whisper-a-nomimo-17

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: 0.0292
• Wer: 9.7222

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: 17
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.9831	1.0	104	0.2031	33.0247
0.3134	2.0	208	0.1138	87.1142
0.0638	3.0	312	0.0907	27.4691
0.0503	4.0	416	0.0745	26.4660
0.0592	5.0	520	0.0361	25.3858
0.0329	6.0	624	0.0674	27.7006
0.0276	7.0	728	0.0524	47.0679
0.0322	8.0	832	0.0321	12.5
0.015	9.0	936	0.0357	13.6574
0.0117	10.0	1040	0.0335	10.3395
0.0075	11.0	1144	0.0387	12.1914
0.0085	12.0	1248	0.0306	9.8765
0.0056	13.0	1352	0.0311	9.0278
0.0031	14.0	1456	0.0274	8.9506
0.0038	15.0	1560	0.0288	9.7222
0.0024	16.0	1664	0.0288	9.4136
0.0013	16.8406	1751	0.0292	9.7222

Framework versions

• Transformers 4.47.0
• Pytorch 2.4.0
• Datasets 3.1.0
• Tokenizers 0.21.0