Hi, as @fciannella said, we will be releasing fine-tuning skills to developer to customize their models for domain adaptation.
Meanwhile, we also have nemotron-3.5-asr-streaming fine-tuning tutorial notebook to help you walk through the fine-tuning workflow step-by-step.
Some other supportive materials:
- If you would like to train another tokenizer from scratch for Chinese, please refer to tokenizer notebook
- If you would like to extend the existing tokenizer for more vocab support, please refer to extend tokenizer notebook
