zuazo/whisper-small-eu-cv22.0

openai/whisper-small

This model is a fine-tuned version of openai/whisper-small on the common_voice_22_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3891
• Wer: 12.1602

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: 3.75e-05
• train_batch_size: 64
• eval_batch_size: 32
• 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: 500
• training_steps: 100000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0113	10.8225	5000	0.2778	12.1163
0.0047	21.6450	10000	0.3096	11.8923
0.0059	32.4675	15000	0.3285	12.1856
0.0028	43.2900	20000	0.3573	11.6886
0.0015	54.1126	25000	0.3549	11.6117
0.0028	64.9351	30000	0.3610	11.9312
0.0013	75.7576	35000	0.3711	11.6683
0.0009	86.5801	40000	0.3756	11.6176
0.0008	97.4026	45000	0.3801	11.8213
0.0009	108.2251	50000	0.3891	12.1602

Framework versions

• Transformers 4.52.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1