You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When I want to run [ visual_instruction_tuning] , as you say in the README.md, "Download the pretrained projector weights from liuhaotian/llava-v1.5-mlp2x-336px-pretrain-vicuna-7b-v1.5 and put it under ./checkpoints/ " .
I have some question:
1st : The llava-v1.5-mlp2x-336px-pretrain-vicuna-7b-v1.5 only has a 42MB mm_projector.bin, the other pretrained weigths such as pytorch_model-00001-of-00002.bin, Where should I put it after I download it?
2nd : I guess I need git clone https://huggingface.co/liuhaotian/llava-v1.5-7b/tree/main, the pretrained model should put under ./checkpoint/, Is it right? and rename this folder to "llava-v1.5-7b-pretrain" , Because it is set up this way in your shell script [Dora_7b.sh
line17].
3rd : the 42MB mm_projector.bin in "llava-v1.5-7b" of 2nd gitted should be replaced by the 42MB mm_projector.bin of 1st?
The text was updated successfully, but these errors were encountered:
When I want to run [ visual_instruction_tuning] , as you say in the README.md, "Download the pretrained projector weights from liuhaotian/llava-v1.5-mlp2x-336px-pretrain-vicuna-7b-v1.5 and put it under ./checkpoints/ " .
I have some question:
1st : The llava-v1.5-mlp2x-336px-pretrain-vicuna-7b-v1.5 only has a 42MB mm_projector.bin, the other pretrained weigths such as pytorch_model-00001-of-00002.bin, Where should I put it after I download it?
2nd : I guess I need git clone https://huggingface.co/liuhaotian/llava-v1.5-7b/tree/main, the pretrained model should put under ./checkpoint/, Is it right? and rename this folder to "llava-v1.5-7b-pretrain" , Because it is set up this way in your shell script [Dora_7b.sh
line17].
3rd : the 42MB mm_projector.bin in "llava-v1.5-7b" of 2nd gitted should be replaced by the 42MB mm_projector.bin of 1st?
The text was updated successfully, but these errors were encountered: