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-baseline-model
-  results: []
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: nyagen
+      type: nyagen
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.31363034847816496
 ---
 
 <!-- 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-baseline-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: 0.5093
-- Wer: 0.3048
+- Loss: 0.4721
+- Wer: 0.3136
 
 ## Model description
 

all_results.json

@@ -0,0 +1,15 @@
+{
+    "epoch": 6.154241645244216,
+    "eval_loss": 0.47213003039360046,
+    "eval_runtime": 117.5727,
+    "eval_samples": 172,
+    "eval_samples_per_second": 1.463,
+    "eval_steps_per_second": 0.731,
+    "eval_wer": 0.31363034847816496,
+    "total_flos": 9.7671889649664e+18,
+    "train_loss": 1.094322391251723,
+    "train_runtime": 3156.6217,
+    "train_samples": 1555,
+    "train_samples_per_second": 12.672,
+    "train_steps_per_second": 1.584
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 6.154241645244216,
+    "eval_loss": 0.47213003039360046,
+    "eval_runtime": 117.5727,
+    "eval_samples": 172,
+    "eval_samples_per_second": 1.463,
+    "eval_steps_per_second": 0.731,
+    "eval_wer": 0.31363034847816496
+}

runs/Feb08_23-33-50_srvrocgpu011.uct.ac.za/events.out.tfevents.1739053980.srvrocgpu011.uct.ac.za

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f75f13b6954ad9d4d8c88b204374f5d627b10519c5b4bf5804ce0f38139f906f
+size 40

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 6.154241645244216,
+    "total_flos": 9.7671889649664e+18,
+    "train_loss": 1.094322391251723,
+    "train_runtime": 3156.6217,
+    "train_samples": 1555,
+    "train_samples_per_second": 12.672,
+    "train_steps_per_second": 1.584
+}

