End of training

all_results.json

@@ -1,14 +1,14 @@
 {
-    "epoch": 15.0,
-    "eval_accuracy": 0.8756855726242065,
-    "eval_loss": 0.810522198677063,
-    "eval_runtime": 3.6004,
+    "epoch": 30.0,
+    "eval_accuracy": 0.9031078815460205,
+    "eval_loss": 0.4978463649749756,
+    "eval_runtime": 3.9767,
     "eval_samples": 1094,
-    "eval_samples_per_second": 303.856,
-    "eval_steps_per_second": 38.051,
-    "train_loss": 1.72243186873619,
-    "train_runtime": 2003.3488,
+    "eval_samples_per_second": 275.1,
+    "eval_steps_per_second": 34.45,
+    "train_loss": 0.7703812461429173,
+    "train_runtime": 4144.3399,
     "train_samples": 10556,
-    "train_samples_per_second": 79.038,
-    "train_steps_per_second": 2.471
+    "train_samples_per_second": 76.413,
+    "train_steps_per_second": 2.389
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 15.0,
-    "eval_accuracy": 0.8756855726242065,
-    "eval_loss": 0.810522198677063,
-    "eval_runtime": 3.6004,
+    "epoch": 30.0,
+    "eval_accuracy": 0.9031078815460205,
+    "eval_loss": 0.4978463649749756,
+    "eval_runtime": 3.9767,
     "eval_samples": 1094,
-    "eval_samples_per_second": 303.856,
-    "eval_steps_per_second": 38.051
+    "eval_samples_per_second": 275.1,
+    "eval_steps_per_second": 34.45
 }

runs/Dec15_15-40-52_jbuhmann/events.out.tfevents.1639582865.jbuhmann.28904.0

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:63cc129a444bc7a72e82f2e0c1d16935e1ce7f330009fe3d1239bfbc9ed68996
-size 29107
+oid sha256:d63d734834def9213d353aa33c52fadcbcf76e337c8773c5868054e65c635ecc
+size 29461

runs/Dec15_15-40-52_jbuhmann/events.out.tfevents.1639587014.jbuhmann.28904.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:26c9c33649a9e22d0ae15626e70632ae0dab2f12afe88208258091716a08f29b
+size 363

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 15.0,
-    "train_loss": 1.72243186873619,
-    "train_runtime": 2003.3488,
+    "epoch": 30.0,
+    "train_loss": 0.7703812461429173,
+    "train_runtime": 4144.3399,
     "train_samples": 10556,
-    "train_samples_per_second": 79.038,
-    "train_steps_per_second": 2.471
+    "train_samples_per_second": 76.413,
+    "train_steps_per_second": 2.389
 }

