Hi, for the released model, it calls LlavaQwenModel class. However, the LlavaQwenModel class uses LlavaQwenConfig, which essentially is Qwen2Config, is a LLM config only. So it could not load the released model correctly... Line here:
|
class LlavaQwenModel(LlavaMetaModel, Qwen2Model): |
Could you please help with this?
Hi, for the released model, it calls LlavaQwenModel class. However, the LlavaQwenModel class uses LlavaQwenConfig, which essentially is Qwen2Config, is a LLM config only. So it could not load the released model correctly... Line here:
LLaVA-ST/llava/model/language_model/llava_qwen.py
Line 41 in bacf6d6
Could you please help with this?