whisper-large-v3-ja

This model is a fine-tuned version of openai/whisper-large-v3 on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.4210
• Wer: 14.6965

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: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1542	1.69	500	0.2712	15.6149
0.0351	3.39	1000	0.3074	16.1866
0.0081	5.08	1500	0.3475	15.3802
0.0049	6.78	2000	0.3427	15.1804
0.001	8.47	2500	0.3851	14.7302
0.0004	10.17	3000	0.4109	14.7254
0.0003	11.86	3500	0.4168	14.6953
0.0003	13.56	4000	0.4210	14.6965

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.2