trainer_state.json

@@ -1,214 +1,409 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 15.0,
-  "global_step": 4950,
+  "epoch": 30.0,
+  "global_step": 9900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.22303473949432373,
-      "eval_loss": 3.9924213886260986,
-      "eval_runtime": 3.5667,
-      "eval_samples_per_second": 306.729,
-      "eval_steps_per_second": 38.411,
+      "eval_accuracy": 0.22486288845539093,
+      "eval_loss": 3.9691851139068604,
+      "eval_runtime": 3.5766,
+      "eval_samples_per_second": 305.88,
+      "eval_steps_per_second": 38.305,
       "step": 330
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.98989898989899e-06,
-      "loss": 4.3795,
+      "learning_rate": 9.494949494949497e-06,
+      "loss": 4.3672,
       "step": 500
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.3957952558994293,
-      "eval_loss": 3.1812195777893066,
-      "eval_runtime": 3.5362,
-      "eval_samples_per_second": 309.369,
-      "eval_steps_per_second": 38.742,
+      "eval_accuracy": 0.4031078517436981,
+      "eval_loss": 3.1312456130981445,
+      "eval_runtime": 3.5565,
+      "eval_samples_per_second": 307.607,
+      "eval_steps_per_second": 38.521,
       "step": 660
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.5511882901191711,
-      "eval_loss": 2.5903849601745605,
-      "eval_runtime": 3.5231,
-      "eval_samples_per_second": 310.519,
-      "eval_steps_per_second": 38.886,
+      "eval_accuracy": 0.5658135414123535,
+      "eval_loss": 2.5068106651306152,
+      "eval_runtime": 3.6412,
+      "eval_samples_per_second": 300.452,
+      "eval_steps_per_second": 37.625,
       "step": 990
     },
     {
       "epoch": 3.03,
-      "learning_rate": 7.97979797979798e-06,
-      "loss": 3.2046,
+      "learning_rate": 8.98989898989899e-06,
+      "loss": 3.1495,
       "step": 1000
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.6407678127288818,
-      "eval_loss": 2.1536314487457275,
-      "eval_runtime": 3.5444,
-      "eval_samples_per_second": 308.659,
-      "eval_steps_per_second": 38.653,
+      "eval_accuracy": 0.6599634289741516,
+      "eval_loss": 2.029958724975586,
+      "eval_runtime": 3.6051,
+      "eval_samples_per_second": 303.46,
+      "eval_steps_per_second": 38.002,
       "step": 1320
     },
     {
       "epoch": 4.55,
-      "learning_rate": 6.969696969696971e-06,
-      "loss": 2.3683,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 2.2491,
       "step": 1500
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.7129799127578735,
-      "eval_loss": 1.8079293966293335,
-      "eval_runtime": 3.555,
-      "eval_samples_per_second": 307.739,
-      "eval_steps_per_second": 38.538,
+      "eval_accuracy": 0.7449725866317749,
+      "eval_loss": 1.651670217514038,
+      "eval_runtime": 3.5564,
+      "eval_samples_per_second": 307.618,
+      "eval_steps_per_second": 38.523,
       "step": 1650
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.7687385678291321,
-      "eval_loss": 1.5419940948486328,
-      "eval_runtime": 3.5056,
-      "eval_samples_per_second": 312.075,
-      "eval_steps_per_second": 39.081,
+      "eval_accuracy": 0.79433274269104,
+      "eval_loss": 1.3604321479797363,
+      "eval_runtime": 3.5912,
+      "eval_samples_per_second": 304.631,
+      "eval_steps_per_second": 38.148,
       "step": 1980
     },
     {
       "epoch": 6.06,
-      "learning_rate": 5.95959595959596e-06,
-      "loss": 1.8065,
+      "learning_rate": 7.97979797979798e-06,
+      "loss": 1.622,
       "step": 2000
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.8053016662597656,
-      "eval_loss": 1.3433690071105957,
-      "eval_runtime": 3.4889,
-      "eval_samples_per_second": 313.564,
-      "eval_steps_per_second": 39.267,
+      "eval_accuracy": 0.8327239751815796,
+      "eval_loss": 1.132811427116394,
+      "eval_runtime": 3.5858,
+      "eval_samples_per_second": 305.092,
+      "eval_steps_per_second": 38.206,
       "step": 2310
     },
     {
       "epoch": 7.58,
-      "learning_rate": 4.94949494949495e-06,
-      "loss": 1.373,
+      "learning_rate": 7.474747474747476e-06,
+      "loss": 1.1252,
       "step": 2500
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.825411319732666,
-      "eval_loss": 1.1881896257400513,
-      "eval_runtime": 3.5056,
-      "eval_samples_per_second": 312.068,
-      "eval_steps_per_second": 39.08,
+      "eval_accuracy": 0.8610603213310242,
+      "eval_loss": 0.9484481811523438,
+      "eval_runtime": 3.5674,
+      "eval_samples_per_second": 306.664,
+      "eval_steps_per_second": 38.403,
       "step": 2640
