MohammadKhosravi/whisper-small

Whisper small- Mohammad Khosravi

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset. It achieves the following results on the evaluation set:

• Loss: 2.0832
• Wer: 105.3763

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: 8
• 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: linear
• lr_scheduler_warmup_steps: 5
• training_steps: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.4286	10	1.9791	109.6774
No log	2.8571	20	1.6973	107.5269
1.023	4.2857	30	1.6941	109.6774
1.023	5.7143	40	1.7788	107.5269
0.1444	7.1429	50	1.8726	104.3011
0.1444	8.5714	60	1.9535	103.2258
0.1444	10.0	70	1.9987	104.3011
0.0166	11.4286	80	2.0563	102.1505
0.0166	12.8571	90	2.0768	105.3763
0.0062	14.2857	100	2.0832	105.3763

Framework versions

• Transformers 4.48.0
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0