End of training

all_results.json

@@ -1,14 +1,14 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.9067641496658325,
-    "eval_loss": 0.7214756608009338,
-    "eval_runtime": 2.2644,
+    "eval_accuracy": 0.8756855726242065,
+    "eval_loss": 0.810522198677063,
+    "eval_runtime": 3.6004,
     "eval_samples": 1094,
-    "eval_samples_per_second": 483.122,
-    "eval_steps_per_second": 60.501,
-    "train_loss": 0.27693930435361286,
-    "train_runtime": 2167.3602,
+    "eval_samples_per_second": 303.856,
+    "eval_steps_per_second": 38.051,
+    "train_loss": 1.72243186873619,
+    "train_runtime": 2003.3488,
     "train_samples": 10556,
-    "train_samples_per_second": 73.057,
-    "train_steps_per_second": 9.136
+    "train_samples_per_second": 79.038,
+    "train_steps_per_second": 2.471
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.9067641496658325,
-    "eval_loss": 0.7214756608009338,
-    "eval_runtime": 2.2644,
+    "eval_accuracy": 0.8756855726242065,
+    "eval_loss": 0.810522198677063,
+    "eval_runtime": 3.6004,
     "eval_samples": 1094,
-    "eval_samples_per_second": 483.122,
-    "eval_steps_per_second": 60.501
+    "eval_samples_per_second": 303.856,
+    "eval_steps_per_second": 38.051
 }

runs/Dec15_14-58-21_jbuhmann/events.out.tfevents.1639580315.jbuhmann.21699.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:16bdf0234f3b04d051d0cec69b08603faf02cf886899f82ed13e82cec7c0eb6a
-size 22692
+oid sha256:992d713823ebc6119fd87774065d5b8297cea2c13ad0b8e9f99aa0c7cd8a3b82
+size 23046

runs/Dec15_14-58-21_jbuhmann/events.out.tfevents.1639582323.jbuhmann.21699.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac566f80e7478542754cbe7ae0fe67741e2873d6ae22ff77b632871885dd66fe
+size 363

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "train_loss": 0.27693930435361286,
-    "train_runtime": 2167.3602,
+    "train_loss": 1.72243186873619,
+    "train_runtime": 2003.3488,
     "train_samples": 10556,
-    "train_samples_per_second": 73.057,
-    "train_steps_per_second": 9.136
+    "train_samples_per_second": 79.038,
+    "train_steps_per_second": 2.471
 }

