mmarron14/whisper-medium-cv17-es-50-steps_proc2-def3

Whisper Medium CV17 Es 50 steps with same configuration as 5-steps_proc2-def3 -processing of text using method 2; with filtering 30sec, and customised optimizer-; but values of training_args similar to 50-steps- María Marrón

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 17.0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1984
• Wer Ortho: 11.3352
• Wer: 6.6179

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: 2
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 32
• 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: 5
• training_steps: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
No log	0.2	10	0.2395	12.9107	7.7479
No log	0.4	20	0.2106	11.6515	6.8111
0.3618	0.6	30	0.2038	11.5459	6.7736
0.3618	0.8	40	0.1997	11.3721	6.6618
0.2271	1.0	50	0.1984	11.3352	6.6179

Framework versions

• Transformers 4.53.2
• Pytorch 2.9.1+cu128
• Datasets 2.14.4
• Tokenizers 0.21.4