finetune_v16

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: 0.2744
• Wer: 154.1280

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: 4
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• 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	6.1538	10	0.3186	11.0165
No log	12.3077	20	0.2781	11.8679
No log	18.4615	30	0.2727	11.6873
No log	24.6154	40	0.2708	31.8369
0.0863	30.7692	50	0.2698	59.2363
0.0863	36.9231	60	0.2722	158.5655
0.0863	43.0769	70	0.2739	156.8627
0.0863	49.2308	80	0.2744	154.1280

Framework versions

• Transformers 4.42.4
• Pytorch 2.2.0
• Datasets 2.20.0
• Tokenizers 0.19.1