# salesforce/LAVIS

LAVIS - A One-stop Library for Language-Vision Intelligence

Repository: https://github.com/salesforce/LAVIS
Canonical: https://ross.abutalabs.com/products/lavis
Language: Jupyter Notebook
License: BSD-3-Clause
License Family: permissive
Topics: deep-learning, deep-learning-library, image-captioning, salesforce, vision-and-language, vision-framework, vision-language-pretraining, vision-language-transformer, visual-question-anwsering, multimodal-datasets, multimodal-deep-learning
Last push: 2026-06-02T18:14:49+00:00

## Health v2 (maintenance only)
Score: 61/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 85, release rhythm 8, longevity 100
- inputs: {"age_days": 1471, "days_push": 92, "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 11262, forks 1108 (observed 2026-08-28T04:10:46.565650+00:00)

## What it is
LAVIS is a Python library from Salesforce AI Research providing a unified toolkit for language-vision (multimodal) intelligence, including pretrained models like BLIP-2, InstructBLIP, and BLIP-Diffusion. It supports tasks such as image captioning, visual question answering, retrieval, and text-to-image generation with datasets, benchmarks, and training utilities.

## Use cases
- generate captions for images
- build a visual question answering model
- run zero-shot vision-language inference with BLIP-2
- do subject-driven text-to-image generation
- pretrain or fine-tune a vision-language transformer
- retrieve images using text queries
- evaluate models on multimodal benchmarks

## When to choose
- you need state-of-the-art pretrained vision-language models in one library
- you want to fine-tune or instruction-tune multimodal models like InstructBLIP
- you're doing research on image-text tasks like captioning, VQA, or retrieval

## When to avoid
- you only need single-modality NLP or computer vision without cross-modal tasks
- you need a production inference server rather than a research library
- you work outside PyTorch/Python ecosystems

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, image-processing, nlp, llm-inference, stable-diffusion
- domain: deep-learning, computer-vision, large-language-models, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: multimodal, vision-language, image-captioning, visual-question-answering, blip, pretrained-models, research-library, natural-language-processing, gpu

## Member repositories
- salesforce/LAVIS (main) score 61

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:46.565650+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 2026-08-29T17:16:24.417723+00:00, confidence not recorded.
  - readme: https://github.com/salesforce/LAVIS (fetched 2026-08-28T04:10:46.565650+00:00, sha acbb7b95be1b)
- Data as of 2026-08-30T08:39:29.467469+00:00.