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.8400365710258484,
-      "eval_loss": 1.0700345039367676,
-      "eval_runtime": 3.4951,
-      "eval_samples_per_second": 313.014,
-      "eval_steps_per_second": 39.198,
+      "eval_accuracy": 0.8756855726242065,
+      "eval_loss": 0.8212197422981262,
+      "eval_runtime": 3.5581,
+      "eval_samples_per_second": 307.464,
+      "eval_steps_per_second": 38.503,
       "step": 2970
     },
     {
       "epoch": 9.09,
-      "learning_rate": 3.93939393939394e-06,
-      "loss": 1.0931,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 0.7969,
       "step": 3000
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.851005494594574,
-      "eval_loss": 0.9790602326393127,
-      "eval_runtime": 3.4892,
-      "eval_samples_per_second": 313.542,
-      "eval_steps_per_second": 39.264,
+      "eval_accuracy": 0.8829981684684753,
+      "eval_loss": 0.7243201732635498,
+      "eval_runtime": 3.5763,
+      "eval_samples_per_second": 305.906,
+      "eval_steps_per_second": 38.308,
       "step": 3300
     },
     {
       "epoch": 10.61,
-      "learning_rate": 2.9292929292929295e-06,
-      "loss": 0.8714,
+      "learning_rate": 6.464646464646466e-06,
+      "loss": 0.5348,
       "step": 3500
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.8619744181632996,
-      "eval_loss": 0.9201710224151611,
-      "eval_runtime": 3.4883,
-      "eval_samples_per_second": 313.616,
-      "eval_steps_per_second": 39.274,
+      "eval_accuracy": 0.8866544961929321,
+      "eval_loss": 0.6596779227256775,
+      "eval_runtime": 3.5885,
+      "eval_samples_per_second": 304.866,
+      "eval_steps_per_second": 38.178,
       "step": 3630
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.868372917175293,
-      "eval_loss": 0.8640827536582947,
-      "eval_runtime": 3.5075,
-      "eval_samples_per_second": 311.902,
-      "eval_steps_per_second": 39.059,
+      "eval_accuracy": 0.8857403993606567,
+      "eval_loss": 0.598337709903717,
+      "eval_runtime": 3.6039,
+      "eval_samples_per_second": 303.56,
+      "eval_steps_per_second": 38.014,
       "step": 3960
     },
     {
       "epoch": 12.12,
-      "learning_rate": 1.9191919191919192e-06,
-      "loss": 0.7428,
+      "learning_rate": 5.95959595959596e-06,
+      "loss": 0.3744,
       "step": 4000
     },
     {
       "epoch": 13.0,
-      "eval_accuracy": 0.8747714757919312,
-      "eval_loss": 0.8372448086738586,
-      "eval_runtime": 3.5209,
-      "eval_samples_per_second": 310.718,
-      "eval_steps_per_second": 38.911,
+      "eval_accuracy": 0.8976234197616577,
+      "eval_loss": 0.5634561777114868,
+      "eval_runtime": 3.5353,
+      "eval_samples_per_second": 309.453,
+      "eval_steps_per_second": 38.752,
       "step": 4290
     },
     {
       "epoch": 13.64,
-      "learning_rate": 9.090909090909091e-07,
-      "loss": 0.6531,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 0.2564,
       "step": 4500
     },
     {
       "epoch": 14.0,
-      "eval_accuracy": 0.8765996098518372,
-      "eval_loss": 0.8168175220489502,
-      "eval_runtime": 3.5294,
-      "eval_samples_per_second": 309.965,
-      "eval_steps_per_second": 38.816,
+      "eval_accuracy": 0.8985374569892883,
+      "eval_loss": 0.5437070727348328,
+      "eval_runtime": 3.5598,
+      "eval_samples_per_second": 307.319,
+      "eval_steps_per_second": 38.485,
       "step": 4620
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.8756855726242065,
-      "eval_loss": 0.810522198677063,
-      "eval_runtime": 3.5261,
-      "eval_samples_per_second": 310.256,
-      "eval_steps_per_second": 38.853,
+      "eval_accuracy": 0.9012796878814697,
+      "eval_loss": 0.5124027132987976,
+      "eval_runtime": 3.5534,
+      "eval_samples_per_second": 307.874,
+      "eval_steps_per_second": 38.555,
       "step": 4950
     },
     {
-      "epoch": 15.0,
-      "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
+      "epoch": 15.15,
+      "learning_rate": 4.94949494949495e-06,
+      "loss": 0.1862,
+      "step": 5000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.9021937847137451,
+      "eval_loss": 0.5074306130409241,
+      "eval_runtime": 3.5582,
+      "eval_samples_per_second": 307.456,
+      "eval_steps_per_second": 38.502,
+      "step": 5280
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.1349,
+      "step": 5500
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.9049360156059265,
+      "eval_loss": 0.5027920007705688,
+      "eval_runtime": 3.5652,
+      "eval_samples_per_second": 306.859,
+      "eval_steps_per_second": 38.427,
+      "step": 5610
