whisper-a-norm-ls-8

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: 0.1538
• Wer: 78.3845

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: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0	70	0.2366	16.4960
0.7791	2.0	140	0.4579	96.5870
0.8786	3.0	210	0.2974	91.4676
0.8786	4.0	280	0.2770	93.1741
0.2773	5.0	350	0.2503	91.8089
0.2596	6.0	420	0.3236	95.2218
0.2596	7.0	490	0.1855	93.0603
0.2108	7.8921	552	0.1538	78.3845

Framework versions

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