alexshengzhili/Llava-Graph-ocr-ft-on-instruct150k

This model is obtained first

1. Feature alignment based on SciCap. The intermediate output is at this link: https://huggingface.co/alexshengzhili/llava-7bv0-mm-projector-ft-with-ocr-caption-prompted-paragraph
2. Instruction Tuning based on OG llava-provided paper