+    },
+    {
+      "epoch": 18.0,
+      "eval_accuracy": 0.9076782464981079,
+      "eval_loss": 0.4876061677932739,
+      "eval_runtime": 3.5354,
+      "eval_samples_per_second": 309.439,
+      "eval_steps_per_second": 38.751,
+      "step": 5940
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 0.0979,
+      "step": 6000
+    },
+    {
+      "epoch": 19.0,
+      "eval_accuracy": 0.9049360156059265,
+      "eval_loss": 0.4970707595348358,
+      "eval_runtime": 3.5724,
+      "eval_samples_per_second": 306.234,
+      "eval_steps_per_second": 38.349,
+      "step": 6270
+    },
+    {
+      "epoch": 19.7,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 0.0763,
+      "step": 6500
+    },
+    {
+      "epoch": 20.0,
+      "eval_accuracy": 0.9021937847137451,
+      "eval_loss": 0.4940781891345978,
+      "eval_runtime": 3.5051,
+      "eval_samples_per_second": 312.116,
+      "eval_steps_per_second": 39.086,
+      "step": 6600
+    },
+    {
+      "epoch": 21.0,
+      "eval_accuracy": 0.9049360156059265,
+      "eval_loss": 0.4956739842891693,
+      "eval_runtime": 3.5501,
+      "eval_samples_per_second": 308.164,
+      "eval_steps_per_second": 38.591,
+      "step": 6930
+    },
+    {
+      "epoch": 21.21,
+      "learning_rate": 2.9292929292929295e-06,
+      "loss": 0.0602,
+      "step": 7000
+    },
+    {
+      "epoch": 22.0,
+      "eval_accuracy": 0.9049360156059265,
+      "eval_loss": 0.4989473819732666,
+      "eval_runtime": 3.5171,
+      "eval_samples_per_second": 311.053,
+      "eval_steps_per_second": 38.953,
+      "step": 7260
+    },
+    {
+      "epoch": 22.73,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 0.0504,
+      "step": 7500
+    },
+    {
+      "epoch": 23.0,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.49591735005378723,
+      "eval_runtime": 3.5103,
+      "eval_samples_per_second": 311.654,
+      "eval_steps_per_second": 39.028,
+      "step": 7590
+    },
+    {
+      "epoch": 24.0,
+      "eval_accuracy": 0.9031078815460205,
+      "eval_loss": 0.4943903982639313,
+      "eval_runtime": 3.7255,
+      "eval_samples_per_second": 293.651,
+      "eval_steps_per_second": 36.774,
+      "step": 7920
+    },
+    {
+      "epoch": 24.24,
+      "learning_rate": 1.9191919191919192e-06,
+      "loss": 0.0422,
+      "step": 8000
+    },
+    {
+      "epoch": 25.0,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.4985043704509735,
+      "eval_runtime": 3.6733,
+      "eval_samples_per_second": 297.824,
+      "eval_steps_per_second": 37.296,
+      "step": 8250
+    },
+    {
+      "epoch": 25.76,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 0.0379,
+      "step": 8500
+    },
+    {
+      "epoch": 26.0,
+      "eval_accuracy": 0.9049360156059265,
+      "eval_loss": 0.4970065951347351,
+      "eval_runtime": 3.7215,
+      "eval_samples_per_second": 293.971,
+      "eval_steps_per_second": 36.814,
+      "step": 8580
+    },
+    {
+      "epoch": 27.0,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.4949011206626892,
+      "eval_runtime": 3.6072,
+      "eval_samples_per_second": 303.279,
+      "eval_steps_per_second": 37.979,
+      "step": 8910
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 0.0351,
+      "step": 9000
+    },
+    {
+      "epoch": 28.0,
+      "eval_accuracy": 0.9040219187736511,
+      "eval_loss": 0.49713513255119324,
+      "eval_runtime": 3.5069,
+      "eval_samples_per_second": 311.955,
+      "eval_steps_per_second": 39.066,
+      "step": 9240
+    },
+    {
+      "epoch": 28.79,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 0.0321,
+      "step": 9500
+    },
+    {
+      "epoch": 29.0,
+      "eval_accuracy": 0.9031078815460205,
+      "eval_loss": 0.49672842025756836,
+      "eval_runtime": 3.506,
+      "eval_samples_per_second": 312.034,
+      "eval_steps_per_second": 39.076,
+      "step": 9570
+    },
+    {
+      "epoch": 30.0,
+      "eval_accuracy": 0.9031078815460205,
+      "eval_loss": 0.4978463649749756,
+      "eval_runtime": 3.6907,
+      "eval_samples_per_second": 296.421,
+      "eval_steps_per_second": 37.12,
+      "step": 9900
+    },
+    {
+      "epoch": 30.0,
+      "step": 9900,
+      "total_flos": 2.086398038559744e+16,
+      "train_loss": 0.7703812461429173,
+      "train_runtime": 4144.3399,
+      "train_samples_per_second": 76.413,
+      "train_steps_per_second": 2.389
     }
   ],
-  "max_steps": 4950,
-  "num_train_epochs": 15,
-  "total_flos": 1.043199019279872e+16,
+  "max_steps": 9900,
+  "num_train_epochs": 30,
+  "total_flos": 2.086398038559744e+16,
   "trial_name": null,
   "trial_params": null
 }