wh_small_cv_trial

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: 0.2449
• Global Wer: 34.3493

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: 5e-06
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• optimizer: Use adamw_torch_fused with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 500

Training results

Training Loss	Epoch	Step	Validation Loss	Global Wer
3.6888	0.1642	50	0.7364	55.3054
2.2599	0.3284	100	0.5582	45.9355
1.8204	0.4926	150	0.4654	45.5305
1.5469	0.6568	200	0.3952	39.6438
1.2098	0.8210	250	0.3165	38.1983
1.0164	0.9852	300	0.2763	37.5495
0.6866	1.1478	350	0.2647	36.5786
0.7090	1.3120	400	0.2593	35.7644
0.6868	1.4762	450	0.2494	34.9806
0.6522	1.6404	500	0.2449	34.3493

Framework versions

• Transformers 5.0.0.dev0
• Pytorch 2.9.0+cu126
• Datasets 3.6.0
• Tokenizers 0.22.2