End of training

README.md

@@ -3,6 +3,7 @@ library_name: transformers
 license: apache-2.0
 base_model: facebook/wav2vec2-base
 tags:
+- audio-classification
 - generated_from_trainer
 metrics:
 - accuracy
@@ -16,10 +17,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # wav2vec2-base-wakeword
 
-This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
+This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the superb dataset.
 It achieves the following results on the evaluation set:
-- Loss: 3.8197
-- Accuracy: 0.7174
+- Loss: 4.0217
+- Accuracy: 0.7473
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
-    "epoch": 5.0,
-    "eval_accuracy": 0.6140350877192983,
-    "eval_loss": 6.055716037750244,
-    "eval_runtime": 1.2768,
-    "eval_samples_per_second": 44.644,
-    "eval_steps_per_second": 1.566,
-    "total_flos": 2.286384817968e+16,
-    "train_loss": 1.5362668991088868,
-    "train_runtime": 34.37,
-    "train_samples_per_second": 73.174,
-    "train_steps_per_second": 0.145
+    "epoch": 4.916201117318436,
+    "eval_accuracy": 0.7472527472527473,
+    "eval_loss": 4.021728038787842,
+    "eval_runtime": 21.9494,
+    "eval_samples_per_second": 58.042,
+    "eval_steps_per_second": 1.822,
+    "total_flos": 2.59294349617536e+17,
+    "train_loss": 5.575903250954368,
+    "train_runtime": 352.7331,
+    "train_samples_per_second": 81.067,
+    "train_steps_per_second": 0.624
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
-    "eval_accuracy": 0.6140350877192983,
-    "eval_loss": 6.055716037750244,
-    "eval_runtime": 1.2768,
-    "eval_samples_per_second": 44.644,
-    "eval_steps_per_second": 1.566
+    "epoch": 4.916201117318436,
+    "eval_accuracy": 0.7472527472527473,
+    "eval_loss": 4.021728038787842,
+    "eval_runtime": 21.9494,
+    "eval_samples_per_second": 58.042,
+    "eval_steps_per_second": 1.822
 }

train_results.json

@@ -1,8 +1,8 @@
 {
-    "epoch": 5.0,
-    "total_flos": 2.286384817968e+16,
-    "train_loss": 1.5362668991088868,
-    "train_runtime": 34.37,
-    "train_samples_per_second": 73.174,
-    "train_steps_per_second": 0.145
+    "epoch": 4.916201117318436,
+    "total_flos": 2.59294349617536e+17,
+    "train_loss": 5.575903250954368,
+    "train_runtime": 352.7331,
+    "train_samples_per_second": 81.067,
+    "train_steps_per_second": 0.624
 }

