Why is there no preprocessor_config.json file in the folder?
#3
by
FloSophorae
- opened
Isn't this model a multimodal model? Why is there no preprocessor_config.json file?
This model was finetuned from https://huggingface.co/lmms-lab/llava-onevision-qwen2-7b-mid-stage-a4, which didn't have the preprocessor_config.json and the training code didn't automatically save it. However, the training code did use the config of the SigLIP encoder from https://huggingface.co/google/siglip-so400m-patch14-384/blob/main/preprocessor_config.json. This was also pretty much the same preprocessor_config.json in the llava-ov's single image and one-vision checkpoint.
ok，thank you