repo_name

This model is a fine-tuned version of openai/whisper-base on the Low Quality Call Voice (Preprocessed) dataset. It achieves the following results on the evaluation set:

• Loss: 0.5266
• Cer: 25.0296

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: 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
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.6412	0.2398	1000	0.6448	65.6017
0.5415	0.4796	2000	0.5695	29.0593
0.5701	0.7194	3000	0.5394	24.7949
0.5208	0.9592	4000	0.5266	25.0296

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1