zuazo/whisper-large-eu-cv22.0

openai/whisper-large

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

• Loss: 0.3446
• Wer: 11.5035

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
0.0153	10.8234	5000	0.2641	10.2881
0.0077	21.6457	10000	0.2892	10.5315
0.0054	32.4680	15000	0.3138	10.5839
0.0058	43.2904	20000	0.3198	11.0335
0.003	54.1127	25000	0.3352	10.7326
0.0047	64.9361	30000	0.3446	11.5035

Framework versions

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