End of training

README.md

@@ -4,11 +4,23 @@ license: apache-2.0
 base_model: openai/whisper-medium
 tags:
 - generated_from_trainer
+datasets:
+- nyagen
 metrics:
 - wer
 model-index:
 - name: whisper-medium-nyagen-balanced-model
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: nyagen
+      type: nyagen
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.23363711681855842
 ---
 
 <!-- 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-nyagen-balanced-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 nyagen dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.3774
-- Wer: 0.1984
+- Loss: 0.3438
+- Wer: 0.2336
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 7.529014844804318,
+    "eval_loss": 0.3438054919242859,
+    "eval_runtime": 121.592,
+    "eval_samples": 177,
+    "eval_samples_per_second": 1.456,
+    "eval_steps_per_second": 0.732,
+    "eval_wer": 0.23363711681855842,
+    "total_flos": 1.138076532375552e+19,
+    "train_loss": 1.1229147970676423,
+    "train_runtime": 3640.8906,
+    "train_samples": 1481,
+    "train_samples_per_second": 10.986,
+    "train_steps_per_second": 1.373
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.529014844804318,
+    "eval_loss": 0.3438054919242859,
+    "eval_runtime": 121.592,
+    "eval_samples": 177,
+    "eval_samples_per_second": 1.456,
+    "eval_steps_per_second": 0.732,
+    "eval_wer": 0.23363711681855842
+}

runs/Jan06_00-25-30_srvrocgpu011.uct.ac.za/events.out.tfevents.1736120005.srvrocgpu011.uct.ac.za

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4dc6fd046566428767334e38e7642d87ec1557d74c9c68f766b959a66ffc5ee2
+size 40

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 7.529014844804318,
+    "total_flos": 1.138076532375552e+19,
+    "train_loss": 1.1229147970676423,
+    "train_runtime": 3640.8906,
+    "train_samples": 1481,
+    "train_samples_per_second": 10.986,
+    "train_steps_per_second": 1.373
+}

