End of training

README.md

@@ -4,11 +4,23 @@ license: apache-2.0
 base_model: openai/whisper-medium
 tags:
 - generated_from_trainer
+datasets:
+- lozgen
 metrics:
 - wer
 model-index:
 - name: whisper-medium-lozgen-male-model
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: lozgen
+      type: lozgen
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.4796960341961529
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -16,10 +28,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # whisper-medium-lozgen-male-model
 
-This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on an unknown dataset.
+This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the lozgen dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.8436
-- Wer: 0.3887
+- Loss: 0.7984
+- Wer: 0.4797
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 11.940298507462687,
+    "eval_loss": 0.7984297275543213,
+    "eval_runtime": 124.2218,
+    "eval_samples": 310,
+    "eval_samples_per_second": 2.496,
+    "eval_steps_per_second": 0.628,
+    "eval_wer": 0.4796960341961529,
+    "total_flos": 6.531871408128e+18,
+    "train_loss": 0.8537300661206245,
+    "train_runtime": 1916.7552,
+    "train_samples": 536,
+    "train_samples_per_second": 20.869,
+    "train_steps_per_second": 2.609
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 11.940298507462687,
+    "eval_loss": 0.7984297275543213,
+    "eval_runtime": 124.2218,
+    "eval_samples": 310,
+    "eval_samples_per_second": 2.496,
+    "eval_steps_per_second": 0.628,
+    "eval_wer": 0.4796960341961529
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 11.940298507462687,
+    "total_flos": 6.531871408128e+18,
+    "train_loss": 0.8537300661206245,
+    "train_runtime": 1916.7552,
+    "train_samples": 536,
+    "train_samples_per_second": 20.869,
+    "train_steps_per_second": 2.609
+}

