Questions about how to fine-tune fastvlm

[zzzz737]

Thank you very much for open-sourcing such an excellent model. I have the following questions regarding fine-tuning:

1. The repo mentions a train_qwen.py file and references the LLaVA repo code, but I haven't found any configuration scripts for passing parameters to train_qwen.py.
2. I haven't found any code related to the FastViTHD visual encoder in the entire repo. Are there plans to open-source this part later?