# OFA-Sys/ONE-PEACE

A general representation model across vision, audio, language modalities. Paper: ONE-PEACE: Exploring One General Representation Model Toward Unlimited Modalities

Repository: https://github.com/OFA-Sys/ONE-PEACE
Canonical: https://ross.abutalabs.com/products/one-peace
Language: Python
License: Apache-2.0
License Family: permissive
Topics: foundation-models, multimodal, representation-learning, vision-language, audio-language, vision-and-language, vision-transformer, contrastive-loss
Last push: 2024-10-06T04:13:22+00:00

## Health v2 (maintenance only)
Score: 29/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 85
- inputs: {"age_days": 1203, "days_push": 696, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1060, forks 71 (observed 2026-08-28T04:03:25.665239+00:00)

## What it is
ONE-PEACE is a general multimodal representation model that jointly encodes vision, audio, and language modalities without initializing from pretrained vision or language models. The repository provides pretrained checkpoints, fine-tuning and inference scripts, embedding and visual grounding APIs, and a multimodal retrieval demo.

## Use cases
- extract joint embeddings for images, audio, and text
- retrieve images using audio, text, or combined audio+text+image queries
- fine-tune a multimodal model on vision-language tasks like VQA and captioning
- fine-tune on audio classification and audio-language tasks
- locate objects in images with visual grounding
- pretrain a modality-agnostic transformer from scratch

## When to choose
- you need a single model embedding multiple modalities in one shared space
- you want strong zero-shot cross-modal retrieval including unpaired modality combinations
- you need a research foundation model for multimodal representation learning

## When to avoid
- you need a lightweight production embedding service with minimal dependencies
- you only need unimodal text or image models
- you require active community support and frequent updates

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, nlp, image-processing, audio-processing, search-engine
- domain: machine-learning, deep-learning, artificial-intelligence, computer-vision
- platform: python
- tags: multimodal, representation-learning, foundation-model, vision-language, contrastive-learning, zero-shot-retrieval, embeddings, natural-language-processing, audio

## Member repositories
- OFA-Sys/ONE-PEACE (main) score 29

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:25.665239+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 2026-08-30T06:56:54.429287+00:00, confidence not recorded.
  - readme: https://github.com/OFA-Sys/ONE-PEACE (fetched 2026-08-28T04:03:25.665239+00:00, sha 75487d7ea45f)
- Data as of 2026-08-30T08:39:29.467469+00:00.