trainer_state.json

@@ -1,70 +1,224 @@
 {
-  "best_metric": 0.6140350877192983,
-  "best_model_checkpoint": "wav2vec2-base-wakeword/checkpoint-5",
-  "epoch": 5.0,
+  "best_metric": 0.7472527472527473,
+  "best_model_checkpoint": "wav2vec2-base-wakeword/checkpoint-179",
+  "epoch": 4.916201117318436,
   "eval_steps": 500,
-  "global_step": 5,
+  "global_step": 220,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.0,
-      "eval_accuracy": 0.0,
-      "eval_loss": 6.203948020935059,
-      "eval_runtime": 1.1877,
-      "eval_samples_per_second": 47.993,
-      "eval_steps_per_second": 1.684,
-      "step": 1
+      "epoch": 0.22346368715083798,
+      "grad_norm": 1.7972055673599243,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 8.6174,
+      "step": 10
     },
     {
-      "epoch": 2.0,
-      "eval_accuracy": 0.017543859649122806,
-      "eval_loss": 6.150973320007324,
-      "eval_runtime": 1.1793,
-      "eval_samples_per_second": 48.332,
-      "eval_steps_per_second": 1.696,
-      "step": 2
+      "epoch": 0.44692737430167595,
+      "grad_norm": 3.6899046897888184,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 8.3156,
+      "step": 20
     },
     {
-      "epoch": 3.0,
-      "eval_accuracy": 0.43859649122807015,
-      "eval_loss": 6.095394611358643,
-      "eval_runtime": 1.2414,
-      "eval_samples_per_second": 45.915,
-      "eval_steps_per_second": 1.611,
-      "step": 3
+      "epoch": 0.6703910614525139,
+      "grad_norm": 4.899014949798584,
+      "learning_rate": 2.8787878787878788e-05,
+      "loss": 7.6827,
+      "step": 30
+    },
+    {
+      "epoch": 0.8938547486033519,
+      "grad_norm": 5.642251014709473,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 7.269,
+      "step": 40
+    },
+    {
+      "epoch": 0.9832402234636871,
+      "eval_accuracy": 0.5274725274725275,
+      "eval_loss": 6.90291690826416,
+      "eval_runtime": 20.6201,
+      "eval_samples_per_second": 61.784,
+      "eval_steps_per_second": 1.94,
+      "step": 44
+    },
+    {
+      "epoch": 1.1173184357541899,
+      "grad_norm": 6.348006248474121,
+      "learning_rate": 2.575757575757576e-05,
+      "loss": 6.9601,
+      "step": 50
+    },
+    {
+      "epoch": 1.3407821229050279,
+      "grad_norm": 6.931854248046875,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 6.6727,
+      "step": 60
+    },
+    {
+      "epoch": 1.564245810055866,
+      "grad_norm": 7.51934814453125,
+      "learning_rate": 2.272727272727273e-05,
+      "loss": 6.3984,
+      "step": 70
+    },
+    {
+      "epoch": 1.7877094972067038,
+      "grad_norm": 8.076024055480957,
+      "learning_rate": 2.121212121212121e-05,
+      "loss": 6.1201,
+      "step": 80
+    },
+    {
+      "epoch": 1.988826815642458,
+      "eval_accuracy": 0.5572998430141287,
+      "eval_loss": 5.679985046386719,
+      "eval_runtime": 21.9537,
+      "eval_samples_per_second": 58.031,
+      "eval_steps_per_second": 1.822,
+      "step": 89
+    },
+    {
+      "epoch": 2.011173184357542,
+      "grad_norm": 8.460003852844238,
+      "learning_rate": 1.9696969696969697e-05,
+      "loss": 5.8571,
+      "step": 90
+    },
+    {
+      "epoch": 2.2346368715083798,
+      "grad_norm": 8.846624374389648,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 5.5999,
+      "step": 100
+    },
+    {
+      "epoch": 2.458100558659218,
+      "grad_norm": 9.555537223815918,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 5.3511,
+      "step": 110
+    },
+    {
+      "epoch": 2.6815642458100557,
+      "grad_norm": 9.668717384338379,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 5.1268,
+      "step": 120
+    },
+    {
+      "epoch": 2.905027932960894,
+      "grad_norm": 9.954819679260254,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 4.9094,
+      "step": 130
+    },
+    {
+      "epoch": 2.994413407821229,
+      "eval_accuracy": 0.6797488226059655,
+      "eval_loss": 4.662096977233887,
+      "eval_runtime": 22.1355,
+      "eval_samples_per_second": 57.555,
+      "eval_steps_per_second": 1.807,
+      "step": 134
+    },
+    {
+      "epoch": 3.1284916201117317,
+      "grad_norm": 10.29443073272705,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 4.7082,
+      "step": 140
+    },
+    {
+      "epoch": 3.35195530726257,
+      "grad_norm": 10.487744331359863,
+      "learning_rate": 1.0606060606060606e-05,
+      "loss": 4.5347,
+      "step": 150
+    },
+    {
+      "epoch": 3.5754189944134076,
+      "grad_norm": 10.81801700592041,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 4.3854,
+      "step": 160
+    },
+    {
+      "epoch": 3.798882681564246,
+      "grad_norm": 10.992441177368164,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 4.2402,
+      "step": 170
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.5789473684210527,
-      "eval_loss": 6.068804740905762,
-      "eval_runtime": 1.2218,
-      "eval_samples_per_second": 46.654,
-      "eval_steps_per_second": 1.637,
-      "step": 4
-    },
-    {
-      "epoch": 5.0,
-      "eval_accuracy": 0.6140350877192983,
-      "eval_loss": 6.055716037750244,
-      "eval_runtime": 1.3159,
-      "eval_samples_per_second": 43.317,
-      "eval_steps_per_second": 1.52,
-      "step": 5
-    },
-    {
-      "epoch": 5.0,
-      "step": 5,
-      "total_flos": 2.286384817968e+16,
-      "train_loss": 1.5362668991088868,
-      "train_runtime": 34.37,
-      "train_samples_per_second": 73.174,
-      "train_steps_per_second": 0.145
+      "eval_accuracy": 0.7472527472527473,
+      "eval_loss": 4.021728038787842,
+      "eval_runtime": 21.4694,
+      "eval_samples_per_second": 59.34,
+      "eval_steps_per_second": 1.863,
+      "step": 179
+    },
+    {
+      "epoch": 4.022346368715084,
+      "grad_norm": 11.058788299560547,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 4.1213,
+      "step": 180
+    },
+    {
+      "epoch": 4.245810055865922,
+      "grad_norm": 11.229594230651855,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 4.0389,
+      "step": 190
+    },
+    {
+      "epoch": 4.4692737430167595,
+      "grad_norm": 11.344269752502441,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 3.9595,
+      "step": 200
+    },
+    {
+      "epoch": 4.692737430167598,
+      "grad_norm": 11.231822967529297,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 3.9156,
+      "step": 210
+    },
+    {
+      "epoch": 4.916201117318436,
+      "grad_norm": 11.280599594116211,
+      "learning_rate": 0.0,
+      "loss": 3.8857,
+      "step": 220
+    },
+    {
+      "epoch": 4.916201117318436,
+      "eval_accuracy": 0.717425431711146,
+      "eval_loss": 3.8197391033172607,
+      "eval_runtime": 21.9944,
+      "eval_samples_per_second": 57.924,
+      "eval_steps_per_second": 1.819,
+      "step": 220
+    },
+    {
+      "epoch": 4.916201117318436,
+      "step": 220,
+      "total_flos": 2.59294349617536e+17,
+      "train_loss": 5.575903250954368,
+      "train_runtime": 352.7331,
+      "train_samples_per_second": 81.067,
+      "train_steps_per_second": 0.624
     }
   ],
   "logging_steps": 10,
-  "max_steps": 5,
+  "max_steps": 220,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 5,
   "save_steps": 500,
@@ -80,8 +234,8 @@
       "attributes": {}
     }
   },
-  "total_flos": 2.286384817968e+16,
-  "train_batch_size": 512,
+  "total_flos": 2.59294349617536e+17,
+  "train_batch_size": 32,
   "trial_name": null,
   "trial_params": null
 }