l3-whisper-tiny-l2v191hc_v5

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

• Loss: 0.8696
• Wer: 64.5463

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: 1
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.1072	200	1.5447	85.0445
No log	0.2145	400	1.1390	76.1530
1.3903	0.3217	600	1.0303	72.4986
1.3903	0.4290	800	0.9720	69.5034
0.7774	0.5362	1000	0.9302	67.0499
0.7774	0.6434	1200	0.9055	65.4626
0.7774	0.7507	1400	0.8864	65.0380
0.6921	0.8579	1600	0.8766	64.4182
0.6921	0.9651	1800	0.8696	64.5463

Framework versions

• Transformers 4.51.3
• Pytorch 2.2.1
• Datasets 3.6.0
• Tokenizers 0.21.2