trainer_state.json

@@ -0,0 +1,506 @@
+{
+  "best_metric": 0.3438054919242859,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/whisper-medium-nyagen-balanced-model/checkpoint-800",
+  "epoch": 7.529014844804318,
+  "eval_steps": 200,
+  "global_step": 1400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.1349527665317139,
+      "grad_norm": 112.41606140136719,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 10.7638,
+      "step": 25
+    },
+    {
+      "epoch": 0.2699055330634278,
+      "grad_norm": 71.51460266113281,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 8.7381,
+      "step": 50
+    },
+    {
+      "epoch": 0.4048582995951417,
+      "grad_norm": 69.4509048461914,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 6.4961,
+      "step": 75
+    },
+    {
+      "epoch": 0.5398110661268556,
+      "grad_norm": 62.58184051513672,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 4.4786,
+      "step": 100
+    },
+    {
+      "epoch": 0.6747638326585695,
+      "grad_norm": 41.15999984741211,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.5462,
+      "step": 125
+    },
+    {
+      "epoch": 0.8097165991902834,
+      "grad_norm": 43.67110061645508,
+      "learning_rate": 2.9e-06,
+      "loss": 2.6498,
+      "step": 150
+    },
+    {
+      "epoch": 0.9446693657219973,
+      "grad_norm": 38.462486267089844,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.3941,
+      "step": 175
+    },
+    {
+      "epoch": 1.0755735492577598,
+      "grad_norm": 38.60875701904297,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 1.8419,
+      "step": 200
+    },
+    {
+      "epoch": 1.0755735492577598,
+      "eval_loss": 0.5249760746955872,
+      "eval_runtime": 125.5624,
+      "eval_samples_per_second": 1.41,
+      "eval_steps_per_second": 0.709,
+      "eval_wer": 0.3744821872410936,
+      "step": 200
+    },
+    {
+      "epoch": 1.2105263157894737,
+      "grad_norm": 38.056907653808594,
+      "learning_rate": 4.4e-06,
+      "loss": 1.8181,
+      "step": 225
+    },
+    {
+      "epoch": 1.3454790823211875,
+      "grad_norm": 55.8536376953125,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.6732,
+      "step": 250
+    },
+    {
+      "epoch": 1.4804318488529016,
+      "grad_norm": 26.30344581604004,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 1.4496,
+      "step": 275
+    },
+    {
+      "epoch": 1.6153846153846154,
+      "grad_norm": 41.35667419433594,
+      "learning_rate": 5.9e-06,
+      "loss": 1.5242,
+      "step": 300
+    },
+    {
+      "epoch": 1.7503373819163293,
+      "grad_norm": 27.333097457885742,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.3668,
+      "step": 325
+    },
+    {
+      "epoch": 1.885290148448043,
+      "grad_norm": 43.94708251953125,
+      "learning_rate": 6.9e-06,
+      "loss": 1.5826,
+      "step": 350
+    },
+    {
+      "epoch": 2.016194331983806,
+      "grad_norm": 19.781164169311523,
+      "learning_rate": 7.4e-06,
+      "loss": 1.1724,
+      "step": 375
+    },
+    {
+      "epoch": 2.1511470985155197,
+      "grad_norm": 21.915298461914062,
+      "learning_rate": 7.9e-06,
+      "loss": 0.6731,
+      "step": 400
+    },
+    {
+      "epoch": 2.1511470985155197,
+      "eval_loss": 0.37383848428726196,
+      "eval_runtime": 125.8509,
+      "eval_samples_per_second": 1.406,
+      "eval_steps_per_second": 0.707,
+      "eval_wer": 0.27009113504556753,
+      "step": 400
+    },
+    {
+      "epoch": 2.2860998650472335,
+      "grad_norm": 21.964094161987305,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.6603,
+      "step": 425
+    },
+    {
+      "epoch": 2.4210526315789473,
+      "grad_norm": 19.82171630859375,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.7134,
+      "step": 450
+    },
+    {
+      "epoch": 2.556005398110661,
+      "grad_norm": 18.731990814208984,
+      "learning_rate": 9.4e-06,
+      "loss": 0.7317,
+      "step": 475
+    },
+    {
+      "epoch": 2.690958164642375,
+      "grad_norm": 29.455059051513672,
