lorenzoncina/whisper-small-en

Whisper Small English

This model is a fine-tuned version of openai/whisper-small on the mozilla-foundation/common_voice_11_0 en dataset. It achieves the following results on the evaluation set:

• Loss: 0.3269
• Wer: 13.0585

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: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1537	0.1	1000	0.4405	17.9276
0.2378	0.2	2000	0.4009	15.9888
0.1709	0.3	3000	0.3852	15.4953
0.2792	0.4	4000	0.3699	14.8758
0.2172	0.5	5000	0.3577	14.2660
0.3616	0.6	6000	0.4042	18.1846
0.2456	0.7	7000	0.3375	13.3091
0.2505	0.8	8000	0.3395	13.6227
0.2563	0.9	9000	0.3305	13.1408
0.2395	1.0	10000	0.3269	13.0585

Framework versions

• Transformers 4.28.0.dev0
• Pytorch 2.0.0+cu117
• Datasets 2.11.1.dev0
• Tokenizers 0.13.2