AI chatbot

LLaVA is a new end-to-end trained large-scale multimodal model that combines visual encoders and Vicuna to achieve impressive chat capabilities, mimicking the spirit of multimodal GPT-4 and achieving new state-of-the-art accuracy in scientific question answering. Usage scenarios of LLaVA include multi-modal chatting for daily user applications and multi-modal reasoning in the scientific field. LLaVA's data, code, and checkpoints are for research use only and are licensed under the CLIP, LLaMA, Vicuna, and GPT-4 licenses.