End of training

all_results.json

@@ -1,15 +1,15 @@
 {
-    "epoch": 5.0,
-    "eval_loss": 26.253612518310547,
-    "eval_runtime": 5.982,
+    "epoch": 20.0,
+    "eval_loss": 6.937458515167236,
+    "eval_runtime": 5.7217,
     "eval_samples": 128,
-    "eval_samples_per_second": 21.397,
-    "eval_steps_per_second": 0.334,
+    "eval_samples_per_second": 22.371,
+    "eval_steps_per_second": 0.35,
     "eval_wer": 1.0,
-    "total_flos": 1.3476444758728704e+17,
-    "train_loss": 16.66825189590454,
-    "train_runtime": 91.9274,
+    "total_flos": 5.430583918308557e+17,
+    "train_loss": 8.69529299736023,
+    "train_runtime": 243.8197,
     "train_samples": 128,
-    "train_samples_per_second": 6.962,
-    "train_steps_per_second": 0.109
+    "train_samples_per_second": 10.5,
+    "train_steps_per_second": 0.164
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 5.0,
-    "eval_loss": 26.253612518310547,
-    "eval_runtime": 5.982,
+    "epoch": 20.0,
+    "eval_loss": 6.937458515167236,
+    "eval_runtime": 5.7217,
     "eval_samples": 128,
-    "eval_samples_per_second": 21.397,
-    "eval_steps_per_second": 0.334,
+    "eval_samples_per_second": 22.371,
+    "eval_steps_per_second": 0.35,
     "eval_wer": 1.0
 }

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 5.0,
-    "total_flos": 1.3476444758728704e+17,
-    "train_loss": 16.66825189590454,
-    "train_runtime": 91.9274,
+    "epoch": 20.0,
+    "total_flos": 5.430583918308557e+17,
+    "train_loss": 8.69529299736023,
+    "train_runtime": 243.8197,
     "train_samples": 128,
-    "train_samples_per_second": 6.962,
-    "train_steps_per_second": 0.109
+    "train_samples_per_second": 10.5,
+    "train_steps_per_second": 0.164
 }

