whisper-small-id

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: 0.4034
• Wer: 13.6494

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: 64
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1014	4.95	500	0.2583	13.6355
0.0058	9.9	1000	0.3169	13.2851
0.0017	14.85	1500	0.3488	13.2251
0.001	19.8	2000	0.3639	13.3542
0.0007	24.75	2500	0.3756	13.5018
0.0005	29.7	3000	0.3844	13.5617
0.0005	34.65	3500	0.3922	13.6401
0.0004	39.6	4000	0.3981	13.6032
0.0003	44.55	4500	0.4019	13.6632
0.0003	49.5	5000	0.4034	13.6494

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2