whisper-small-nomo

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: 0.0001
• Wer: 4.0426

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 132
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.1746	2.0833	100	0.3119	318.9362
0.3844	4.1667	200	0.2805	23.8298
0.2172	6.25	300	0.1609	26.5957
0.1388	8.3333	400	0.0868	56.1702
0.098	10.4167	500	0.0621	14.8936
0.0726	12.5	600	0.0523	17.0213
0.0573	14.5833	700	0.0255	7.6596
0.0359	16.6667	800	0.0084	47.4468
0.0199	18.75	900	0.0020	50.0
0.0158	20.8333	1000	0.0030	50.0
0.0037	22.9167	1100	0.0008	49.1489
0.0005	25.0	1200	0.0002	3.6170
0.0004	27.0833	1300	0.0001	4.0426
0.0002	29.1667	1400	0.0001	4.0426

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0
• Datasets 2.21.0
• Tokenizers 0.19.1