trainer_state.json

@@ -1,175 +1,325 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 5.0,
-  "global_step": 10,
+  "epoch": 20.0,
+  "global_step": 40,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "learning_rate": 7.5e-05,
-      "loss": 18.7839,
-      "step": 1
+      "loss": 19.0933,
+      "step": 2
     },
     {
-      "epoch": 0.5,
+      "epoch": 1.0,
       "eval_loss": 32.55470657348633,
-      "eval_runtime": 5.5933,
-      "eval_samples_per_second": 22.884,
-      "eval_steps_per_second": 0.358,
+      "eval_runtime": 5.3015,
+      "eval_samples_per_second": 24.144,
+      "eval_steps_per_second": 0.377,
       "eval_wer": 1.0,
-      "step": 1
+      "step": 2
     },
     {
-      "epoch": 1.0,
-      "learning_rate": 7.5e-05,
-      "loss": 19.4045,
-      "step": 2
+      "epoch": 2.0,
+      "learning_rate": 7.125e-05,
+      "loss": 18.6757,
+      "step": 4
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 32.55470657348633,
-      "eval_runtime": 5.6123,
-      "eval_samples_per_second": 22.807,
-      "eval_steps_per_second": 0.356,
+      "epoch": 2.0,
+      "eval_loss": 31.040802001953125,
+      "eval_runtime": 5.2939,
+      "eval_samples_per_second": 24.179,
+      "eval_steps_per_second": 0.378,
       "eval_wer": 1.0,
-      "step": 2
+      "step": 4
     },
     {
-      "epoch": 1.5,
-      "learning_rate": 6.75e-05,
-      "loss": 19.6279,
-      "step": 3
+      "epoch": 3.0,
+      "learning_rate": 6.937499999999999e-05,
+      "loss": 17.1649,
+      "step": 6
     },
     {
-      "epoch": 1.5,
-      "eval_loss": 31.884950637817383,
-      "eval_runtime": 5.4543,
-      "eval_samples_per_second": 23.468,
-      "eval_steps_per_second": 0.367,
+      "epoch": 3.0,
+      "eval_loss": 29.57763671875,
+      "eval_runtime": 5.2818,
+      "eval_samples_per_second": 24.234,
+      "eval_steps_per_second": 0.379,
       "eval_wer": 1.0,
-      "step": 3
+      "step": 6
     },
     {
-      "epoch": 2.0,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 17.7216,
-      "step": 4
+      "epoch": 4.0,
+      "learning_rate": 6.5625e-05,
+      "loss": 14.7415,
+      "step": 8
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 31.10820960998535,
-      "eval_runtime": 5.4982,
-      "eval_samples_per_second": 23.28,
-      "eval_steps_per_second": 0.364,
+      "epoch": 4.0,
+      "eval_loss": 22.918315887451172,
+      "eval_runtime": 5.4291,
+      "eval_samples_per_second": 23.577,
+      "eval_steps_per_second": 0.368,
       "eval_wer": 1.0,
-      "step": 4
+      "step": 8
     },
     {
-      "epoch": 2.5,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 17.3103,
-      "step": 5
+      "epoch": 5.0,
+      "learning_rate": 6.187499999999999e-05,
+      "loss": 11.8071,
+      "step": 10
     },
     {
-      "epoch": 2.5,
-      "eval_loss": 31.10820960998535,
-      "eval_runtime": 5.9789,
-      "eval_samples_per_second": 21.409,
-      "eval_steps_per_second": 0.335,
+      "epoch": 5.0,
+      "eval_loss": 17.507305145263672,
+      "eval_runtime": 5.3146,
+      "eval_samples_per_second": 24.085,
+      "eval_steps_per_second": 0.376,
       "eval_wer": 1.0,
-      "step": 5
+      "step": 10
     },
     {
-      "epoch": 3.0,
-      "learning_rate": 5.2499999999999995e-05,
-      "loss": 17.2026,
-      "step": 6
+      "epoch": 6.0,
+      "learning_rate": 5.8124999999999997e-05,
+      "loss": 9.7675,
+      "step": 12
     },
     {
-      "epoch": 3.0,
-      "eval_loss": 29.966575622558594,
-      "eval_runtime": 5.3861,
-      "eval_samples_per_second": 23.765,
-      "eval_steps_per_second": 0.371,
+      "epoch": 6.0,
+      "eval_loss": 14.17426872253418,
+      "eval_runtime": 5.4054,
+      "eval_samples_per_second": 23.68,
+      "eval_steps_per_second": 0.37,
       "eval_wer": 1.0,
-      "step": 6
+      "step": 12
     },
     {
-      "epoch": 3.5,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 16.7911,
-      "step": 7
+      "epoch": 7.0,
+      "learning_rate": 5.4374999999999994e-05,
+      "loss": 8.4193,
+      "step": 14
     },
     {
-      "epoch": 3.5,
-      "eval_loss": 28.54442596435547,
-      "eval_runtime": 5.434,
-      "eval_samples_per_second": 23.555,
-      "eval_steps_per_second": 0.368,
+      "epoch": 7.0,
+      "eval_loss": 12.122542381286621,
+      "eval_runtime": 5.2782,
+      "eval_samples_per_second": 24.251,
+      "eval_steps_per_second": 0.379,
       "eval_wer": 1.0,
-      "step": 7
+      "step": 14
     },
     {
-      "epoch": 4.0,
-      "learning_rate": 3.75e-05,
-      "loss": 13.7364,
-      "step": 8
+      "epoch": 8.0,
+      "learning_rate": 5.0625e-05,
+      "loss": 7.4746,
+      "step": 16
     },
     {
-      "epoch": 4.0,
-      "eval_loss": 25.79315185546875,
-      "eval_runtime": 5.4455,
-      "eval_samples_per_second": 23.506,
-      "eval_steps_per_second": 0.367,
+      "epoch": 8.0,
+      "eval_loss": 10.744585037231445,
+      "eval_runtime": 5.3374,
+      "eval_samples_per_second": 23.982,
+      "eval_steps_per_second": 0.375,
       "eval_wer": 1.0,
-      "step": 8
+      "step": 16
     },
     {
-      "epoch": 4.5,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 13.1602,
-      "step": 9
+      "epoch": 9.0,
