l3-whisper-base-l2c_v4

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

• Loss: 0.4519
• Wer: 35.5361

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 OptimizerNames.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: 500
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6946	0.0933	1000	0.7307	55.1069
0.5268	0.1866	2000	0.6148	47.0617
0.4649	0.2799	3000	0.5589	42.9543
0.4261	0.3731	4000	0.5232	40.9576
0.4018	0.4664	5000	0.5010	39.9343
0.3831	0.5597	6000	0.4867	38.1922
0.3715	0.6530	7000	0.4710	36.8359
0.3659	0.7463	8000	0.4637	36.4900
0.3547	0.8396	9000	0.4558	35.9241
0.3503	0.9328	10000	0.4519	35.5361

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.1+cu128
• Datasets 3.6.0
• Tokenizers 0.21.4