parambharat/whisper-small-te

Whisper Small Te - Bharat Ramanathan

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: 0.1863
• Wer: 31.6456

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
0.1637	0.1	500	0.2092	42.9406
0.1459	0.2	1000	0.2025	35.9299
0.1348	0.3	1500	0.1990	35.4917
0.1309	0.4	2000	0.1974	33.7390
0.1253	0.5	2500	0.1974	34.0312
0.1209	0.6	3000	0.1909	32.4732
0.1139	1.05	3500	0.1899	31.7916
0.1043	1.15	4000	0.1868	31.6456
0.0996	1.25	4500	0.1874	31.6943
0.1002	1.35	5000	0.1863	31.6456

Framework versions

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