metadata

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-AU
          split: train
          args: en-AU
        metrics:
          - name: Wer
            type: wer
            value: 0.1655499720826354

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: 0.4245
Wer Ortho: 0.1714
Wer: 0.1655

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: 16
eval_batch_size: 16
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant_with_warmup
lr_scheduler_warmup_steps: 5
training_steps: 400

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
No log	0.36	10	3.1022	0.3282	0.1960
No log	0.71	20	1.6867	0.2399	0.1865
2.9245	1.07	30	0.6685	0.2332	0.1982
2.9245	1.43	40	0.4912	0.2017	0.1848
0.6297	1.79	50	0.4243	0.1865	0.1753
0.6297	2.14	60	0.3895	0.1801	0.1689
0.6297	2.5	70	0.3678	0.1769	0.1669
0.3045	2.86	80	0.3570	0.1746	0.1689
0.3045	3.21	90	0.3496	0.1720	0.1647
0.1949	3.57	100	0.3451	0.1746	0.1661
0.1949	3.93	110	0.3407	0.1804	0.1700
0.1949	4.29	120	0.3439	0.1778	0.1695
0.1099	4.64	130	0.3501	0.1743	0.1689
0.1099	5.0	140	0.3488	0.1737	0.1667
0.0583	5.36	150	0.3554	0.1778	0.1697
0.0583	5.71	160	0.3595	0.1708	0.1628
0.0583	6.07	170	0.3514	0.1746	0.1661
0.032	6.43	180	0.3672	0.1755	0.1683
0.032	6.79	190	0.3676	0.1676	0.1602
0.0146	7.14	200	0.3791	0.1658	0.1600
0.0146	7.5	210	0.3825	0.1676	0.1625
0.0146	7.86	220	0.3799	0.1702	0.1650
0.0084	8.21	230	0.3827	0.1702	0.1655
0.0084	8.57	240	0.3869	0.1778	0.1714
0.0043	8.93	250	0.3951	0.1740	0.1686
0.0043	9.29	260	0.3958	0.1720	0.1672
0.0043	9.64	270	0.3968	0.1758	0.1706
0.003	10.0	280	0.3978	0.1725	0.1672
0.003	10.36	290	0.4012	0.1734	0.1681
0.0023	10.71	300	0.4068	0.1728	0.1678
0.0023	11.07	310	0.4097	0.1752	0.1697
0.0023	11.43	320	0.4113	0.1746	0.1692
0.0018	11.79	330	0.4120	0.1737	0.1681
0.0018	12.14	340	0.4141	0.1740	0.1683
0.0016	12.5	350	0.4172	0.1731	0.1678
0.0016	12.86	360	0.4193	0.1740	0.1681
0.0016	13.21	370	0.4197	0.1731	0.1672
0.0014	13.57	380	0.4215	0.1731	0.1672
0.0014	13.93	390	0.4228	0.1720	0.1664
0.0012	14.29	400	0.4245	0.1714	0.1655

Framework versions

Transformers 4.31.0.dev0
Pytorch 2.0.1+cu118
Datasets 2.13.1
Tokenizers 0.13.3