whisper-base-aug-20-april-lightning-v1

This model is a fine-tuned version of openai/whisper-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.1044
• Wer: 85.2539

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• lr_scheduler_warmup_steps: 1000
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.5768	1.0	1424	0.2062	98.4412
0.1775	2.0	2848	0.1505	89.7549
0.1321	3.0	4272	0.1304	86.5233
0.109	4.0	5696	0.1184	87.7905
0.0935	5.0	7120	0.1108	83.8661
0.0815	6.0	8544	0.1072	85.3635
0.0722	7.0	9968	0.1058	84.4405
0.0644	8.0	11392	0.1049	82.3862
0.0575	9.0	12816	0.1049	84.2761
0.0521	9.9933	14230	0.1044	85.2539

Framework versions

• Transformers 4.51.3
• Pytorch 2.2.1+cu121
• Datasets 3.5.0
• Tokenizers 0.21.1