End of training

README.md

@@ -4,11 +4,23 @@ license: apache-2.0
 base_model: openai/whisper-medium
 tags:
 - generated_from_trainer
+datasets:
+- toigen
 metrics:
 - wer
 model-index:
 - name: whisper-medium-toigen-combined-model
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: toigen
+      type: toigen
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.4497528830313015
 ---
 
 <!-- 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-toigen-combined-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 toigen dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7208
-- Wer: 0.3864
+- Loss: 0.6425
+- Wer: 0.4498
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 6.6059171597633135,
+    "eval_loss": 0.6425132155418396,
+    "eval_runtime": 253.9056,
+    "eval_samples": 392,
+    "eval_samples_per_second": 1.544,
+    "eval_steps_per_second": 0.772,
+    "eval_wer": 0.4497528830313015,
+    "total_flos": 1.138790955810816e+19,
+    "train_loss": 1.9424426797458103,
+    "train_runtime": 4520.3552,
+    "train_samples": 1689,
+    "train_samples_per_second": 11.209,
+    "train_steps_per_second": 1.4
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 6.6059171597633135,
+    "eval_loss": 0.6425132155418396,
+    "eval_runtime": 253.9056,
+    "eval_samples": 392,
+    "eval_samples_per_second": 1.544,
+    "eval_steps_per_second": 0.772,
+    "eval_wer": 0.4497528830313015
+}

runs/Jan05_20-15-26_srvrocgpu011.uct.ac.za/events.out.tfevents.1736105977.srvrocgpu011.uct.ac.za

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3a6686ce5c20a83557d8b7728c09481c0db8abd443b9200214745f048cae3471
+size 40

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 6.6059171597633135,
+    "total_flos": 1.138790955810816e+19,
+    "train_loss": 1.9424426797458103,
+    "train_runtime": 4520.3552,
+    "train_samples": 1689,
+    "train_samples_per_second": 11.209,
+    "train_steps_per_second": 1.4
+}

