zuazo/whisper-large-v2-eu-cv21.0

Whisper Large-V2 Basque

This model is a fine-tuned version of openai/whisper-large-v2 on the common_voice_21_0_eu dataset. It achieves the following results on the evaluation set:

• Loss: 0.2048
• Wer: 8.6823

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: 32
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• 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
0.0086	11.1112	5000	0.2048	8.6823
0.0049	22.2225	10000	0.2296	9.1852
0.0026	33.3337	15000	0.2459	9.0196
0.004	44.4449	20000	0.2476	9.1453
0.0029	55.5562	25000	0.2631	9.7765
0.0017	66.6674	30000	0.2687	9.0057

Framework versions

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