Cristhian2430/whisper-large-coes-v8

Whisper Large SEIN - COES SEIN - Version 8

This model is a fine-tuned version of openai/whisper-large-v3-turbo on the SEIN COES dataset. It achieves the following results on the evaluation set:

• Loss: 3.6411
• Wer: 67.2340

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: 16
• eval_batch_size: 8
• 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0831	45.4545	500	2.7691	70.0
0.0117	90.9091	1000	3.1980	68.3511
0.0037	136.3636	1500	3.3787	68.7234
0.0003	181.8182	2000	3.5221	67.8191
0.0002	227.2727	2500	3.5781	67.7660
0.0002	272.7273	3000	3.6119	67.5
0.0002	318.1818	3500	3.6331	67.1809
0.0002	363.6364	4000	3.6411	67.2340

Framework versions

• Transformers 4.56.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 4.0.0
• Tokenizers 0.21.4