trainer_state.json

@@ -0,0 +1,506 @@
+{
+  "best_metric": 0.6425132155418396,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/whisper-medium-toigen-combined-model/checkpoint-800",
+  "epoch": 6.6059171597633135,
+  "eval_steps": 200,
+  "global_step": 1400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.11834319526627218,
+      "grad_norm": 120.00074768066406,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 13.7413,
+      "step": 25
+    },
+    {
+      "epoch": 0.23668639053254437,
+      "grad_norm": 93.74044799804688,
+      "learning_rate": 9.000000000000001e-07,
+      "loss": 10.9596,
+      "step": 50
+    },
+    {
+      "epoch": 0.35502958579881655,
+      "grad_norm": 83.35017395019531,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 8.1662,
+      "step": 75
+    },
+    {
+      "epoch": 0.47337278106508873,
+      "grad_norm": 69.67986297607422,
+      "learning_rate": 1.9000000000000002e-06,
+      "loss": 6.1015,
+      "step": 100
+    },
+    {
+      "epoch": 0.591715976331361,
+      "grad_norm": 65.54053497314453,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 4.9343,
+      "step": 125
+    },
+    {
+      "epoch": 0.7100591715976331,
+      "grad_norm": 51.996883392333984,
+      "learning_rate": 2.9e-06,
+      "loss": 4.21,
+      "step": 150
+    },
+    {
+      "epoch": 0.8284023668639053,
+      "grad_norm": 52.11565017700195,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 4.4511,
+      "step": 175
+    },
+    {
+      "epoch": 0.9467455621301775,
+      "grad_norm": 70.60668182373047,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 3.9586,
+      "step": 200
+    },
+    {
+      "epoch": 0.9467455621301775,
+      "eval_loss": 0.8733280897140503,
+      "eval_runtime": 258.2817,
+      "eval_samples_per_second": 1.518,
+      "eval_steps_per_second": 0.759,
+      "eval_wer": 0.5994351612144034,
+      "step": 200
+    },
+    {
+      "epoch": 1.0615384615384615,
+      "grad_norm": 54.279884338378906,
+      "learning_rate": 4.4e-06,
+      "loss": 3.2145,
+      "step": 225
+    },
+    {
+      "epoch": 1.1798816568047337,
+      "grad_norm": 50.64502716064453,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 3.1628,
+      "step": 250
+    },
+    {
+      "epoch": 1.298224852071006,
+      "grad_norm": 62.229068756103516,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 2.9053,
+      "step": 275
+    },
+    {
+      "epoch": 1.4165680473372781,
+      "grad_norm": 48.52084732055664,
+      "learning_rate": 5.9e-06,
+      "loss": 2.8052,
+      "step": 300
+    },
+    {
+      "epoch": 1.5349112426035503,
+      "grad_norm": 45.87370300292969,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.7131,
+      "step": 325
+    },
+    {
+      "epoch": 1.6532544378698226,
+      "grad_norm": 60.52415084838867,
+      "learning_rate": 6.9e-06,
+      "loss": 2.7482,
+      "step": 350
+    },
+    {
+      "epoch": 1.7715976331360945,
+      "grad_norm": 51.26884078979492,
+      "learning_rate": 7.4e-06,
+      "loss": 2.8582,
+      "step": 375
+    },
+    {
+      "epoch": 1.8899408284023669,
+      "grad_norm": 51.80022430419922,
+      "learning_rate": 7.9e-06,
+      "loss": 2.4999,
+      "step": 400
+    },
+    {
+      "epoch": 1.8899408284023669,
+      "eval_loss": 0.6725602149963379,
+      "eval_runtime": 255.9417,
+      "eval_samples_per_second": 1.532,
+      "eval_steps_per_second": 0.766,
+      "eval_wer": 0.46481525064721113,
+      "step": 400
+    },
+    {
+      "epoch": 2.0047337278106507,
+      "grad_norm": 35.06719207763672,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.1474,
+      "step": 425
+    },
+    {
+      "epoch": 2.123076923076923,
+      "grad_norm": 40.39971160888672,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 1.4558,
+      "step": 450
+    },
+    {
+      "epoch": 2.2414201183431954,
+      "grad_norm": 37.043128967285156,
+      "learning_rate": 9.4e-06,
+      "loss": 1.6915,
+      "step": 475
+    },
+    {
+      "epoch": 2.3597633136094673,
+      "grad_norm": 38.598167419433594,
+      "learning_rate": 9.9e-06,
