Blgn94/whisper-large-v3-turbo-mn-lora
0134
Update to 2-epoch model: eval loss 0.330->0.283, WER 35.5%->27.4% (200-clip subset)
Add merged Mongolian fine-tune (whisper-large-v3-turbo + LoRA) with model card
initial commit
Update to 2-epoch model: eval loss 0.330->0.283, WER 35.5%->27.4% (200-clip subset)
Add merged Mongolian fine-tune (whisper-large-v3-turbo + LoRA) with model card
initial commit