+      "learning_rate": 4.6874999999999994e-05,
+      "loss": 6.8442,
+      "step": 18
     },
     {
-      "epoch": 4.5,
-      "eval_loss": 23.46338653564453,
-      "eval_runtime": 5.3714,
-      "eval_samples_per_second": 23.83,
-      "eval_steps_per_second": 0.372,
+      "epoch": 9.0,
+      "eval_loss": 9.794867515563965,
+      "eval_runtime": 5.2577,
+      "eval_samples_per_second": 24.345,
+      "eval_steps_per_second": 0.38,
       "eval_wer": 1.0,
-      "step": 9
+      "step": 18
     },
     {
-      "epoch": 5.0,
-      "learning_rate": 2.2499999999999998e-05,
-      "loss": 12.9439,
-      "step": 10
+      "epoch": 10.0,
+      "learning_rate": 4.312499999999999e-05,
+      "loss": 6.3765,
+      "step": 20
     },
     {
-      "epoch": 5.0,
-      "eval_loss": 21.93878936767578,
-      "eval_runtime": 5.4116,
-      "eval_samples_per_second": 23.653,
-      "eval_steps_per_second": 0.37,
+      "epoch": 10.0,
+      "eval_loss": 9.114768981933594,
+      "eval_runtime": 5.2827,
+      "eval_samples_per_second": 24.23,
+      "eval_steps_per_second": 0.379,
       "eval_wer": 1.0,
-      "step": 10
+      "step": 20
     },
     {
-      "epoch": 5.0,
-      "step": 10,
-      "total_flos": 1.3476444758728704e+17,
-      "train_loss": 16.66825189590454,
-      "train_runtime": 91.9274,
-      "train_samples_per_second": 6.962,
-      "train_steps_per_second": 0.109
+      "epoch": 11.0,
+      "learning_rate": 3.9374999999999995e-05,
+      "loss": 6.0321,
+      "step": 22
+    },
+    {
+      "epoch": 11.0,
+      "eval_loss": 8.59341812133789,
+      "eval_runtime": 5.2171,
+      "eval_samples_per_second": 24.535,
+      "eval_steps_per_second": 0.383,
+      "eval_wer": 1.0,
+      "step": 22
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.5625e-05,
+      "loss": 5.7783,
+      "step": 24
+    },
+    {
+      "epoch": 12.0,
+      "eval_loss": 8.198293685913086,
+      "eval_runtime": 5.3204,
+      "eval_samples_per_second": 24.058,
+      "eval_steps_per_second": 0.376,
+      "eval_wer": 1.0,
+      "step": 24
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 3.1874999999999996e-05,
+      "loss": 5.5827,
+      "step": 26
+    },
+    {
+      "epoch": 13.0,
+      "eval_loss": 7.877962112426758,
+      "eval_runtime": 5.2188,
+      "eval_samples_per_second": 24.527,
+      "eval_steps_per_second": 0.383,
+      "eval_wer": 1.0,
+      "step": 26
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.8125e-05,
+      "loss": 5.4249,
+      "step": 28
+    },
+    {
+      "epoch": 14.0,
+      "eval_loss": 7.628803730010986,
+      "eval_runtime": 5.2811,
+      "eval_samples_per_second": 24.238,
+      "eval_steps_per_second": 0.379,
+      "eval_wer": 1.0,
+      "step": 28
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 2.4375e-05,
+      "loss": 5.3088,
+      "step": 30
+    },
+    {
+      "epoch": 15.0,
+      "eval_loss": 7.427917003631592,
+      "eval_runtime": 5.236,
+      "eval_samples_per_second": 24.446,
+      "eval_steps_per_second": 0.382,
+      "eval_wer": 1.0,
+      "step": 30
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 2.0625e-05,
+      "loss": 5.2078,
+      "step": 32
+    },
+    {
+      "epoch": 16.0,
+      "eval_loss": 7.268764972686768,
+      "eval_runtime": 5.2941,
+      "eval_samples_per_second": 24.178,
+      "eval_steps_per_second": 0.378,
+      "eval_wer": 1.0,
+      "step": 32
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 1.6875e-05,
+      "loss": 5.1289,
+      "step": 34
+    },
+    {
+      "epoch": 17.0,
+      "eval_loss": 7.145933628082275,
+      "eval_runtime": 5.2685,
+      "eval_samples_per_second": 24.295,
+      "eval_steps_per_second": 0.38,
+      "eval_wer": 1.0,
+      "step": 34
+    },
+    {
+      "epoch": 18.0,
+      "learning_rate": 1.3124999999999999e-05,
+      "loss": 5.0697,
+      "step": 36
+    },
+    {
+      "epoch": 18.0,
+      "eval_loss": 7.052780628204346,
+      "eval_runtime": 5.263,
+      "eval_samples_per_second": 24.321,
+      "eval_steps_per_second": 0.38,
+      "eval_wer": 1.0,
+      "step": 36
+    },
+    {
+      "epoch": 19.0,
+      "learning_rate": 9.375e-06,
+      "loss": 5.0227,
+      "step": 38
+    },
+    {
+      "epoch": 19.0,
+      "eval_loss": 6.983470916748047,
+      "eval_runtime": 5.2829,
+      "eval_samples_per_second": 24.229,
+      "eval_steps_per_second": 0.379,
+      "eval_wer": 1.0,
+      "step": 38
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 5.6249999999999995e-06,
+      "loss": 4.9853,
+      "step": 40
+    },
+    {
+      "epoch": 20.0,
+      "eval_loss": 6.937458515167236,
+      "eval_runtime": 5.3165,
+      "eval_samples_per_second": 24.076,
+      "eval_steps_per_second": 0.376,
+      "eval_wer": 1.0,
+      "step": 40
+    },
+    {
+      "epoch": 20.0,
+      "step": 40,
+      "total_flos": 5.430583918308557e+17,
+      "train_loss": 8.69529299736023,
+      "train_runtime": 243.8197,
+      "train_samples_per_second": 10.5,
+      "train_steps_per_second": 0.164
     }
   ],
-  "max_steps": 10,
-  "num_train_epochs": 5,
-  "total_flos": 1.3476444758728704e+17,
+  "max_steps": 40,
+  "num_train_epochs": 20,
+  "total_flos": 5.430583918308557e+17,
   "trial_name": null,
   "trial_params": null
 }