# InternRobotics/PointLLM

[ECCV 2024 Best Paper Candidate & TPAMI 2025] PointLLM: Empowering Large Language Models to Understand Point Clouds

Repository: https://github.com/InternRobotics/PointLLM
Canonical: https://ross.abutalabs.com/products/pointllm
Homepage: https://runsenxu.com/projects/PointLLM
Language: Python
License Family: other
Topics: 3d, chatbot, foundation-models, gpt-4, large-language-models, llama, multimodal, objaverse, point-cloud, representation-learning, vision-and-language, pointllm
Last push: 2026-05-15T07:35:59+00:00

## Health v2 (maintenance only)
Score: 65/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 82, release rhythm 35, longevity 79
- inputs: {"age_days": 1112, "days_push": 110, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1051, forks 58 (observed 2026-08-28T04:03:23.123119+00:00)

## What it is
PointLLM is a multimodal large language model that understands colored 3D point clouds of objects, built on a point cloud encoder fused with an LLM (LLaMA-based). It ships with a 660K+70K point-text instruction dataset, two-stage training code, and benchmarks for generative 3D classification and captioning.

## Use cases
- chat with an LLM about 3D point clouds of objects
- generate captions describing 3D object point clouds
- classify 3D objects generatively from point clouds
- train a multimodal LLM on point-text instruction data
- evaluate 3D perception of large language models
- understand geometry and appearance without depth ambiguity

## When to choose
- you need LLM-based understanding of raw 3D point clouds rather than 2D images
- you want a research foundation model and benchmark for 3D-language tasks
- you have point cloud data and want instruction-following dialogue or captioning

## When to avoid
- you need production-ready, commercially licensed software (no license is provided)
- you lack GPU resources for large multimodal model inference or training
- you need 2D image or video understanding rather than point clouds

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, machine-learning, nlp, computer-vision, chatbot
- domain: large-language-models, computer-vision, artificial-intelligence, deep-learning
- platform: python
- tags: point-cloud, multimodal, 3d-understanding, instruction-tuning, llama, research-model, eccv-2024, gpu, linux

## Member repositories
- InternRobotics/PointLLM (main) score 65

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:23.123119+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 2026-08-30T06:59:47.161777+00:00, confidence not recorded.
  - readme: https://github.com/InternRobotics/PointLLM (fetched 2026-08-28T04:03:23.123119+00:00, sha c8719a47de3f)
  - homepage: https://runsenxu.com/projects/PointLLM (fetched 2026-08-29T13:01:30.771055+00:00, sha d7a0964125d7)
- Data as of 2026-08-30T08:39:29.467469+00:00.