samples_per_second": 527.383,
-      "eval_steps_per_second": 66.043,
+      "eval_accuracy": 0.8912248611450195,
+      "eval_loss": 0.650793194770813,
+      "eval_runtime": 2.0892,
+      "eval_samples_per_second": 523.641,
+      "eval_steps_per_second": 65.575,
       "step": 6600
     },
     {
       "epoch": 5.3,
-      "learning_rate": 6.464646464646466e-06,
-      "loss": 0.5107,
+      "learning_rate": 3.232323232323233e-05,
+      "loss": 0.0352,
       "step": 7000
     },
     {
       "epoch": 5.68,
-      "learning_rate": 6.212121212121213e-06,
-      "loss": 0.4456,
+      "learning_rate": 3.106060606060606e-05,
+      "loss": 0.079,
       "step": 7500
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.8912248611450195,
-      "eval_loss": 0.5804843902587891,
-      "eval_runtime": 2.0875,
-      "eval_samples_per_second": 524.077,
-      "eval_steps_per_second": 65.629,
+      "eval_accuracy": 0.8930529952049255,
+      "eval_loss": 0.6763377785682678,
+      "eval_runtime": 2.0996,
+      "eval_samples_per_second": 521.052,
+      "eval_steps_per_second": 65.251,
       "step": 7920
     },
     {
       "epoch": 6.06,
-      "learning_rate": 5.95959595959596e-06,
-      "loss": 0.3836,
+      "learning_rate": 2.9797979797979796e-05,
+      "loss": 0.054,
       "step": 8000
     },
     {
       "epoch": 6.44,
-      "learning_rate": 5.7070707070707075e-06,
-      "loss": 0.2848,
+      "learning_rate": 2.8535353535353536e-05,
+      "loss": 0.0388,
       "step": 8500
     },
     {
       "epoch": 6.82,
-      "learning_rate": 5.4545454545454545e-06,
-      "loss": 0.2563,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.0413,
       "step": 9000
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.8994515538215637,
-      "eval_loss": 0.5123510956764221,
-      "eval_runtime": 2.1263,
-      "eval_samples_per_second": 514.504,
-      "eval_steps_per_second": 64.431,
+      "eval_accuracy": 0.9021937847137451,
+      "eval_loss": 0.6991991400718689,
+      "eval_runtime": 2.1868,
+      "eval_samples_per_second": 500.277,
+      "eval_steps_per_second": 62.649,
       "step": 9240
     },
     {
       "epoch": 7.2,
-      "learning_rate": 5.202020202020202e-06,
-      "loss": 0.2322,
+      "learning_rate": 2.6010101010101013e-05,
+      "loss": 0.046,
       "step": 9500
     },
     {
       "epoch": 7.58,
-      "learning_rate": 4.94949494949495e-06,
-      "loss": 0.1742,
+      "learning_rate": 2.474747474747475e-05,
+      "loss": 0.0295,
       "step": 10000
     },
     {
       "epoch": 7.95,
-      "learning_rate": 4.696969696969698e-06,
-      "loss": 0.1686,
+      "learning_rate": 2.3484848484848487e-05,
+      "loss": 0.0291,
       "step": 10500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.9021937847137451,
-      "eval_loss": 0.49525973200798035,
-      "eval_runtime": 2.0716,
-      "eval_samples_per_second": 528.088,
-      "eval_steps_per_second": 66.132,
+      "eval_accuracy": 0.8921389579772949,
+      "eval_loss": 0.7643230557441711,
+      "eval_runtime": 2.2005,
+      "eval_samples_per_second": 497.161,
+      "eval_steps_per_second": 62.259,
       "step": 10560
     },
     {
       "epoch": 8.33,
-      "learning_rate": 4.444444444444444e-06,
-      "loss": 0.1313,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.0236,
       "step": 11000
     },
     {
       "epoch": 8.71,
-      "learning_rate": 4.191919191919192e-06,
-      "loss": 0.1135,
+      "learning_rate": 2.095959595959596e-05,
+      "loss": 0.032,
       "step": 11500
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.9021937847137451,
-      "eval_loss": 0.486285001039505,
-      "eval_runtime": 2.0803,
-      "eval_samples_per_second": 525.893,
-      "eval_steps_per_second": 65.857,
+      "eval_accuracy": 0.9113345742225647,
+      "eval_loss": 0.6661025285720825,
+      "eval_runtime": 2.0869,
+      "eval_samples_per_second": 524.224,
+      "eval_steps_per_second": 65.648,
       "step": 11880
     },
     {
       "epoch": 9.09,
-      "learning_rate": 3.93939393939394e-06,
-      "loss": 0.1161,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 0.0323,
       "step": 12000
     },
     {
       "epoch": 9.47,
-      "learning_rate": 3.686868686868687e-06,
-      "loss": 0.0839,
+      "learning_rate": 1.8434343434343433e-05,
+      "loss": 0.0158,
       "step": 12500
     },
     {
       "epoch": 9.85,
-      "learning_rate": 3.4343434343434347e-06,
-      "loss": 0.0846,
+      "learning_rate": 1.7171717171717173e-05,
+      "loss": 0.027,
       "step": 13000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.9049360156059265,
-      "eval_loss": 0.48929038643836975,
-      "eval_runtime": 2.0752,
-      "eval_samples_per_second": 527.187,
-      "eval_steps_per_second": 66.019,
+      "eval_accuracy": 0.9085923433303833,
+      "eval_loss": 0.6882250308990479,
+      "eval_runtime": 2.2029,
+      "eval_samples_per_second": 496.616,
+      "eval_steps_per_second": 62.19,
       "step": 13200
     },
     {
       "epoch": 10.23,
