zuazo/whisper-medium-eu-cv22.0

openai/whisper-medium

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

• Loss: 0.3322
• Wer: 11.1992

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: 64
• eval_batch_size: 32
• seed: 42
• 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
0.0128	10.8225	5000	0.2586	10.4875
0.0073	21.6450	10000	0.2837	10.6540
0.0053	32.4675	15000	0.2982	10.8248
0.0039	43.2900	20000	0.3295	10.9025
0.0018	54.1126	25000	0.3297	10.7386
0.0044	64.9351	30000	0.3322	11.1992

Framework versions

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