question about the --version parameter

microsoft / LLaVA-Med

Large Language-and-Vision Assistant for Biomedicine, built towards multimodal GPT-4 level capabilities.

Other

1.48k stars 180 forks source link

question about the --version parameter #69

Open Yanllan opened 5 months ago

Yanllan commented 5 months ago

What values should be set for the --version parameters in the graph? Using it directly causes the following error:

CinKKKyo commented 5 months ago

What values should be set for the --version parameters in the graph? Using it directly causes the following error:

you can type "v0", "1"... any character after the XX.sh like: ./XX.sh v0

Yanllan commented 5 months ago

How does v0,v1 differ, and what are they for? and If I am using qwen1.5, what should I set to? In my experiments, if I use V1 in fine-tuning, I get the following error (pre-training is also V1, but no error) :

Thank you very much for your reply!