sidleal/distil-whisper-coraa-mupe-asr-1

CORAA-MUPE-ASR distil-whisper fine-tuned

This model is a fine-tuned version of RodrigoLimaRFL/distil-whisper-nurc-sp-fine-tuned on the CORAA-MUPE-ASR dataset. It achieves the following results on the evaluation set:

• Loss: 0.3488
• Wer: 15.2736

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4451	0.0558	1000	0.4482	18.5598
0.4006	0.1116	2000	0.4095	17.3061
0.2992	0.1674	3000	0.3848	16.5660
0.2781	0.2232	4000	0.3609	15.5857
0.2839	0.2790	5000	0.3488	15.2736

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1