+      "loss": 1.7167,
+      "step": 500
+    },
+    {
+      "epoch": 2.4781065088757397,
+      "grad_norm": 46.40306091308594,
+      "learning_rate": 9.965694682675816e-06,
+      "loss": 1.6127,
+      "step": 525
+    },
+    {
+      "epoch": 2.596449704142012,
+      "grad_norm": 32.87551498413086,
+      "learning_rate": 9.922813036020584e-06,
+      "loss": 1.6845,
+      "step": 550
+    },
+    {
+      "epoch": 2.714792899408284,
+      "grad_norm": 19.153915405273438,
+      "learning_rate": 9.879931389365352e-06,
+      "loss": 1.4961,
+      "step": 575
+    },
+    {
+      "epoch": 2.8331360946745563,
+      "grad_norm": 28.79606819152832,
+      "learning_rate": 9.837049742710121e-06,
+      "loss": 1.7047,
+      "step": 600
+    },
+    {
+      "epoch": 2.8331360946745563,
+      "eval_loss": 0.6522520184516907,
+      "eval_runtime": 254.5133,
+      "eval_samples_per_second": 1.54,
+      "eval_steps_per_second": 0.77,
+      "eval_wer": 0.45846081430924923,
+      "step": 600
+    },
+    {
+      "epoch": 2.9514792899408286,
+      "grad_norm": 39.831016540527344,
+      "learning_rate": 9.794168096054889e-06,
+      "loss": 1.8359,
+      "step": 625
+    },
+    {
+      "epoch": 3.0662721893491125,
+      "grad_norm": 32.2681999206543,
+      "learning_rate": 9.751286449399657e-06,
+      "loss": 1.1873,
+      "step": 650
+    },
+    {
+      "epoch": 3.184615384615385,
+      "grad_norm": 23.113107681274414,
+      "learning_rate": 9.708404802744426e-06,
+      "loss": 0.8456,
+      "step": 675
+    },
+    {
+      "epoch": 3.3029585798816568,
+      "grad_norm": 24.971742630004883,
+      "learning_rate": 9.665523156089196e-06,
+      "loss": 0.7569,
+      "step": 700
+    },
+    {
+      "epoch": 3.421301775147929,
+      "grad_norm": 18.853492736816406,
+      "learning_rate": 9.622641509433963e-06,
+      "loss": 0.9036,
+      "step": 725
+    },
+    {
+      "epoch": 3.5396449704142015,
+      "grad_norm": 23.712282180786133,
+      "learning_rate": 9.579759862778731e-06,
+      "loss": 0.7134,
+      "step": 750
+    },
+    {
+      "epoch": 3.6579881656804734,
+      "grad_norm": 38.5533332824707,
+      "learning_rate": 9.536878216123499e-06,
+      "loss": 0.9859,
+      "step": 775
+    },
+    {
+      "epoch": 3.7763313609467457,
+      "grad_norm": 38.32489776611328,
+      "learning_rate": 9.493996569468268e-06,
+      "loss": 0.9573,
+      "step": 800
+    },
+    {
+      "epoch": 3.7763313609467457,
+      "eval_loss": 0.6425132155418396,
+      "eval_runtime": 261.773,
+      "eval_samples_per_second": 1.497,
+      "eval_steps_per_second": 0.749,
+      "eval_wer": 0.4497528830313015,
+      "step": 800
+    },
+    {
+      "epoch": 3.8946745562130176,
+      "grad_norm": 22.206430435180664,
+      "learning_rate": 9.451114922813038e-06,
+      "loss": 0.876,
+      "step": 825
+    },
+    {
+      "epoch": 4.0094674556213015,
+      "grad_norm": 17.389280319213867,
+      "learning_rate": 9.408233276157806e-06,
+      "loss": 0.8495,
+      "step": 850
+    },
+    {
+      "epoch": 4.127810650887574,
+      "grad_norm": 15.826939582824707,
+      "learning_rate": 9.365351629502573e-06,
+      "loss": 0.3702,
+      "step": 875
+    },
+    {
+      "epoch": 4.246153846153846,
+      "grad_norm": 19.050931930541992,
+      "learning_rate": 9.322469982847341e-06,
+      "loss": 0.4092,
+      "step": 900
+    },
+    {
+      "epoch": 4.364497041420118,
+      "grad_norm": 18.2451114654541,
+      "learning_rate": 9.27958833619211e-06,
+      "loss": 0.3743,
+      "step": 925
+    },
+    {
+      "epoch": 4.482840236686391,
+      "grad_norm": 12.22165298461914,
+      "learning_rate": 9.236706689536878e-06,
+      "loss": 0.4114,
+      "step": 950
+    },
+    {
+      "epoch": 4.601183431952663,
+      "grad_norm": 14.346109390258789,
+      "learning_rate": 9.193825042881648e-06,
+      "loss": 0.3816,
+      "step": 975
+    },
+    {
+      "epoch": 4.719526627218935,
+      "grad_norm": 56.37358474731445,
+      "learning_rate": 9.150943396226416e-06,
+      "loss": 0.4029,
+      "step": 1000
+    },