trainer_state.json

@@ -0,0 +1,311 @@
+{
+  "best_metric": 0.7984297275543213,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/whisper-medium-lozgen-male-model/checkpoint-200",
+  "epoch": 11.940298507462687,
+  "eval_steps": 200,
+  "global_step": 800,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.373134328358209,
+      "grad_norm": 67.36864471435547,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 5.9515,
+      "step": 25
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 50.873260498046875,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 4.7813,
+      "step": 50
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 34.35158157348633,
+      "learning_rate": 1.42e-06,
+      "loss": 3.468,
+      "step": 75
+    },
+    {
+      "epoch": 1.4925373134328357,
+      "grad_norm": 42.986968994140625,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 2.385,
+      "step": 100
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 29.906461715698242,
+      "learning_rate": 2.42e-06,
+      "loss": 2.0003,
+      "step": 125
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 22.721763610839844,
+      "learning_rate": 2.92e-06,
+      "loss": 1.5359,
+      "step": 150
+    },
+    {
+      "epoch": 2.611940298507463,
+      "grad_norm": 23.2641658782959,
+      "learning_rate": 3.4200000000000007e-06,
+      "loss": 1.2475,
+      "step": 175
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 26.96329116821289,
+      "learning_rate": 3.920000000000001e-06,
+      "loss": 1.2961,
+      "step": 200
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "eval_loss": 0.7984297275543213,
+      "eval_runtime": 127.2992,
+      "eval_samples_per_second": 2.435,
+      "eval_steps_per_second": 0.613,
+      "eval_wer": 0.4796960341961529,
+      "step": 200
+    },
+    {
+      "epoch": 3.3582089552238807,
+      "grad_norm": 12.527202606201172,
+      "learning_rate": 4.42e-06,
+      "loss": 0.7541,
+      "step": 225
+    },
+    {
+      "epoch": 3.7313432835820897,
+      "grad_norm": 21.142616271972656,
+      "learning_rate": 4.92e-06,
+      "loss": 0.7149,
+      "step": 250
+    },
+    {
+      "epoch": 4.104477611940299,
+      "grad_norm": 9.611239433288574,
+      "learning_rate": 5.420000000000001e-06,
+      "loss": 0.5993,
+      "step": 275
+    },
+    {
+      "epoch": 4.477611940298507,
+      "grad_norm": 12.687724113464355,
+      "learning_rate": 5.92e-06,
+      "loss": 0.3943,
+      "step": 300
+    },
+    {
+      "epoch": 4.850746268656716,
+      "grad_norm": 19.18768310546875,
+      "learning_rate": 6.42e-06,
+      "loss": 0.3778,
+      "step": 325
+    },
+    {
+      "epoch": 5.223880597014926,
+      "grad_norm": 15.153021812438965,
+      "learning_rate": 6.92e-06,
+      "loss": 0.208,
+      "step": 350
+    },
+    {
+      "epoch": 5.597014925373134,
+      "grad_norm": 10.51365852355957,
+      "learning_rate": 7.420000000000001e-06,
+      "loss": 0.1744,
+      "step": 375
+    },
+    {
+      "epoch": 5.970149253731344,
+      "grad_norm": 9.576289176940918,
+      "learning_rate": 7.92e-06,
+      "loss": 0.2221,
+      "step": 400
+    },
+    {
+      "epoch": 5.970149253731344,
+      "eval_loss": 0.8183491230010986,
+      "eval_runtime": 129.9768,
+      "eval_samples_per_second": 2.385,
+      "eval_steps_per_second": 0.6,
+      "eval_wer": 0.4540489194965566,
+      "step": 400
+    },
+    {
+      "epoch": 6.343283582089552,
+      "grad_norm": 20.339366912841797,
+      "learning_rate": 8.42e-06,
+      "loss": 0.12,
+      "step": 425
+    },
+    {
+      "epoch": 6.7164179104477615,
+      "grad_norm": 6.8028669357299805,
+      "learning_rate": 8.920000000000001e-06,
+      "loss": 0.1084,
+      "step": 450
+    },
+    {
+      "epoch": 7.08955223880597,
+      "grad_norm": 16.080068588256836,
+      "learning_rate": 9.42e-06,
+      "loss": 0.1097,
+      "step": 475
+    },
+    {
+      "epoch": 7.462686567164179,
+      "grad_norm": 6.925537109375,
+      "learning_rate": 9.920000000000002e-06,
+      "loss": 0.0899,
+      "step": 500
+    },
+    {
+      "epoch": 7.835820895522388,
+      "grad_norm": 15.699053764343262,
+      "learning_rate": 9.953333333333333e-06,
+      "loss": 0.1049,
+      "step": 525
+    },
+    {
+      "epoch": 8.208955223880597,
+      "grad_norm": 10.956867218017578,
+      "learning_rate": 9.89777777777778e-06,
+      "loss": 0.0883,
+      "step": 550
+    },
+    {
+      "epoch": 8.582089552238806,
+      "grad_norm": 10.130331993103027,
+      "learning_rate": 9.842222222222223e-06,
+      "loss": 0.0952,
+      "step": 575
+    },
+    {
+      "epoch": 8.955223880597014,
+      "grad_norm": 16.053123474121094,
+      "learning_rate": 9.786666666666667e-06,
+      "loss": 0.0963,
+      "step": 600
+    },
+    {
+      "epoch": 8.955223880597014,
+      "eval_loss": 0.8390884399414062,
+      "eval_runtime": 124.2772,
+      "eval_samples_per_second": 2.494,
+      "eval_steps_per_second": 0.628,
+      "eval_wer": 0.3889812396105438,
+      "step": 600
+    },
+    {
+      "epoch": 9.328358208955224,
+      "grad_norm": 5.976753234863281,
+      "learning_rate": 9.731111111111113e-06,
+      "loss": 0.0706,
+      "step": 625
+    },
+    {
+      "epoch": 9.701492537313433,
+      "grad_norm": 4.946258544921875,
+      "learning_rate": 9.675555555555555e-06,
+      "loss": 0.0533,
+      "step": 650
+    },
+    {
+      "epoch": 10.074626865671641,
+      "grad_norm": 2.1235852241516113,
+      "learning_rate": 9.620000000000001e-06,
+      "loss": 0.0535,
+      "step": 675
+    },
+    {
+      "epoch": 10.447761194029852,
+      "grad_norm": 6.762497425079346,
+      "learning_rate": 9.564444444444445e-06,
+      "loss": 0.0426,
+      "step": 700
+    },
+    {
+      "epoch": 10.82089552238806,
+      "grad_norm": 10.996703147888184,
+      "learning_rate": 9.508888888888889e-06,
+      "loss": 0.037,
+      "step": 725
+    },
+    {
+      "epoch": 11.194029850746269,
+      "grad_norm": 11.038252830505371,
+      "learning_rate": 9.453333333333335e-06,
+      "loss": 0.0537,
+      "step": 750
+    },
+    {
+      "epoch": 11.567164179104477,
+      "grad_norm": 5.437051296234131,
+      "learning_rate": 9.397777777777779e-06,
+      "loss": 0.0398,
+      "step": 775
+    },
+    {
+      "epoch": 11.940298507462687,
+      "grad_norm": 39.8909797668457,
+      "learning_rate": 9.342222222222223e-06,
+      "loss": 0.0457,
+      "step": 800
+    },
+    {
+      "epoch": 11.940298507462687,
+      "eval_loss": 0.8435798287391663,
+      "eval_runtime": 129.9146,
+      "eval_samples_per_second": 2.386,
+      "eval_steps_per_second": 0.6,
+      "eval_wer": 0.3887437663262883,
+      "step": 800
+    },
+    {
+      "epoch": 11.940298507462687,
+      "step": 800,
+      "total_flos": 6.531871408128e+18,
+      "train_loss": 0.8537300661206245,
+      "train_runtime": 1916.7552,
+      "train_samples_per_second": 20.869,
+      "train_steps_per_second": 2.609
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 75,
+  "save_steps": 200,
+  "stateful_callbacks": {
+    "EarlyStoppingCallback": {
+      "args": {
+        "early_stopping_patience": 3,
+        "early_stopping_threshold": 0.0
+      },
+      "attributes": {
+        "early_stopping_patience_counter": 3
+      }
+    },
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.531871408128e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}