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: 0.1706
Wer Ortho: 10.3861
Wer: 5.9690

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: 50
training_steps: 500
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.2136	0.2	100	0.1925	11.7012	7.0463
0.1862	0.4	200	0.1846	11.0360	6.4247
0.1899	0.6	300	0.1784	10.6973	6.2042
0.1746	0.8	400	0.1735	10.6776	5.9754
0.1794	1.0	500	0.1706	10.3861	5.9690

Framework versions

Transformers 4.53.2
Pytorch 2.9.1+cu128
Datasets 2.14.4
Tokenizers 0.21.4

mmarron14/whisper-medium-cv17-es-500-steps_proc2-def3

Whisper Medium CV17 Es 500 steps with same configuration as 5-steps_proc2-def3 -processing of text using method 2; with filtering 30sec, and customised optimizer-; but with values of training_args similar to our another 500-steps model- María Marrón

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

mmarron14/
whisper-medium-cv17-es-500-steps_proc2-def3