Qwen2-VL Training

[rburchcp]

Hi, Qwen2-VL is mentioned in the README but there does not appear to be any code in the repo that allows us to use Qwen2-VL in training or inference. My understanding is that Qwen2-VL is the only model among the three mentioned that works on videos, and that's primarily why I'm interested in that model. Do y'all plan on introducing code soon that allows Qwen2-VL training on videos and associated descriptions?