# LargeWorldModel/LWM

Large World Model -- Modeling Text and Video with Millions Context

Repository: https://github.com/LargeWorldModel/LWM
Canonical: https://ross.abutalabs.com/products/lwm
Homepage: https://largeworldmodel.github.io/
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2024-10-19T03:27:38+00:00

## Health v2 (maintenance only)
Score: 25/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 66
- inputs: {"age_days": 937, "days_push": 683, "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 7425, forks 563 (observed 2026-08-28T04:09:59.813894+00:00)

## What it is
Large World Model (LWM) is a family of open-source 7B-parameter multimodal autoregressive transformer models trained on long videos and books with RingAttention, supporting context up to 1M tokens. The repository provides training and inference code for language, image, and video understanding and generation.

## Use cases
- answer questions about hour-long YouTube videos
- retrieve facts across a 1M-token context
- chat with images and videos
- generate videos and images from text
- train transformers on million-length multimodal sequences
- process entire books with a large-context language model

## When to choose
- you need a model that understands very long videos or documents
- you want open weights for million-token-context multimodal research
- you need a reference implementation of RingAttention and masked sequence packing

## When to avoid
- you need a production-ready, well-supported inference stack
- you only need short-context text chat
- you lack multi-GPU resources, since training targets large clusters

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-training, llm-inference, video-processing, nlp
- domain: large-language-models, deep-learning, machine-learning
- platform: python
- tags: multimodal, ring-attention, long-context, vision-language-model, video-understanding, research, video, natural-language-processing, linux, gpu

## Member repositories
- LargeWorldModel/LWM (main) score 25

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:09:59.813894+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 2026-08-29T17:37:44.696502+00:00, confidence not recorded.
  - readme: https://github.com/LargeWorldModel/LWM (fetched 2026-08-28T04:09:59.813894+00:00, sha d166a98023a9)
  - homepage: https://largeworldmodel.github.io/ (fetched 2026-08-29T08:33:21.806377+00:00, sha 15f4e987afc0)
- Data as of 2026-08-30T08:39:29.467469+00:00.