whisper-a-clp-ls

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: 0.0240
• Wer: 10.0629

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: 0.0004
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Use 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: 132
• num_epochs: 11
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	40	0.1713	46.9602
No log	2.0	80	0.0920	28.3019
1.2158	3.0	120	0.1828	31.2369
1.2158	4.0	160	0.2743	42.3480
0.1604	5.0	200	0.1326	62.8931
0.1604	6.0	240	0.0734	25.7862
0.1604	7.0	280	0.0510	15.7233
0.0502	8.0	320	0.0262	10.4822
0.0502	9.0	360	0.0320	11.9497
0.0202	10.0	400	0.0229	7.1279
0.0202	10.7342	429	0.0240	10.0629

Framework versions

• Transformers 4.47.0.dev0
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0