End of training

all_results.json

@@ -1,14 +1,14 @@
 {
-    "epoch": 49.98,
-    "eval_loss": 2.8934524059295654,
-    "eval_runtime": 13.0771,
+    "epoch": 50.0,
+    "eval_loss": 0.9580045938491821,
+    "eval_runtime": 13.35,
     "eval_samples": 341,
-    "eval_samples_per_second": 26.076,
-    "eval_steps_per_second": 3.288,
-    "eval_wer": 0.9874538745387453,
-    "train_loss": 5.6772947265625,
-    "train_runtime": 2084.0288,
+    "eval_samples_per_second": 25.543,
+    "eval_steps_per_second": 3.221,
+    "eval_wer": 0.652029520295203,
+    "train_loss": 2.3068374424354703,
+    "train_runtime": 2918.903,
     "train_samples": 810,
-    "train_samples_per_second": 19.434,
-    "train_steps_per_second": 0.6
+    "train_samples_per_second": 13.875,
+    "train_steps_per_second": 0.874
 }

eval_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 49.98,
-    "eval_loss": 2.8934524059295654,
-    "eval_runtime": 13.0771,
+    "epoch": 50.0,
+    "eval_loss": 0.9580045938491821,
+    "eval_runtime": 13.35,
     "eval_samples": 341,
-    "eval_samples_per_second": 26.076,
-    "eval_steps_per_second": 3.288,
-    "eval_wer": 0.9874538745387453
+    "eval_samples_per_second": 25.543,
+    "eval_steps_per_second": 3.221,
+    "eval_wer": 0.652029520295203
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 49.98,
-    "train_loss": 5.6772947265625,
-    "train_runtime": 2084.0288,
+    "epoch": 50.0,
+    "train_loss": 2.3068374424354703,
+    "train_runtime": 2918.903,
     "train_samples": 810,
-    "train_samples_per_second": 19.434,
-    "train_steps_per_second": 0.6
+    "train_samples_per_second": 13.875,
+    "train_steps_per_second": 0.874
 }

trainer_state.json

@@ -1,115 +1,400 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 49.98039215686274,
-  "global_step": 1250,
+  "epoch": 50.0,
+  "global_step": 2550,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 3.98,
-      "learning_rate": 3.675e-06,
-      "loss": 20.0794,
+      "epoch": 1.96,
+      "learning_rate": 7.358999999999999e-05,
+      "loss": 9.5036,
       "step": 100
     },
     {
-      "epoch": 7.98,
-      "learning_rate": 7.425e-06,
-      "loss": 10.5776,
+      "epoch": 1.96,
+      "eval_loss": 4.053760528564453,
+      "eval_runtime": 13.4693,
+      "eval_samples_per_second": 25.317,
+      "eval_steps_per_second": 3.192,
+      "eval_wer": 1.0,
+      "step": 100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 7.059e-05,
+      "loss": 3.3669,
       "step": 200
     },
     {
-      "epoch": 11.98,
-      "learning_rate": 1.1174999999999999e-05,
-      "loss": 6.6033,
+      "epoch": 3.92,
+      "eval_loss": 3.204110622406006,
+      "eval_runtime": 13.5228,
+      "eval_samples_per_second": 25.217,
+      "eval_steps_per_second": 3.18,
+      "eval_wer": 1.0,
+      "step": 200
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 6.759e-05,
+      "loss": 3.1499,
+      "step": 300
+    },
+    {
+      "epoch": 5.88,
+      "eval_loss": 3.1220219135284424,
+      "eval_runtime": 13.3359,
+      "eval_samples_per_second": 25.57,
+      "eval_steps_per_second": 3.224,
+      "eval_wer": 1.0,
       "step": 300
     },
     {
-      "epoch": 15.98,
-      "learning_rate": 1.4925e-05,
-      "loss": 5.3857,
+      "epoch": 7.84,