trainer_state.json

@@ -2,393 +2,213 @@
   "best_metric": null,
   "best_model_checkpoint": null,
   "epoch": 15.0,
-  "global_step": 19800,
+  "global_step": 4950,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
-    {
-      "epoch": 0.38,
-      "learning_rate": 4.8737373737373736e-05,
-      "loss": 3.8927,
-      "step": 500
-    },
-    {
-      "epoch": 0.76,
-      "learning_rate": 4.7474747474747476e-05,
-      "loss": 2.3719,
-      "step": 1000
-    },
     {
       "epoch": 1.0,
-      "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": 4.621212121212121e-05,
-      "loss": 1.4046,
-      "step": 1500
+      "eval_accuracy": 0.22303473949432373,
+      "eval_loss": 3.9924213886260986,
+      "eval_runtime": 3.5667,
+      "eval_samples_per_second": 306.729,
+      "eval_steps_per_second": 38.411,
+      "step": 330
     },
     {
       "epoch": 1.52,
-      "learning_rate": 4.494949494949495e-05,
-      "loss": 0.7837,
-      "step": 2000
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 4.368686868686869e-05,
-      "loss": 0.5941,
-      "step": 2500
+      "learning_rate": 8.98989898989899e-06,
+      "loss": 4.3795,
+      "step": 500
     },
     {
       "epoch": 2.0,
-      "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": 4.242424242424243e-05,
-      "loss": 0.3401,
-      "step": 3000
-    },
-    {
-      "epoch": 2.65,
-      "learning_rate": 4.116161616161616e-05,
-      "loss": 0.2557,
-      "step": 3500
+      "eval_accuracy": 0.3957952558994293,
+      "eval_loss": 3.1812195777893066,
+      "eval_runtime": 3.5362,
+      "eval_samples_per_second": 309.369,
+      "eval_steps_per_second": 38.742,
+      "step": 660
     },
     {
       "epoch": 3.0,
-      "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
+      "eval_accuracy": 0.5511882901191711,
+      "eval_loss": 2.5903849601745605,
+      "eval_runtime": 3.5231,
+      "eval_samples_per_second": 310.519,
+      "eval_steps_per_second": 38.886,
+      "step": 990
     },
     {
       "epoch": 3.03,
-      "learning_rate": 3.98989898989899e-05,
-      "loss": 0.2031,
-      "step": 4000
-    },
-    {
-      "epoch": 3.41,
-      "learning_rate": 3.8636363636363636e-05,
-      "loss": 0.1193,
-      "step": 4500
-    },
-    {
-      "epoch": 3.79,
-      "learning_rate": 3.7373737373737376e-05,
-      "loss": 0.1178,
-      "step": 5000
+      "learning_rate": 7.97979797979798e-06,
+      "loss": 3.2046,
+      "step": 1000
     },
     {
       "epoch": 4.0,
-      "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": 3.611111111111111e-05,
-      "loss": 0.1101,
-      "step": 5500
+      "eval_accuracy": 0.6407678127288818,
+      "eval_loss": 2.1536314487457275,
+      "eval_runtime": 3.5444,
+      "eval_samples_per_second": 308.659,
+      "eval_steps_per_second": 38.653,
+      "step": 1320
     },
     {
       "epoch": 4.55,
-      "learning_rate": 3.484848484848485e-05,
-      "loss": 0.078,
-      "step": 6000
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 3.358585858585859e-05,
-      "loss": 0.0948,
-      "step": 6500
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 2.3683,
+      "step": 1500
     },
     {
       "epoch": 5.0,
-      "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": 3.232323232323233e-05,
-      "loss": 0.0352,
-      "step": 7000
-    },
-    {
-      "epoch": 5.68,
-      "learning_rate": 3.106060606060606e-05,
-      "loss": 0.079,
-      "step": 7500
+      "eval_accuracy": 0.7129799127578735,
+      "eval_loss": 1.8079293966293335,
+      "eval_runtime": 3.555,
+      "eval_samples_per_second": 307.739,
+      "eval_steps_per_second": 38.538,
+      "step": 1650
     },
     {
       "epoch": 6.0,
-      "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
+      "eval_accuracy": 0.7687385678291321,
+      "eval_loss": 1.5419940948486328,
+      "eval_runtime": 3.5056,
+      "eval_samples_per_second": 312.075,
+      "eval_steps_per_second": 39.081,
+      "step": 1980
     },
     {
       "epoch": 6.06,
-      "learning_rate": 2.9797979797979796e-05,
-      "loss": 0.054,
-      "step": 8000
-    },
-    {
-      "epoch": 6.44,
-      "learning_rate": 2.8535353535353536e-05,
-      "loss": 0.0388,
-      "step": 8500
-    },
-    {
-      "epoch": 6.82,
-      "learning_rate": 2.7272727272727273e-05,
-      "loss": 0.0413,
-      "step": 9000
+      "learning_rate": 5.95959595959596e-06,
+      "loss": 1.8065,
+      "step": 2000
     },
     {
       "epoch": 7.0,
-      "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": 2.6010101010101013e-05,
-      "loss": 0.046,
-      "step": 9500
+      "eval_accuracy": 0.8053016662597656,
+      "eval_loss": 1.3433690071105957,
+      "eval_runtime": 3.4889,
+      "eval_samples_per_second": 313.564,
+      "eval_steps_per_second": 39.267,
+      "step": 2310
     },
     {
       "epoch": 7.58,
-      "learning_rate": 2.474747474747475e-05,
-      "loss": 0.0295,
-      "step": 10000
-    },
-    {
-      "epoch": 7.95,
-      "learning_rate": 2.3484848484848487e-05,
-      "loss": 0.0291,
-      "step": 10500
+      "learning_rate": 4.94949494949495e-06,
+      "loss": 1.373,
+      "step": 2500
     },
     {
       "epoch": 8.0,
-      "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": 2.2222222222222223e-05,
-      "loss": 0.0236,
-      "step": 11000
-    },
-    {
-      "epoch": 8.71,
-      "learning_rate": 2.095959595959596e-05,
