# X-Square-Robot/wall-x

Building General-Purpose Robots Based on Embodied Foundation Model

Repository: https://github.com/X-Square-Robot/wall-x
Canonical: https://ross.abutalabs.com/products/wall-x
Homepage: https://x2robot.com/en/research
Language: Python
License: Apache-2.0
License Family: permissive
Last push: 2026-08-18T06:26:09+00:00

## Health v2 (maintenance only)
Score: 61/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 98, release rhythm 35, longevity 25
- inputs: {"age_days": 363, "days_push": 15, "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 1236, forks 98 (observed 2026-08-28T04:04:05.199333+00:00)

## What it is
Wall-X is the open-source training and inference stack for X Square Robot's WALL series of embodied foundation models (VLAs) for general-purpose robot manipulation. It includes LeRobot data preparation, model configuration, flow-matching and FAST action branches, serving/evaluation utilities, and CUDA operators compiled at install time.

## Use cases
- train a vision-language-action model for robot manipulation
- fine-tune an embodied foundation model on custom robot demonstration data
- run inference on a real robot arm from a pretrained VLA checkpoint
- prepare LeRobot-format datasets for embodied model training
- evaluate robot manipulation policies in closed-loop rollouts
- adapt a pretrained VLA to new manipulation tasks

## When to choose
- you need an open-source VLA training/inference stack for real-robot manipulation
- you want to fine-tune or deploy WALL-OSS models on your own robot hardware
- you work with LeRobot datasets and want a flow-matching or FAST action head

## When to avoid
- you need a lightweight simulation-only robotics toolkit without GPU-heavy training
- your platform is not Linux with CUDA 12.x GPUs
- you want a plug-and-play robot controller with no model training involved

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, deep-learning, llm-training, llm-inference
- domain: robotics, artificial-intelligence, machine-learning, deep-learning
- platform: python
- tags: embodied-ai, vla, vision-language-action, robot-manipulation, foundation-model, flow-matching, action-tokenizer, lerobot, world-model, linux, gpu

## Member repositories
- X-Square-Robot/wall-x (main) score 61

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:05.199333+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 2026-08-30T08:21:57.050189+00:00, confidence not recorded.
  - readme: https://github.com/X-Square-Robot/wall-x (fetched 2026-08-28T04:04:05.199333+00:00, sha c7db6eeaa1d5)
  - homepage: https://x2robot.com/en/research (fetched 2026-08-29T12:21:45.517296+00:00, sha 8c631abacb77)
  - site_page: https://x2robot.com/en/about (fetched 2026-08-29T12:21:45.520029+00:00, sha dc00c777ec56)
  - site_page: https://x2robot.com/about (fetched 2026-08-29T12:21:45.522245+00:00, sha bfe8a270d5fc)
- Data as of 2026-08-30T08:39:29.467469+00:00.