Official repository of SegDAC: Visual Generalization in Reinforcement Learning via Dynamic Object Tokens
-
Updated
Mar 13, 2026 - Jupyter Notebook
Official repository of SegDAC: Visual Generalization in Reinforcement Learning via Dynamic Object Tokens
Minimum example showing how to setup Singularity with Maniskill3 to run training headless on a cluster.
Very minimal Implementation of Pi-Zero (https://www.pi.website/download/pi0.pdf). Vision Action Flow matching model
Convert ManiSkill HDF5 demonstration trajectories to RLDS / TFDS format for robot learning and VLA data pipelines.
This repository is a modified fork of the official Octo project.It extends the original Octo codebase with tools for adapting Octo to ManiSkill simulation tasks, including ManiSkill-to-RLDS data conversion, Octo fine-tuning scripts, and ManiSkill-based evaluation utilities.
MOPS Data Generation
To associate your repository with the maniskill3 topic, visit your repo's landing page and select "manage topics."