MagiBoss/Whisper-Small-Thai-Lora

Whisper Small Thai Lora - Magi Boss

This model is a fine-tuned version of openai/whisper-small on the Common Voice 11.0 dataset (Training Set 20000 row, Validation Set 500 row). It achieves the following results on the evaluation set:

• Loss: 0.8313
• WER: 1.1186
• TER: 111.8553
• ChrF: 66.9454
• CER: 0.2283
• BLEU: 3.6586
• SubER: 1.1628

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 32
• eval_batch_size: 16
• seed: 42
• optimizer: AdamW
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 25
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Ter	Chrf	Cer	Bleu	SubER
0.1990	0.4	250	0.8732	1.1969	119.6879	65.8239	0.2487	4.2583	1.2745
0.1902	0.8	500	0.8353	1.1232	112.3175	66.5794	0.2430	3.9823	1.1698
0.1873	1	625	0.8313	1.1186	111.8553	66.9454	0.2283	3.6586	1.1628

Framework versions

• PEFT 0.12.1.dev0
• Transformers 4.45.0.dev0
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1