-      "loss": 0.032,
-      "step": 11500
+      "eval_accuracy": 0.825411319732666,
+      "eval_loss": 1.1881896257400513,
+      "eval_runtime": 3.5056,
+      "eval_samples_per_second": 312.068,
+      "eval_steps_per_second": 39.08,
+      "step": 2640
     },
     {
       "epoch": 9.0,
-      "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
+      "eval_accuracy": 0.8400365710258484,
+      "eval_loss": 1.0700345039367676,
+      "eval_runtime": 3.4951,
+      "eval_samples_per_second": 313.014,
+      "eval_steps_per_second": 39.198,
+      "step": 2970
     },
     {
       "epoch": 9.09,
-      "learning_rate": 1.9696969696969697e-05,
-      "loss": 0.0323,
-      "step": 12000
-    },
-    {
-      "epoch": 9.47,
-      "learning_rate": 1.8434343434343433e-05,
-      "loss": 0.0158,
-      "step": 12500
-    },
-    {
-      "epoch": 9.85,
-      "learning_rate": 1.7171717171717173e-05,
-      "loss": 0.027,
-      "step": 13000
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 1.0931,
+      "step": 3000
     },
     {
       "epoch": 10.0,
-      "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": 1.590909090909091e-05,
-      "loss": 0.0147,
-      "step": 13500
+      "eval_accuracy": 0.851005494594574,
+      "eval_loss": 0.9790602326393127,
+      "eval_runtime": 3.4892,
+      "eval_samples_per_second": 313.542,
+      "eval_steps_per_second": 39.264,
+      "step": 3300
     },
     {
       "epoch": 10.61,
-      "learning_rate": 1.4646464646464647e-05,
-      "loss": 0.0125,
-      "step": 14000
-    },
-    {
-      "epoch": 10.98,
-      "learning_rate": 1.3383838383838385e-05,
-      "loss": 0.0241,
-      "step": 14500
+      "learning_rate": 2.9292929292929295e-06,
+      "loss": 0.8714,
+      "step": 3500
     },
     {
       "epoch": 11.0,
-      "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": 1.2121212121212122e-05,
-      "loss": 0.0108,
-      "step": 15000
-    },
-    {
-      "epoch": 11.74,
-      "learning_rate": 1.085858585858586e-05,
-      "loss": 0.0109,
-      "step": 15500
+      "eval_accuracy": 0.8619744181632996,
+      "eval_loss": 0.9201710224151611,
+      "eval_runtime": 3.4883,
+      "eval_samples_per_second": 313.616,
+      "eval_steps_per_second": 39.274,
+      "step": 3630
     },
     {
       "epoch": 12.0,
-      "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
+      "eval_accuracy": 0.868372917175293,
+      "eval_loss": 0.8640827536582947,
+      "eval_runtime": 3.5075,
+      "eval_samples_per_second": 311.902,
+      "eval_steps_per_second": 39.059,
+      "step": 3960
     },
     {
       "epoch": 12.12,
-      "learning_rate": 9.595959595959595e-06,
-      "loss": 0.0122,
-      "step": 16000
-    },
-    {
-      "epoch": 12.5,
-      "learning_rate": 8.333333333333334e-06,
-      "loss": 0.0058,
-      "step": 16500
-    },
-    {
-      "epoch": 12.88,
-      "learning_rate": 7.0707070707070704e-06,
-      "loss": 0.0056,
-      "step": 17000
+      "learning_rate": 1.9191919191919192e-06,
+      "loss": 0.7428,
+      "step": 4000
     },
     {
       "epoch": 13.0,
-      "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": 5.808080808080808e-06,
-      "loss": 0.0055,
-      "step": 17500
+      "eval_accuracy": 0.8747714757919312,
+      "eval_loss": 0.8372448086738586,
+      "eval_runtime": 3.5209,
+      "eval_samples_per_second": 310.718,
+      "eval_steps_per_second": 38.911,
+      "step": 4290
     },
     {
       "epoch": 13.64,
-      "learning_rate": 4.5454545454545455e-06,
-      "loss": 0.0018,
-      "step": 18000
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.6531,
+      "step": 4500
     },
     {
       "epoch": 14.0,
-      "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": 3.2828282828282835e-06,
-      "loss": 0.0067,
-      "step": 18500
-    },
-    {
-      "epoch": 14.39,
-      "learning_rate": 2.0202020202020206e-06,
-      "loss": 0.0023,
-      "step": 19000
-    },
-    {
-      "epoch": 14.77,
-      "learning_rate": 7.575757575757576e-07,
-      "loss": 0.0017,
-      "step": 19500
+      "eval_accuracy": 0.8765996098518372,
+      "eval_loss": 0.8168175220489502,
+      "eval_runtime": 3.5294,
+      "eval_samples_per_second": 309.965,
+      "eval_steps_per_second": 38.816,
+      "step": 4620
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.9067641496658325,
-      "eval_loss": 0.7214756608009338,
-      "eval_runtime": 2.1084,
-      "eval_samples_per_second": 518.881,
-      "eval_steps_per_second": 64.979,
-      "step": 19800
+      "eval_accuracy": 0.8756855726242065,
+      "eval_loss": 0.810522198677063,
+      "eval_runtime": 3.5261,
+      "eval_samples_per_second": 310.256,
+      "eval_steps_per_second": 38.853,
+      "step": 4950
     },
     {
       "epoch": 15.0,
-      "step": 19800,
-      "total_flos": 5215995096399360.0,
-      "train_loss": 0.27693930435361286,
-      "train_runtime": 2167.3602,
-      "train_samples_per_second": 73.057,
-      "train_steps_per_second": 9.136
+      "step": 4950,
+      "total_flos": 1.043199019279872e+16,
+      "train_loss": 1.72243186873619,
+      "train_runtime": 2003.3488,
+      "train_samples_per_second": 79.038,
+      "train_steps_per_second": 2.471
     }
   ],
-  "max_steps": 19800,
+  "max_steps": 4950,
   "num_train_epochs": 15,
-  "total_flos": 5215995096399360.0,
+  "total_flos": 1.043199019279872e+16,
   "trial_name": null,
   "trial_params": null
 }