whisper-small-mupe-3

This model is a fine-tuned version of lejonck/whisper-small-mupe-2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7351
• Wer: 0.2430
• Cer: 0.5095

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: 8
• eval_batch_size: 2
• 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: 100
• num_epochs: 12

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
0.2677	1.0	1000	0.4901	0.2760	0.5167
0.1301	2.0	2000	0.5221	0.2478	0.5118
0.0496	3.0	3000	0.5680	0.2489	0.5119
0.0204	4.0	4000	0.6225	0.2662	0.5170
0.0103	5.0	5000	0.6579	0.2456	0.5110
0.0091	6.0	6000	0.6706	0.2502	0.5122
0.0012	7.0	7000	0.6920	0.2427	0.5111
0.0005	8.0	8000	0.7184	0.2422	0.5098
0.0003	9.0	9000	0.7554	0.2417	0.5112
0.0003	10.0	10000	0.7458	0.2478	0.5122
0.0002	11.0	11000	0.7628	0.2523	0.5121
0.0002	12.0	12000	0.7785	0.2453	0.5119

Framework versions

• Transformers 4.55.2
• Pytorch 2.7.0+cu126
• Datasets 4.0.0
• Tokenizers 0.21.4