+      "learning_rate": 6.458999999999999e-05,
+      "loss": 3.0271,
+      "step": 400
+    },
+    {
+      "epoch": 7.84,
+      "eval_loss": 2.9935410022735596,
+      "eval_runtime": 13.4609,
+      "eval_samples_per_second": 25.333,
+      "eval_steps_per_second": 3.194,
+      "eval_wer": 0.9970479704797048,
       "step": 400
     },
     {
-      "epoch": 19.98,
-      "learning_rate": 1.8675e-05,
-      "loss": 4.4431,
+      "epoch": 9.8,
+      "learning_rate": 6.159e-05,
+      "loss": 2.9565,
       "step": 500
     },
     {
-      "epoch": 19.98,
-      "eval_loss": 4.144110202789307,
-      "eval_runtime": 12.9526,
-      "eval_samples_per_second": 26.327,
-      "eval_steps_per_second": 3.32,
-      "eval_wer": 1.0,
+      "epoch": 9.8,
+      "eval_loss": 2.935685157775879,
+      "eval_runtime": 13.2089,
+      "eval_samples_per_second": 25.816,
+      "eval_steps_per_second": 3.255,
+      "eval_wer": 0.9992619926199262,
       "step": 500
     },
     {
-      "epoch": 23.98,
-      "learning_rate": 2.2424999999999996e-05,
-      "loss": 3.7383,
+      "epoch": 11.76,
+      "learning_rate": 5.858999999999999e-05,
+      "loss": 2.9184,
       "step": 600
     },
     {
-      "epoch": 27.98,
-      "learning_rate": 2.6174999999999996e-05,
-      "loss": 3.361,
+      "epoch": 11.76,
+      "eval_loss": 2.916475296020508,
+      "eval_runtime": 13.2506,
+      "eval_samples_per_second": 25.735,
+      "eval_steps_per_second": 3.245,
+      "eval_wer": 0.996309963099631,
+      "step": 600
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 5.5589999999999995e-05,
+      "loss": 2.8832,
+      "step": 700
+    },
+    {
+      "epoch": 13.73,
+      "eval_loss": 2.8761985301971436,
+      "eval_runtime": 13.271,
+      "eval_samples_per_second": 25.695,
+      "eval_steps_per_second": 3.24,
+      "eval_wer": 0.9911439114391144,
       "step": 700
     },
     {
-      "epoch": 31.98,
-      "learning_rate": 2.9925e-05,
-      "loss": 3.2219,
+      "epoch": 15.69,
+      "learning_rate": 5.2589999999999996e-05,
+      "loss": 2.8407,
       "step": 800
     },
     {
-      "epoch": 35.98,
-      "learning_rate": 3.3675e-05,
-      "loss": 3.104,
+      "epoch": 15.69,
+      "eval_loss": 2.810206890106201,
+      "eval_runtime": 13.1264,
+      "eval_samples_per_second": 25.978,
+      "eval_steps_per_second": 3.276,
+      "eval_wer": 0.9970479704797048,
+      "step": 800
+    },
+    {
+      "epoch": 17.65,
+      "learning_rate": 4.959e-05,
+      "loss": 2.7007,
+      "step": 900
+    },
+    {
+      "epoch": 17.65,
+      "eval_loss": 2.436384439468384,
+      "eval_runtime": 13.0662,
+      "eval_samples_per_second": 26.098,
+      "eval_steps_per_second": 3.291,
+      "eval_wer": 0.996309963099631,
       "step": 900
     },
     {
-      "epoch": 39.98,
-      "learning_rate": 3.7424999999999995e-05,
-      "loss": 3.0399,
+      "epoch": 19.61,
+      "learning_rate": 4.658999999999999e-05,
+      "loss": 2.4206,
       "step": 1000
     },
     {
-      "epoch": 39.98,
-      "eval_loss": 2.9768528938293457,
-      "eval_runtime": 12.8843,
-      "eval_samples_per_second": 26.466,
-      "eval_steps_per_second": 3.337,
-      "eval_wer": 0.9988929889298893,
+      "epoch": 19.61,
+      "eval_loss": 1.985196590423584,
