VladS159/whisper-medium-6000-steps_10_percent_4_speakers_synthetic_data_16_02_2026

Whisper Medium Ro - PEFT

This model is a fine-tuned version of openai/whisper-medium on the Common Voice 17.0 10% Synthetic 4 speakers dataset. It achieves the following results on the evaluation set:

• Loss: 0.1623
• Wer: 8.3516

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: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch_fused 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: 600
• training_steps: 6000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.2456	0.2267	1000	0.2116	10.6105
0.2176	0.4533	2000	0.1821	9.4078
0.2059	0.6800	3000	0.1719	8.8431
0.1889	0.9066	4000	0.1665	8.5867
0.194	1.1333	5000	0.1632	8.3455
0.1729	1.3599	6000	0.1623	8.3516

Framework versions

• PEFT 0.18.1.dev0
• Transformers 4.57.1
• Pytorch 2.9.1+rocm6.4
• Datasets 3.6.0
• Tokenizers 0.22.1