whisper_new_ver5

This model is a fine-tuned version of Curiousfox/whisper_new_ver4 on the common_voice_12_0 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7340
• Wer: 86.5922

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-06
• train_batch_size: 8
• eval_batch_size: 16
• 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: 1000
• training_steps: 8000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3476	0.4073	800	0.5630	86.3890
0.2172	0.8147	1600	0.6254	85.6780
0.1864	1.2220	2400	0.6920	86.2875
0.1359	1.6293	3200	0.6817	84.3575
0.123	2.0367	4000	0.7072	83.7481
0.0894	2.4440	4800	0.7293	85.8812
0.086	2.8513	5600	0.7340	86.5922

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1