+      "learning_rate": 9.9e-06,
+      "loss": 0.8073,
+      "step": 500
+    },
+    {
+      "epoch": 2.8259109311740893,
+      "grad_norm": 39.07381057739258,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 0.8497,
+      "step": 525
+    },
+    {
+      "epoch": 2.960863697705803,
+      "grad_norm": 20.394155502319336,
+      "learning_rate": 9.9e-06,
+      "loss": 0.8654,
+      "step": 550
+    },
+    {
+      "epoch": 3.0917678812415654,
+      "grad_norm": 14.952836036682129,
+      "learning_rate": 9.844444444444446e-06,
+      "loss": 0.4818,
+      "step": 575
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "grad_norm": 14.173761367797852,
+      "learning_rate": 9.78888888888889e-06,
+      "loss": 0.3159,
+      "step": 600
+    },
+    {
+      "epoch": 3.2267206477732793,
+      "eval_loss": 0.353611558675766,
+      "eval_runtime": 123.7082,
+      "eval_samples_per_second": 1.431,
+      "eval_steps_per_second": 0.719,
+      "eval_wer": 0.2448218724109362,
+      "step": 600
+    },
+    {
+      "epoch": 3.361673414304993,
+      "grad_norm": 17.980554580688477,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.4002,
+      "step": 625
+    },
+    {
+      "epoch": 3.4966261808367074,
+      "grad_norm": 24.047901153564453,
+      "learning_rate": 9.677777777777778e-06,
+      "loss": 0.366,
+      "step": 650
+    },
+    {
+      "epoch": 3.6315789473684212,
+      "grad_norm": 16.637474060058594,
+      "learning_rate": 9.622222222222222e-06,
+      "loss": 0.3679,
+      "step": 675
+    },
+    {
+      "epoch": 3.766531713900135,
+      "grad_norm": 10.418351173400879,
+      "learning_rate": 9.566666666666668e-06,
+      "loss": 0.3904,
+      "step": 700
+    },
+    {
+      "epoch": 3.901484480431849,
+      "grad_norm": 16.89838981628418,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 0.4296,
+      "step": 725
+    },
+    {
+      "epoch": 4.032388663967612,
+      "grad_norm": 24.11263084411621,
+      "learning_rate": 9.455555555555557e-06,
+      "loss": 0.3219,
+      "step": 750
+    },
+    {
+      "epoch": 4.1673414304993255,
+      "grad_norm": 13.350052833557129,
+      "learning_rate": 9.4e-06,
+      "loss": 0.1524,
+      "step": 775
+    },
+    {
+      "epoch": 4.302294197031039,
+      "grad_norm": 8.87996768951416,
+      "learning_rate": 9.344444444444446e-06,
+      "loss": 0.2231,
+      "step": 800
+    },
+    {
+      "epoch": 4.302294197031039,
+      "eval_loss": 0.3438054919242859,
+      "eval_runtime": 121.9867,
+      "eval_samples_per_second": 1.451,
+      "eval_steps_per_second": 0.73,
+      "eval_wer": 0.23363711681855842,
+      "step": 800
+    },
+    {
+      "epoch": 4.437246963562753,
+      "grad_norm": 11.531867027282715,
+      "learning_rate": 9.28888888888889e-06,
+      "loss": 0.24,
+      "step": 825
+    },
+    {
+      "epoch": 4.572199730094467,
+      "grad_norm": 11.15186882019043,
+      "learning_rate": 9.233333333333334e-06,
+      "loss": 0.2351,
+      "step": 850
+    },
+    {
+      "epoch": 4.707152496626181,
+      "grad_norm": 20.399194717407227,
+      "learning_rate": 9.17777777777778e-06,
+      "loss": 0.1775,
+      "step": 875
+    },
+    {
+      "epoch": 4.842105263157895,
+      "grad_norm": 14.302651405334473,
+      "learning_rate": 9.122222222222223e-06,
+      "loss": 0.2056,
+      "step": 900
+    },
+    {
+      "epoch": 4.9770580296896085,
+      "grad_norm": 14.608598709106445,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.1961,
+      "step": 925
+    },
+    {
+      "epoch": 5.107962213225371,
+      "grad_norm": 16.94548225402832,
+      "learning_rate": 9.011111111111111e-06,
+      "loss": 0.1131,
+      "step": 950
+    },
+    {
+      "epoch": 5.242914979757085,
+      "grad_norm": 8.520947456359863,
+      "learning_rate": 8.955555555555555e-06,
+      "loss": 0.1135,
+      "step": 975
+    },
+    {
+      "epoch": 5.377867746288799,
+      "grad_norm": 14.151520729064941,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.1319,
+      "step": 1000
