finetune_v7

This model is a fine-tuned version of openai/whisper-large-v3 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6387
• Wer: 81.7276

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: 8
• eval_batch_size: 4
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 4
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 5
• training_steps: 80
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	6.6667	10	0.6616	27.2425
No log	13.3333	20	0.6074	28.5714
No log	20.0	30	0.6377	28.5714
No log	26.6667	40	0.6221	32.5581
0.2362	33.3333	50	0.6255	103.9867
0.2362	40.0	60	0.6309	36.2126
0.2362	46.6667	70	0.6362	37.2093
0.2362	53.3333	80	0.6387	81.7276

Framework versions

• Transformers 4.42.3
• Pytorch 2.2.0
• Datasets 2.20.0
• Tokenizers 0.19.1