# haotian-liu/LLaVA

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.

Repository: https://github.com/haotian-liu/LLaVA
Canonical: https://ross.abutalabs.com/products/llava
Homepage: https://llava.hliu.cc
Language: Python
License: Apache-2.0
License Family: permissive
Topics: gpt-4, chatbot, chatgpt, llama, multimodal, llava, foundation-models, instruction-tuning, multi-modality, visual-language-learning, llama-2, llama2, vision-language-model
Last push: 2024-08-12T09:52:38+00:00

## Health v2 (maintenance only)
Score: 20/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 8, longevity 88
- inputs: {"age_days": 1234, "days_push": 751, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 25000, forks 2778 (observed 2026-08-28T04:11:37.761631+00:00)

## What it is
LLaVA (Large Language and Vision Assistant) is an open-source multimodal large language model framework implementing visual instruction tuning, combining vision encoders with LLMs like LLaMA for image understanding and conversation. It provides training code, pretrained checkpoints, and a demo toward GPT-4V-level vision-language capabilities.

## Use cases
- build a chatbot that understands images
- fine-tune a vision-language model on custom data
- run multimodal LLM inference locally
- ask questions about images with an open-source model
- train a multimodal model with visual instruction tuning
- evaluate large multimodal models on benchmarks

## When to choose
- you need an open-source, self-hosted alternative to GPT-4V for image+text tasks
- you want to fine-tune or extend a proven vision-language model
- you need pretrained multimodal checkpoints with permissive Apache-2.0 licensing

## When to avoid
- you only need text-only LLM capabilities
- you lack GPU resources for training or inference
- you need a fully managed multimodal API rather than self-hosted models

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, llm-training, machine-learning, deep-learning, chatbot, transformers
- domain: large-language-models, computer-vision, artificial-intelligence, deep-learning
- platform: python
- tags: multimodal, vision-language-model, visual-instruction-tuning, gpt-4v, foundation-models, image-understanding, natural-language-processing, gpu, linux, docker

## Member repositories
- haotian-liu/LLaVA (main) score 20

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:37.761631+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T16:56:07.675260+00:00, confidence not recorded.
  - readme: https://github.com/haotian-liu/LLaVA (fetched 2026-08-28T04:11:37.761631+00:00, sha 8b8b4b4ea008)
- Data as of 2026-08-30T08:39:29.467469+00:00.