trainer_state.json

@@ -0,0 +1,441 @@
+{
+  "best_metric": 0.47213003039360046,
+  "best_model_checkpoint": "/scratch/skscla001/speech/results/whisper-medium-nyagen-baseline-model/checkpoint-600",
+  "epoch": 6.154241645244216,
+  "eval_steps": 200,
+  "global_step": 1200,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.12853470437017994,
+      "grad_norm": 106.13851928710938,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 9.6589,
+      "step": 25
+    },
+    {
+      "epoch": 0.2570694087403599,
+      "grad_norm": 82.48638916015625,
+      "learning_rate": 9.200000000000001e-07,
+      "loss": 7.8012,
+      "step": 50
+    },
+    {
+      "epoch": 0.3856041131105398,
+      "grad_norm": 47.58100891113281,
+      "learning_rate": 1.42e-06,
+      "loss": 5.3963,
+      "step": 75
+    },
+    {
+      "epoch": 0.5141388174807198,
+      "grad_norm": 51.5584831237793,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 3.6524,
+      "step": 100
+    },
+    {
+      "epoch": 0.6426735218508998,
+      "grad_norm": 40.65543746948242,
+      "learning_rate": 2.42e-06,
+      "loss": 2.9645,
+      "step": 125
+    },
+    {
+      "epoch": 0.7712082262210797,
+      "grad_norm": 48.534339904785156,
+      "learning_rate": 2.9e-06,
+      "loss": 2.4303,
+      "step": 150
+    },
+    {
+      "epoch": 0.8997429305912596,
+      "grad_norm": 42.68857192993164,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 1.7682,
+      "step": 175
+    },
+    {
+      "epoch": 1.025706940874036,
+      "grad_norm": 33.803348541259766,
+      "learning_rate": 3.900000000000001e-06,
+      "loss": 1.6573,
+      "step": 200
+    },
+    {
+      "epoch": 1.025706940874036,
+      "eval_loss": 0.747204601764679,
+      "eval_runtime": 117.9895,
+      "eval_samples_per_second": 1.458,
+      "eval_steps_per_second": 0.729,
+      "eval_wer": 0.4662549625055139,
+      "step": 200
+    },
+    {
+      "epoch": 1.154241645244216,
+      "grad_norm": 30.34743881225586,
+      "learning_rate": 4.4e-06,
+      "loss": 1.184,
+      "step": 225
+    },
+    {
+      "epoch": 1.2827763496143958,
+      "grad_norm": 27.770002365112305,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 1.1498,
+      "step": 250
+    },
+    {
+      "epoch": 1.411311053984576,
+      "grad_norm": 31.53769874572754,
+      "learning_rate": 5.400000000000001e-06,
+      "loss": 1.2871,
+      "step": 275
+    },
+    {
+      "epoch": 1.5398457583547558,
+      "grad_norm": 31.70905303955078,
+      "learning_rate": 5.9e-06,
+      "loss": 1.2743,
+      "step": 300
+    },
+    {
+      "epoch": 1.6683804627249357,
+      "grad_norm": 46.74306106567383,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 1.2064,
+      "step": 325
+    },
+    {
+      "epoch": 1.7969151670951158,
+      "grad_norm": 35.37800216674805,
+      "learning_rate": 6.9e-06,
+      "loss": 1.0643,
+      "step": 350
+    },
+    {
+      "epoch": 1.9254498714652957,
+      "grad_norm": 26.779239654541016,
+      "learning_rate": 7.4e-06,
+      "loss": 1.0283,
+      "step": 375
+    },
+    {
+      "epoch": 2.051413881748072,
+      "grad_norm": 23.027477264404297,
+      "learning_rate": 7.9e-06,
+      "loss": 0.8271,
+      "step": 400
+    },
+    {
+      "epoch": 2.051413881748072,
+      "eval_loss": 0.49850764870643616,
+      "eval_runtime": 123.2873,
+      "eval_samples_per_second": 1.395,
+      "eval_steps_per_second": 0.698,
+      "eval_wer": 0.43890604322893695,
+      "step": 400
+    },
+    {
+      "epoch": 2.179948586118252,
+      "grad_norm": 12.114481925964355,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 0.4631,
+      "step": 425
+    },
+    {
+      "epoch": 2.308483290488432,
+      "grad_norm": 27.12820053100586,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.4889,
+      "step": 450
+    },
+    {
+      "epoch": 2.437017994858612,
+      "grad_norm": 23.574703216552734,
+      "learning_rate": 9.4e-06,
+      "loss": 0.5423,
+      "step": 475
+    },
+    {
+      "epoch": 2.5655526992287916,
+      "grad_norm": 19.00645637512207,
+      "learning_rate": 9.9e-06,
+      "loss": 0.6414,
+      "step": 500
+    },
+    {
+      "epoch": 2.6940874035989717,
+      "grad_norm": 23.830310821533203,
+      "learning_rate": 9.955555555555556e-06,
+      "loss": 0.5073,
+      "step": 525
+    },
+    {
+      "epoch": 2.822622107969152,
+      "grad_norm": 20.477584838867188,
+      "learning_rate": 9.9e-06,
+      "loss": 0.6408,
+      "step": 550
+    },
+    {
+      "epoch": 2.9511568123393315,
+      "grad_norm": 17.4973201751709,
+      "learning_rate": 9.844444444444446e-06,
+      "loss": 0.5744,
+      "step": 575
+    },
+    {
+      "epoch": 3.077120822622108,
+      "grad_norm": 20.835445404052734,
+      "learning_rate": 9.78888888888889e-06,
+      "loss": 0.4472,
+      "step": 600
+    },
+    {
+      "epoch": 3.077120822622108,
+      "eval_loss": 0.47213003039360046,
+      "eval_runtime": 119.7194,
+      "eval_samples_per_second": 1.437,
+      "eval_steps_per_second": 0.718,
+      "eval_wer": 0.31363034847816496,
+      "step": 600
+    },
+    {
+      "epoch": 3.2056555269922877,
+      "grad_norm": 6.538525581359863,
+      "learning_rate": 9.733333333333334e-06,
+      "loss": 0.1986,
+      "step": 625