+    {
+      "epoch": 4.719526627218935,
+      "eval_loss": 0.6656792163848877,
+      "eval_runtime": 253.5608,
+      "eval_samples_per_second": 1.546,
+      "eval_steps_per_second": 0.773,
+      "eval_wer": 0.404330430689574,
+      "step": 1000
+    },
+    {
+      "epoch": 4.8378698224852075,
+      "grad_norm": 23.6176700592041,
+      "learning_rate": 9.108061749571185e-06,
+      "loss": 0.4506,
+      "step": 1025
+    },
+    {
+      "epoch": 4.956213017751479,
+      "grad_norm": 23.03661346435547,
+      "learning_rate": 9.065180102915953e-06,
+      "loss": 0.4928,
+      "step": 1050
+    },
+    {
+      "epoch": 5.071005917159764,
+      "grad_norm": 25.316198348999023,
+      "learning_rate": 9.02229845626072e-06,
+      "loss": 0.2739,
+      "step": 1075
+    },
+    {
+      "epoch": 5.189349112426036,
+      "grad_norm": 6.165097713470459,
+      "learning_rate": 8.97941680960549e-06,
+      "loss": 0.1914,
+      "step": 1100
+    },
+    {
+      "epoch": 5.3076923076923075,
+      "grad_norm": 20.19268226623535,
+      "learning_rate": 8.936535162950258e-06,
+      "loss": 0.221,
+      "step": 1125
+    },
+    {
+      "epoch": 5.42603550295858,
+      "grad_norm": 12.469457626342773,
+      "learning_rate": 8.893653516295027e-06,
+      "loss": 0.199,
+      "step": 1150
+    },
+    {
+      "epoch": 5.544378698224852,
+      "grad_norm": 14.28165054321289,
+      "learning_rate": 8.850771869639795e-06,
+      "loss": 0.2235,
+      "step": 1175
+    },
+    {
+      "epoch": 5.662721893491124,
+      "grad_norm": 15.683744430541992,
+      "learning_rate": 8.807890222984563e-06,
+      "loss": 0.2311,
+      "step": 1200
+    },
+    {
+      "epoch": 5.662721893491124,
+      "eval_loss": 0.6909866333007812,
+      "eval_runtime": 253.3291,
+      "eval_samples_per_second": 1.547,
+      "eval_steps_per_second": 0.774,
+      "eval_wer": 0.41868674982348786,
+      "step": 1200
+    },
+    {
+      "epoch": 5.781065088757396,
+      "grad_norm": 11.54914665222168,
+      "learning_rate": 8.765008576329332e-06,
+      "loss": 0.2098,
+      "step": 1225
+    },
+    {
+      "epoch": 5.899408284023669,
+      "grad_norm": 18.604154586791992,
+      "learning_rate": 8.722126929674101e-06,
+      "loss": 0.1964,
+      "step": 1250
+    },
+    {
+      "epoch": 6.014201183431953,
+      "grad_norm": 13.384803771972656,
+      "learning_rate": 8.67924528301887e-06,
+      "loss": 0.2031,
+      "step": 1275
+    },
+    {
+      "epoch": 6.132544378698225,
+      "grad_norm": 9.78126049041748,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 0.1747,
+      "step": 1300
+    },
+    {
+      "epoch": 6.250887573964497,
+      "grad_norm": 8.569636344909668,
+      "learning_rate": 8.593481989708405e-06,
+      "loss": 0.1078,
+      "step": 1325
+    },
+    {
+      "epoch": 6.36923076923077,
+      "grad_norm": 16.902490615844727,
+      "learning_rate": 8.550600343053174e-06,
+      "loss": 0.1253,
+      "step": 1350
+    },
+    {
+      "epoch": 6.487573964497042,
+      "grad_norm": 15.703929901123047,
+      "learning_rate": 8.507718696397942e-06,
+      "loss": 0.1213,
+      "step": 1375
+    },
+    {
+      "epoch": 6.6059171597633135,
+      "grad_norm": 10.21462631225586,
+      "learning_rate": 8.464837049742711e-06,
+      "loss": 0.1545,
+      "step": 1400
+    },
+    {
+      "epoch": 6.6059171597633135,
+      "eval_loss": 0.720788836479187,
+      "eval_runtime": 248.7794,
+      "eval_samples_per_second": 1.576,
+      "eval_steps_per_second": 0.788,
+      "eval_wer": 0.38644386914568135,
+      "step": 1400
+    },
+    {
+      "epoch": 6.6059171597633135,
+      "step": 1400,
+      "total_flos": 1.138790955810816e+19,
+      "train_loss": 1.9424426797458103,
+      "train_runtime": 4520.3552,
+      "train_samples_per_second": 11.209,
+      "train_steps_per_second": 1.4
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 6330,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "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": 1.138790955810816e+19,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}