whisper-medium-nya

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

• Loss: 0.8813
• Wer: 37.8500

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: 2
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.1179	0.25	500	1.8786	126.6155
0.5971	0.49	1000	1.2148	57.5546
0.4307	0.74	1500	1.0730	44.9380
0.3661	0.99	2000	0.9695	41.0278
0.299	1.23	2500	0.9517	38.9014
0.2619	1.48	3000	0.9244	36.2197
0.2476	1.72	3500	0.8762	41.6657
0.2262	1.97	4000	0.8813	37.8500

Framework versions

• Transformers 4.27.0.dev0
• Pytorch 1.13.1+cu116
• Datasets 2.9.0
• Tokenizers 0.13.2