metadata

language:
  - en
license: apache-2.0
base_model: openai/whisper-tiny
tags:
  - generated_from_trainer
datasets:
  - PolyAI/minds14
metrics:
  - wer
model-index:
  - name: Whisper tiny
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: minds14
          type: PolyAI/minds14
          config: en-US
          split: train
          args: en-US
        metrics:
          - name: Wer
            type: wer
            value: 32.88075560802833

Whisper tiny

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

Loss: 0.6734
Wer Ortho: 32.6959
Wer: 32.8808

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: 5e-05
train_batch_size: 8
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant_with_warmup
lr_scheduler_warmup_steps: 8
training_steps: 225
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
2.8035	0.25	14	0.7206	42.1345	40.4368
0.5469	0.5	28	0.5327	36.0888	36.3046
0.4968	0.75	42	0.5195	34.7933	34.7107
0.5012	1.0	56	0.5551	33.5595	33.7072
0.1879	1.25	70	0.5353	31.5854	31.4640
0.239	1.5	84	0.5303	37.4460	40.6139
0.2082	1.75	98	0.5565	31.0302	31.2279
0.2244	2.0	112	0.5540	28.5626	28.6305
0.0679	2.25	126	0.5759	28.5009	28.6305
0.0637	2.5	140	0.6192	50.7094	54.0732
0.072	2.75	154	0.6093	31.2770	30.9327
0.0506	3.0	168	0.6302	35.2869	35.5372
0.029	3.25	182	0.6299	33.4978	33.5891
0.0405	3.5	196	0.6159	30.1049	30.4014
0.022	3.75	210	0.6441	30.7218	30.9917
0.0332	4.0	224	0.6734	32.6959	32.8808

Framework versions

Transformers 4.40.2
Pytorch 2.3.1+cu121
Datasets 2.20.0
Tokenizers 0.19.1