misterkissi/whisper-small-tigre

wisper-small-tigre

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

• Loss: 0.5606
• Wer: 0.5816

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: 64
• eval_batch_size: 16
• seed: 42
• optimizer: Use 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: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4124	6.7568	500	0.2653	1.2609
0.0549	13.5135	1000	0.3898	0.7472
0.0073	20.2703	1500	0.4585	0.6313
0.0027	27.0270	2000	0.4898	0.5996
0.001	33.7838	2500	0.5110	0.5899
0.0002	40.5405	3000	0.5326	0.5899
0.0001	47.2973	3500	0.5452	0.5868
0.0001	54.0541	4000	0.5534	0.5854
0.0	60.8108	4500	0.5586	0.5849
0.0	67.5676	5000	0.5606	0.5816

Framework versions

• Transformers 4.53.2
• Pytorch 2.6.0+cu124
• Datasets 2.18.0
• Tokenizers 0.21.2