w_large

This model is a fine-tuned version of openai/whisper-large on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.7156
• Wer: 66.4973

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: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7439	0.4548	1000	0.7228	107.9816
0.6638	0.9095	2000	0.6496	82.4336
0.413	1.3643	3000	0.6292	76.3384
0.4303	1.8190	4000	0.6144	69.9421
0.3339	2.2738	5000	0.6557	71.5521
0.3224	2.7285	6000	0.6553	63.5360
0.1991	3.1833	7000	0.7058	64.2753
0.1752	3.6380	8000	0.7156	66.4973

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu118
• Datasets 3.0.0
• Tokenizers 0.19.1