+      "eval_runtime": 13.5307,
+      "eval_samples_per_second": 25.202,
+      "eval_steps_per_second": 3.178,
+      "eval_wer": 0.9420664206642066,
       "step": 1000
     },
     {
-      "epoch": 43.98,
-      "learning_rate": 4.1175e-05,
-      "loss": 2.9893,
+      "epoch": 21.57,
+      "learning_rate": 4.359e-05,
+      "loss": 2.0699,
       "step": 1100
     },
     {
-      "epoch": 47.98,
-      "learning_rate": 4.4924999999999994e-05,
-      "loss": 2.953,
+      "epoch": 21.57,
+      "eval_loss": 1.4849445819854736,
+      "eval_runtime": 13.2138,
+      "eval_samples_per_second": 25.806,
+      "eval_steps_per_second": 3.254,
+      "eval_wer": 0.8343173431734318,
+      "step": 1100
+    },
+    {
+      "epoch": 23.53,
+      "learning_rate": 4.0589999999999996e-05,
+      "loss": 1.8311,
       "step": 1200
     },
     {
-      "epoch": 49.98,
-      "step": 1250,
-      "total_flos": 4.919672600088422e+18,
-      "train_loss": 5.6772947265625,
-      "train_runtime": 2084.0288,
-      "train_samples_per_second": 19.434,
-      "train_steps_per_second": 0.6
+      "epoch": 23.53,
+      "eval_loss": 1.3083971738815308,
+      "eval_runtime": 13.1253,
+      "eval_samples_per_second": 25.98,
+      "eval_steps_per_second": 3.276,
+      "eval_wer": 0.7800738007380074,
+      "step": 1200
+    },
+    {
+      "epoch": 25.49,
+      "learning_rate": 3.759e-05,
+      "loss": 1.7127,
+      "step": 1300
+    },
+    {
+      "epoch": 25.49,
+      "eval_loss": 1.203996181488037,
+      "eval_runtime": 14.0537,
+      "eval_samples_per_second": 24.264,
+      "eval_steps_per_second": 3.06,
+      "eval_wer": 0.744649446494465,
+      "step": 1300
+    },
+    {
+      "epoch": 27.45,
+      "learning_rate": 3.459e-05,
+      "loss": 1.6239,
+      "step": 1400
+    },
+    {
+      "epoch": 27.45,
+      "eval_loss": 1.1359221935272217,
+      "eval_runtime": 13.3447,
+      "eval_samples_per_second": 25.553,
+      "eval_steps_per_second": 3.222,
+      "eval_wer": 0.7280442804428044,
+      "step": 1400
+    },
+    {
+      "epoch": 29.41,
+      "learning_rate": 3.159e-05,
+      "loss": 1.5654,
+      "step": 1500
+    },
+    {
+      "epoch": 29.41,
+      "eval_loss": 1.068825364112854,
+      "eval_runtime": 13.1322,
+      "eval_samples_per_second": 25.967,
+      "eval_steps_per_second": 3.274,
+      "eval_wer": 0.7158671586715867,
+      "step": 1500
+    },
+    {
+      "epoch": 31.37,
+      "learning_rate": 2.8589999999999996e-05,
+      "loss": 1.4965,
+      "step": 1600
+    },
+    {
+      "epoch": 31.37,
+      "eval_loss": 1.052024245262146,
+      "eval_runtime": 13.0376,
+      "eval_samples_per_second": 26.155,
+      "eval_steps_per_second": 3.298,
+      "eval_wer": 0.6985239852398524,
+      "step": 1600
+    },
+    {
+      "epoch": 33.33,
+      "learning_rate": 2.5589999999999997e-05,
+      "loss": 1.445,
+      "step": 1700
+    },
+    {
+      "epoch": 33.33,
+      "eval_loss": 1.0314019918441772,
+      "eval_runtime": 13.256,
+      "eval_samples_per_second": 25.724,
+      "eval_steps_per_second": 3.244,
+      "eval_wer": 0.6878228782287823,
+      "step": 1700
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 2.259e-05,
+      "loss": 1.4095,
+      "step": 1800
+    },
+    {
+      "epoch": 35.29,
