whisper-a-nomimo

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: 0.0351
• Wer: 22.9167

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: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.9699	0.9662	100	0.1287	46.6821
0.1644	1.9275	200	0.1624	445.0617
0.236	2.8889	300	0.0895	23.1481
0.0518	3.8502	400	0.0479	18.4414
0.0321	4.8116	500	0.0426	14.8148
0.03	5.7729	600	0.0482	19.4444
0.0218	6.7343	700	0.0325	11.6512
0.0143	7.6957	800	0.0439	15.2778
0.0147	8.6570	900	0.0339	11.9599
0.0104	9.6184	1000	0.0391	14.5833
0.0079	10.5797	1100	0.0338	33.9506
0.0054	11.5411	1200	0.0293	20.4475
0.0032	12.5024	1300	0.0357	14.3519
0.002	13.4638	1400	0.0327	18.0556
0.0023	14.4251	1500	0.0351	22.9167

Framework versions

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