+    },
+    {
+      "epoch": 3.334190231362468,
+      "grad_norm": 15.117712020874023,
+      "learning_rate": 9.677777777777778e-06,
+      "loss": 0.2394,
+      "step": 650
+    },
+    {
+      "epoch": 3.462724935732648,
+      "grad_norm": 15.175765037536621,
+      "learning_rate": 9.622222222222222e-06,
+      "loss": 0.2637,
+      "step": 675
+    },
+    {
+      "epoch": 3.5912596401028276,
+      "grad_norm": 13.438119888305664,
+      "learning_rate": 9.566666666666668e-06,
+      "loss": 0.2456,
+      "step": 700
+    },
+    {
+      "epoch": 3.7197943444730077,
+      "grad_norm": 21.882404327392578,
+      "learning_rate": 9.511111111111112e-06,
+      "loss": 0.273,
+      "step": 725
+    },
+    {
+      "epoch": 3.848329048843188,
+      "grad_norm": 15.743215560913086,
+      "learning_rate": 9.455555555555557e-06,
+      "loss": 0.41,
+      "step": 750
+    },
+    {
+      "epoch": 3.9768637532133675,
+      "grad_norm": 12.749256134033203,
+      "learning_rate": 9.4e-06,
+      "loss": 0.2412,
+      "step": 775
+    },
+    {
+      "epoch": 4.102827763496144,
+      "grad_norm": 12.16617202758789,
+      "learning_rate": 9.344444444444446e-06,
+      "loss": 0.1499,
+      "step": 800
+    },
+    {
+      "epoch": 4.102827763496144,
+      "eval_loss": 0.4967658817768097,
+      "eval_runtime": 118.6311,
+      "eval_samples_per_second": 1.45,
+      "eval_steps_per_second": 0.725,
+      "eval_wer": 0.30921923246581384,
+      "step": 800
+    },
+    {
+      "epoch": 4.231362467866324,
+      "grad_norm": 12.10049819946289,
+      "learning_rate": 9.28888888888889e-06,
+      "loss": 0.1118,
+      "step": 825
+    },
+    {
+      "epoch": 4.359897172236504,
+      "grad_norm": 6.198291778564453,
+      "learning_rate": 9.233333333333334e-06,
+      "loss": 0.1684,
+      "step": 850
+    },
+    {
+      "epoch": 4.4884318766066835,
+      "grad_norm": 4.662709712982178,
+      "learning_rate": 9.17777777777778e-06,
+      "loss": 0.1103,
+      "step": 875
+    },
+    {
+      "epoch": 4.616966580976864,
+      "grad_norm": 7.5239787101745605,
+      "learning_rate": 9.122222222222223e-06,
+      "loss": 0.1963,
+      "step": 900
+    },
+    {
+      "epoch": 4.745501285347044,
+      "grad_norm": 11.247775077819824,
+      "learning_rate": 9.066666666666667e-06,
+      "loss": 0.1525,
+      "step": 925
+    },
+    {
+      "epoch": 4.874035989717224,
+      "grad_norm": 17.83033561706543,
+      "learning_rate": 9.011111111111111e-06,
+      "loss": 0.1736,
+      "step": 950
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 6.933111190795898,
+      "learning_rate": 8.955555555555555e-06,
+      "loss": 0.1397,
+      "step": 975
+    },
+    {
+      "epoch": 5.12853470437018,
+      "grad_norm": 6.672903060913086,
+      "learning_rate": 8.900000000000001e-06,
+      "loss": 0.064,
+      "step": 1000
+    },
+    {
+      "epoch": 5.12853470437018,
+      "eval_loss": 0.48916691541671753,
+      "eval_runtime": 123.2475,
+      "eval_samples_per_second": 1.396,
+      "eval_steps_per_second": 0.698,
+      "eval_wer": 0.333480370533745,
+      "step": 1000
+    },
+    {
+      "epoch": 5.25706940874036,
+      "grad_norm": 17.61103057861328,
+      "learning_rate": 8.844444444444445e-06,
+      "loss": 0.0843,
+      "step": 1025
+    },
+    {
+      "epoch": 5.385604113110539,
+      "grad_norm": 17.312223434448242,
+      "learning_rate": 8.788888888888891e-06,
+      "loss": 0.0925,
+      "step": 1050
+    },
+    {
+      "epoch": 5.5141388174807195,
+      "grad_norm": 13.309065818786621,
+      "learning_rate": 8.733333333333333e-06,
+      "loss": 0.1091,
+      "step": 1075
+    },
+    {
+      "epoch": 5.6426735218509,
+      "grad_norm": 11.82165241241455,
+      "learning_rate": 8.677777777777779e-06,
+      "loss": 0.1296,
+      "step": 1100
+    },
+    {
+      "epoch": 5.77120822622108,
+      "grad_norm": 4.475037097930908,
+      "learning_rate": 8.622222222222223e-06,
+      "loss": 0.1161,
+      "step": 1125
+    },
+    {
+      "epoch": 5.89974293059126,
+      "grad_norm": 6.766268730163574,
+      "learning_rate": 8.566666666666667e-06,
+      "loss": 0.0754,
+      "step": 1150
+    },
+    {
+      "epoch": 6.025706940874036,
+      "grad_norm": 2.879678726196289,
+      "learning_rate": 8.511111111111113e-06,
+      "loss": 0.075,
+      "step": 1175
+    },
+    {
+      "epoch": 6.154241645244216,
+      "grad_norm": 4.57487678527832,
+      "learning_rate": 8.455555555555555e-06,
+      "loss": 0.0521,
+      "step": 1200
+    },
+    {
+      "epoch": 6.154241645244216,
+      "eval_loss": 0.5093271136283875,
+      "eval_runtime": 117.7178,
+      "eval_samples_per_second": 1.461,
+      "eval_steps_per_second": 0.731,
+      "eval_wer": 0.30480811645346273,
+      "step": 1200
+    },
+    {
+      "epoch": 6.154241645244216,
+      "step": 1200,
+      "total_flos": 9.7671889649664e+18,
+      "train_loss": 1.094322391251723,
+      "train_runtime": 3156.6217,
+      "train_samples_per_second": 12.672,
+      "train_steps_per_second": 1.584
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 26,
+  "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": 9.7671889649664e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}