Reproduce “LLaVA-LE: Large Language-and-Vision Assistant for Lunar Exploration” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Gokce Inal, Pouyan Navard, Alper Yilmaz
LLaVA-LE adapts a general vision-language assistant to lunar geological image interpretation by training it first on terrain captions and then on question-answer examples. Captions use panchromatic imagery with gravity and slope data, but evaluation supplies only an image and question. On 190 held-out questions, Stage 2 scores 0.921 overall, versus 0.278 and 0.443 for Base LLaVA and Stage 1.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.