Skip to content
RobotWorld
ZH
← Back to Videos

ndm-gym: Implementation Factors for Model-Based Robot Learning

Loading video
Loading video

ndm-gym isolates implementation choices in model-based reinforcement learning and world models, comparing tracking quality, failure rates and training speed.

ndm-gymModel-Based RLWorld ModelsHumanoid Motion TrackingOpen SourceBPTTRobot Learning
Category: reinforcement-learning
Author: @frankzydou
Date: 2026-11-06T00:00:00
Duration: 22.4s
Source: https://x.com/frankzydou/status/2118773030344392936
Reference: https://github.com/Frank-ZY-Dou/ndm-gym

Related Videos

QF3: Fast Off-Policy RL for Flow Policies with Filtered Q-Gradients
0:06

QF3: Fast Off-Policy RL for Flow Policies with Filtered Q-Gradients

@robotsdigest · 220 views · 2026-10-11
QF3Flow Matching
WARL: Wrench-Augmented Reinforcement Learning (IROS 2026)
0:31

WARL: Wrench-Augmented Reinforcement Learning (IROS 2026)

@kyoneda777 · 432 views · 2026-07-28
RLIROS2026
Frameboy Explores Atlas Worlds at 15 FPS
0:46

Frameboy Explores Atlas Worlds at 15 FPS

@theworldlabs · 71 views · 2026-11-05
World ModelsInteractive 3D
ID-WM Plans Thousands of Steps with Hierarchical World Models
0:07

ID-WM Plans Thousands of Steps with Hierarchical World Models

@GaoyueZhou · 125 views · 2026-11-04
World ModelsLong-Horizon Planning
3D World Simulator Runs Interactive Robot Demos in Browser
0:47

3D World Simulator Runs Interactive Robot Demos in Browser

@YunzhuLiYZ · 136 views · 2026-11-03
World Models3D Interaction
3D World Simulator Predicts Geometry and Touch
0:11

3D World Simulator Predicts Geometry and Touch

@robotsdigest · 94 views · 2026-11-03
World Models3D Scene Prediction

As an Amazon Associate, we earn from qualifying purchases.