whisper-small-vi-vlsp

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

• Loss: 0.0808
• Wer: 14.2857

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
0.1227	0.2216	500	0.1274	21.2294
0.1044	0.4433	1000	0.1056	19.4552
0.0888	0.6649	1500	0.0954	16.9265
0.0877	0.8865	2000	0.0894	15.7455
0.0592	1.1082	2500	0.0859	15.1250
0.0541	1.3298	3000	0.0836	15.6218
0.0542	1.5514	3500	0.0820	16.8766
0.0574	1.7730	4000	0.0808	14.2857

Framework versions

• Transformers 4.48.0
• Pytorch 2.8.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4