whisper-tiny-aug-19-april-lightning-v1

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1330
• Wer: 85.8714

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: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0683	1.0	712	0.4517	100.8319
0.3469	2.0	1424	0.2621	97.0808
0.24	3.0	2136	0.2112	94.0364
0.1958	4.0	2848	0.1838	91.4772
0.1692	5.0	3560	0.1672	90.0219
0.1506	6.0	4272	0.1565	90.5312
0.1367	7.0	4984	0.1479	88.0180
0.1252	8.0	5696	0.1422	88.6304
0.1162	9.0	6408	0.1372	85.7419
0.1085	10.0	7120	0.1330	85.8714

Framework versions

• Transformers 4.51.3
• Pytorch 2.2.1+cu121
• Datasets 3.5.0
• Tokenizers 0.21.1