+      "eval_loss": 1.0063327550888062,
+      "eval_runtime": 13.1463,
+      "eval_samples_per_second": 25.939,
+      "eval_steps_per_second": 3.271,
+      "eval_wer": 0.6712177121771218,
+      "step": 1800
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 1.9589999999999997e-05,
+      "loss": 1.3853,
+      "step": 1900
+    },
+    {
+      "epoch": 37.25,
+      "eval_loss": 0.9847874045372009,
+      "eval_runtime": 13.6657,
+      "eval_samples_per_second": 24.953,
+      "eval_steps_per_second": 3.147,
+      "eval_wer": 0.6701107011070111,
+      "step": 1900
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 1.659e-05,
+      "loss": 1.3558,
+      "step": 2000
+    },
+    {
+      "epoch": 39.22,
+      "eval_loss": 0.9738264679908752,
+      "eval_runtime": 13.0206,
+      "eval_samples_per_second": 26.189,
+      "eval_steps_per_second": 3.302,
+      "eval_wer": 0.6730627306273063,
+      "step": 2000
+    },
+    {
+      "epoch": 41.18,
+      "learning_rate": 1.3589999999999999e-05,
+      "loss": 1.3415,
+      "step": 2100
+    },
+    {
+      "epoch": 41.18,
+      "eval_loss": 0.9656162858009338,
+      "eval_runtime": 13.0695,
+      "eval_samples_per_second": 26.091,
+      "eval_steps_per_second": 3.29,
+      "eval_wer": 0.6645756457564576,
+      "step": 2100
+    },
+    {
+      "epoch": 43.14,
+      "learning_rate": 1.0589999999999999e-05,
+      "loss": 1.3102,
+      "step": 2200
+    },
+    {
+      "epoch": 43.14,
+      "eval_loss": 0.9631739854812622,
+      "eval_runtime": 13.1258,
+      "eval_samples_per_second": 25.979,
+      "eval_steps_per_second": 3.276,
+      "eval_wer": 0.655719557195572,
+      "step": 2200
+    },
+    {
+      "epoch": 45.1,
+      "learning_rate": 7.589999999999999e-06,
+      "loss": 1.309,
+      "step": 2300
+    },
+    {
+      "epoch": 45.1,
+      "eval_loss": 0.9495635032653809,
+      "eval_runtime": 12.963,
+      "eval_samples_per_second": 26.306,
+      "eval_steps_per_second": 3.317,
+      "eval_wer": 0.655719557195572,
+      "step": 2300
+    },
+    {
+      "epoch": 47.06,
+      "learning_rate": 4.589999999999999e-06,
+      "loss": 1.2993,
+      "step": 2400
+    },
+    {
+      "epoch": 47.06,
+      "eval_loss": 0.960854709148407,
+      "eval_runtime": 13.0799,
+      "eval_samples_per_second": 26.071,
+      "eval_steps_per_second": 3.287,
+      "eval_wer": 0.6549815498154982,
+      "step": 2400
+    },
+    {
+      "epoch": 49.02,
+      "learning_rate": 1.5899999999999998e-06,
+      "loss": 1.2695,
+      "step": 2500
+    },
+    {
+      "epoch": 49.02,
+      "eval_loss": 0.9603787660598755,
+      "eval_runtime": 13.0779,
+      "eval_samples_per_second": 26.075,
+      "eval_steps_per_second": 3.288,
+      "eval_wer": 0.6542435424354244,
+      "step": 2500
+    },
+    {
+      "epoch": 50.0,
+      "step": 2550,
+      "total_flos": 4.920553620909019e+18,
+      "train_loss": 2.3068374424354703,
+      "train_runtime": 2918.903,
+      "train_samples_per_second": 13.875,
+      "train_steps_per_second": 0.874
     }
   ],
-  "max_steps": 1250,
+  "max_steps": 2550,
   "num_train_epochs": 50,
-  "total_flos": 4.919672600088422e+18,
+  "total_flos": 4.920553620909019e+18,
   "trial_name": null,
   "trial_params": null
 }