+    },
+    {
+      "epoch": 5.377867746288799,
+      "eval_loss": 0.35178500413894653,
+      "eval_runtime": 122.4778,
+      "eval_samples_per_second": 1.445,
+      "eval_steps_per_second": 0.727,
+      "eval_wer": 0.21830985915492956,
+      "step": 1000
+    },
+    {
+      "epoch": 5.512820512820513,
+      "grad_norm": 7.279934883117676,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 0.0928,
+      "step": 1025
+    },
+    {
+      "epoch": 5.647773279352227,
+      "grad_norm": 3.711681604385376,
+      "learning_rate": 8.788888888888891e-06,
+      "loss": 0.1271,
+      "step": 1050
+    },
+    {
+      "epoch": 5.7827260458839405,
+      "grad_norm": 12.580814361572266,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 0.096,
+      "step": 1075
+    },
+    {
+      "epoch": 5.917678812415654,
+      "grad_norm": 12.59216594696045,
+      "learning_rate": 8.677777777777779e-06,
+      "loss": 0.1243,
+      "step": 1100
+    },
+    {
+      "epoch": 6.048582995951417,
+      "grad_norm": 8.682173728942871,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 0.0811,
+      "step": 1125
+    },
+    {
+      "epoch": 6.183535762483131,
+      "grad_norm": 11.549365043640137,
+      "learning_rate": 8.566666666666667e-06,
+      "loss": 0.049,
+      "step": 1150
+    },
+    {
+      "epoch": 6.318488529014845,
+      "grad_norm": 4.770676612854004,
+      "learning_rate": 8.511111111111113e-06,
+      "loss": 0.0864,
+      "step": 1175
+    },
+    {
+      "epoch": 6.4534412955465585,
+      "grad_norm": 1.4411147832870483,
+      "learning_rate": 8.455555555555555e-06,
+      "loss": 0.0553,
+      "step": 1200
+    },
+    {
+      "epoch": 6.4534412955465585,
+      "eval_loss": 0.36358916759490967,
+      "eval_runtime": 126.3553,
+      "eval_samples_per_second": 1.401,
+      "eval_steps_per_second": 0.704,
+      "eval_wer": 0.29536039768019884,
+      "step": 1200
+    },
+    {
+      "epoch": 6.588394062078272,
+      "grad_norm": 9.254927635192871,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.075,
+      "step": 1225
+    },
+    {
+      "epoch": 6.723346828609986,
+      "grad_norm": 4.376775741577148,
+      "learning_rate": 8.344444444444445e-06,
+      "loss": 0.0959,
+      "step": 1250
+    },
+    {
+      "epoch": 6.8582995951417,
+      "grad_norm": 9.643482208251953,
+      "learning_rate": 8.288888888888889e-06,
+      "loss": 0.0829,
+      "step": 1275
+    },
+    {
+      "epoch": 6.993252361673415,
+      "grad_norm": 6.23452615737915,
+      "learning_rate": 8.233333333333335e-06,
+      "loss": 0.0654,
+      "step": 1300
+    },
+    {
+      "epoch": 7.124156545209177,
+      "grad_norm": 8.643274307250977,
+      "learning_rate": 8.177777777777779e-06,
+      "loss": 0.0403,
+      "step": 1325
+    },
+    {
+      "epoch": 7.2591093117408905,
+      "grad_norm": 9.578027725219727,
+      "learning_rate": 8.122222222222223e-06,
+      "loss": 0.0361,
+      "step": 1350
+    },
+    {
+      "epoch": 7.394062078272604,
+      "grad_norm": 19.954702377319336,
+      "learning_rate": 8.066666666666667e-06,
+      "loss": 0.0551,
+      "step": 1375
+    },
+    {
+      "epoch": 7.529014844804318,
+      "grad_norm": 3.9316067695617676,
+      "learning_rate": 8.011111111111113e-06,
+      "loss": 0.062,
+      "step": 1400
+    },
+    {
+      "epoch": 7.529014844804318,
+      "eval_loss": 0.37738800048828125,
+      "eval_runtime": 122.1805,
+      "eval_samples_per_second": 1.449,
+      "eval_steps_per_second": 0.728,
+      "eval_wer": 0.1984258492129246,
+      "step": 1400
+    },
+    {
+      "epoch": 7.529014844804318,
+      "step": 1400,
+      "total_flos": 1.138076532375552e+19,
+      "train_loss": 1.1229147970676423,
+      "train_runtime": 3640.8906,
+      "train_samples_per_second": 10.986,
+      "train_steps_per_second": 1.373
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 28,
+  "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": 1.138076532375552e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}