-      "learning_rate": 3.181818181818182e-06,
-      "loss": 0.0658,
+      "learning_rate": 1.590909090909091e-05,
+      "loss": 0.0147,
       "step": 13500
     },
     {
       "epoch": 10.61,
-      "learning_rate": 2.9292929292929295e-06,
-      "loss": 0.0663,
+      "learning_rate": 1.4646464646464647e-05,
+      "loss": 0.0125,
       "step": 14000
     },
     {
       "epoch": 10.98,
-      "learning_rate": 2.676767676767677e-06,
-      "loss": 0.07,
+      "learning_rate": 1.3383838383838385e-05,
+      "loss": 0.0241,
       "step": 14500
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.9076782464981079,
-      "eval_loss": 0.485994428396225,
-      "eval_runtime": 2.1075,
-      "eval_samples_per_second": 519.1,
-      "eval_steps_per_second": 65.006,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.7250856757164001,
+      "eval_runtime": 2.0725,
+      "eval_samples_per_second": 527.862,
+      "eval_steps_per_second": 66.103,
       "step": 14520
     },
     {
       "epoch": 11.36,
-      "learning_rate": 2.4242424242424244e-06,
-      "loss": 0.0489,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.0108,
       "step": 15000
     },
     {
       "epoch": 11.74,
-      "learning_rate": 2.171717171717172e-06,
-      "loss": 0.0487,
+      "learning_rate": 1.085858585858586e-05,
+      "loss": 0.0109,
       "step": 15500
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.9058501124382019,
-      "eval_loss": 0.4888700246810913,
-      "eval_runtime": 2.0826,
-      "eval_samples_per_second": 525.296,
-      "eval_steps_per_second": 65.782,
+      "eval_accuracy": 0.8976234197616577,
+      "eval_loss": 0.7604121565818787,
+      "eval_runtime": 2.2757,
+      "eval_samples_per_second": 480.738,
+      "eval_steps_per_second": 60.202,
       "step": 15840
     },
     {
       "epoch": 12.12,
-      "learning_rate": 1.9191919191919192e-06,
-      "loss": 0.049,
+      "learning_rate": 9.595959595959595e-06,
+      "loss": 0.0122,
       "step": 16000
     },
     {
       "epoch": 12.5,
-      "learning_rate": 1.6666666666666667e-06,
-      "loss": 0.0476,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.0058,
       "step": 16500
     },
     {
       "epoch": 12.88,
-      "learning_rate": 1.4141414141414143e-06,
-      "loss": 0.0342,
+      "learning_rate": 7.0707070707070704e-06,
+      "loss": 0.0056,
       "step": 17000
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.9067641496658325,
-      "eval_loss": 0.4926171898841858,
-      "eval_runtime": 2.3107,
-      "eval_samples_per_second": 473.444,
-      "eval_steps_per_second": 59.289,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.7356353998184204,
+      "eval_runtime": 2.1586,
+      "eval_samples_per_second": 506.799,
+      "eval_steps_per_second": 63.466,
       "step": 17160
     },
     {
       "epoch": 13.26,
-      "learning_rate": 1.1616161616161617e-06,
-      "loss": 0.0392,
+      "learning_rate": 5.808080808080808e-06,
+      "loss": 0.0055,
       "step": 17500
     },
     {
       "epoch": 13.64,
-      "learning_rate": 9.090909090909091e-07,
-      "loss": 0.0332,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 0.0018,
       "step": 18000
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.9067641496658325,
-      "eval_loss": 0.495064914226532,
-      "eval_runtime": 2.1748,
-      "eval_samples_per_second": 503.039,
-      "eval_steps_per_second": 62.995,
+      "eval_accuracy": 0.9076782464981079,
+      "eval_loss": 0.7189434170722961,
+      "eval_runtime": 2.0825,
+      "eval_samples_per_second": 525.332,
+      "eval_steps_per_second": 65.787,
       "step": 18480
     },
     {
       "epoch": 14.02,
-      "learning_rate": 6.565656565656567e-07,
-      "loss": 0.0411,
+      "learning_rate": 3.2828282828282835e-06,
+      "loss": 0.0067,
       "step": 18500
     },
     {
       "epoch": 14.39,
-      "learning_rate": 4.040404040404041e-07,
-      "loss": 0.0368,
+      "learning_rate": 2.0202020202020206e-06,
+      "loss": 0.0023,
       "step": 19000
     },
     {
       "epoch": 14.77,
-      "learning_rate": 1.5151515151515152e-07,
-      "loss": 0.029,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 0.0017,
       "step": 19500
     },
     {
       "epoch": 15.0,
       "eval_accuracy": 0.9067641496658325,
-      "eval_loss": 0.4963526129722595,
-      "eval_runtime": 2.0713,
-      "eval_samples_per_second": 528.178,
-      "eval_steps_per_second": 66.143,
+      "eval_loss": 0.7214756608009338,
+      "eval_runtime": 2.1084,
+      "eval_samples_per_second": 518.881,
+      "eval_steps_per_second": 64.979,
       "step": 19800
     },
     {
       "epoch": 15.0,
       "step": 19800,
       "total_flos": 5215995096399360.0,
-      "train_loss": 0.7935866661746093,
-      "train_runtime": 2196.4523,
-      "train_samples_per_second": 72.089,
-      "train_steps_per_second": 9.015
+      "train_loss": 0.27693930435361286,
+      "train_runtime": 2167.3602,
+      "train_samples_per_second": 73.057,
+      "train_steps_per_second": 9.136
     }
   ],
   "max_steps": 19800,