How do I load a pre-trained model for fine-tuning?

[LORDPQK]

How do I load a pre-trained model for fine-tuning?

[lsztzp]

I've encountered the same issue as well. I'm not sure if you've managed to resolve it. I'm thinking of adding an additional LORA on top of the existing model to make it suitable for my task.

[CurryYuan]

Can we just change the path here?

Sa2VA/projects/llava_sam2/configs/sa2va_4b.py

Line 31 in c490564

path = './pretrained/InternVL2_5-4B'

[LORDPQK]

Can we just change the path here?

Sa2VA/projects/llava_sam2/configs/sa2va_4b.py

Line 31 in c490564

path = './pretrained/InternVL2_5-4B'

can i ask another problem,did you meet "Floating point exception (core dumped)" when using demo.py or quick start? @CurryYuan

[CurryYuan]

I did not meet this problem. The inference is ok.

Can we just change the path here?
Sa2VA/projects/llava_sam2/configs/sa2va_4b.py
Line 31 in c490564
path = './pretrained/InternVL2_5-4B'

can i ask another problem,did you meet "Floating point exception (core dumped)" when using demo.py or quick start? @CurryYuan

[CurryYuan]

However I got error FileNotFoundError: can't find *_optim_states.pt files in directory '/huggingface/ByteDance/Sa2VA-4B' when set the pretrained_pth to the released weight.

Sa2VA/projects/llava_sam2/configs/sa2va_4b.py

Line 32 in c490564

pretrained_pth = None