README.md

metadata

library_name: transformers
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - PolyAI/minds14
metrics:
  - wer
model-index:
  - name: whisper-tiny-en-US
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: PolyAI/minds14
          type: PolyAI/minds14
          config: en-US
          split: train[450:]
          args: en-US
        metrics:
          - name: Wer
            type: wer
            value: 0.34946871310507677

whisper-tiny-en-US

This model is a fine-tuned version of openai/whisper-tiny on the PolyAI/minds14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7334
• Wer Ortho: 0.3701
• Wer: 0.3495

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 48
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• num_epochs: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
2.5445	1.0	10	2.4472	0.5379	0.3985
2.0492	2.0	20	1.8586	0.5287	0.3973
1.3657	3.0	30	1.1065	0.4867	0.4038
0.6326	4.0	40	0.5885	0.4769	0.4115
0.3984	5.0	50	0.5155	0.4399	0.3861
0.2907	6.0	60	0.4921	0.3849	0.3347
0.236	7.0	70	0.4864	0.3886	0.3459
0.14	8.0	80	0.4936	0.3677	0.3264
0.106	9.0	90	0.5082	0.3917	0.3518
0.0837	10.0	100	0.5316	0.3819	0.3347
0.0458	11.0	110	0.5475	0.3899	0.3489
0.0201	12.0	120	0.5706	0.3893	0.3536
0.0099	13.0	130	0.5851	0.3831	0.3495
0.0067	14.0	140	0.6010	0.3769	0.3489
0.0036	15.0	150	0.6196	0.3819	0.3506
0.0021	16.0	160	0.6377	0.3782	0.3530
0.0013	17.0	170	0.6539	0.3708	0.3453
0.0006	18.0	180	0.6831	0.3720	0.3506
0.0004	19.0	190	0.7018	0.3732	0.3512
0.0003	20.0	200	0.7334	0.3701	0.3495

Framework versions

• Transformers 4.46.1
• Pytorch 2.3.1
• Datasets 3.0.2
• Tokenizers 0.20.1