seniruk/whisper-medium-si

Sinscribe Medium

This model is a fine-tuned version of openai/whisper-medium on the Sinhala CSV + FLACs dataset. It achieves the following results on the evaluation set:

• Loss: 0.0751
• Wer: 28.7421

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: 5e-06
• train_batch_size: 4
• eval_batch_size: 16
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 16
• 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: 250
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1168	0.3307	3000	0.1161	37.6130
0.0676	0.3307	6000	0.1078	35.7945
0.0873	0.6613	9000	0.0870	32.0812
0.0813	0.9920	12000	0.0786	30.1238
0.0545	1.3226	15000	0.0779	29.2746
0.0494	1.6533	18000	0.0751	28.7421

Framework versions

• Transformers 4.54.0
• Pytorch 2.3.0+cu121
• Datasets 3.6.0
• Tokenizers 0.21.4