Skip to content
RobothecaBeta
Log in

EasyWAM

  • AI model
  • Humanoids and robot learning

A unified research codebase for training and evaluating World Action Models, covering several model designs, backbones and robot benchmarks such as LIBERO and RoboTwin.

  • Non-commercial weights

Made by OpenMOSS (SII), China (checked on github.com/OpenMOSS (opens in a new tab)).

Source on GitHub: github.com/OpenMOSS/EasyWAM (opens in a new tab). Facts read at commit dd5b061 (opens in a new tab).

Licenses

Code
MIT (opens in a new tab)

Read by Robotheca from LICENSE (opens in a new tab) (MIT) at commit dd5b061 (opens in a new tab), on 16 Oct 2026.

Commercial use
No

The weights license does not allow commercial use.

Labels

  • Non-commercial weightsThe weights license (CC-BY-NC-4.0) does not allow commercial use.

What it does

  • Provides a shared workflow for model implementation, data processing, distributed training, LoRA fine-tuning and large-scale evaluation.
  • Offers five model variants (Unified, MoT, MoT-Joint, MoT-IDM, Hidden) that can be trained fully or with LoRA.
  • Supports Wan2.2-TI2V-5B, Cosmos-Predict2.5-2B and FLUX.2 Klein-4B as backbones.
  • Integrates FlashAttention 2/3/4 when installed, falling back to PyTorch SDPA, plus BF16, gradient checkpointing and DeepSpeed ZeRO.
  • Runs evaluations on LIBERO, LIBERO-Plus, LIBERO-Pro, RoboTwin, RoboDojo and RoboCasa365 benchmarks.
  • Uses Hydra configuration with launchers for training and evaluation, including multi-node options.

Written by Robotheca from the project's README (opens in a new tab) at commit dd5b061.

Getting started

Difficulty
Advanced

Maintenance

Status
Active: a commit in the last 90 days
Newest commit
9 Oct 2026 on main
Latest release
None published on GitHub

Checked on GitHub on 16 Oct 2026.

Links

Robots that can run it

No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.

Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.