Unofficial implementation of "Generative Modeling via Drifting" on PushT, built on Diffusion Policy
-
Updated
Mar 25, 2026 - Python
Unofficial implementation of "Generative Modeling via Drifting" on PushT, built on Diffusion Policy
A beginner-friendly VLA project starter: CPU smoke, ACT + PushT-style imitation learning, rollout eval, and resume-ready docs.
Closed-loop Diffusion Policy vs ACT comparison on PushT using LeRobot
Push-T VW2-DirectAct falsification code and artifacts, including subgoal-distillation hard-stop results.
Matched-compute diffusion policy study: local teacher fidelity fails to predict closed-loop PushT success.
Controlled study of ego/world latent factorization in a JEPA world model for planning on PushT.
Small reproducible robot-learning and policy-evaluation experiments with provenance and honest metrics
PushT world-model planning with a frozen dynamics model: look ahead to choose actions, then re-anchor and replan when prediction and reality diverge.
PushT experiments for AF-LeWM-lite, a factorized-latent LeWorldModel variant
A fixed-observation Push-T counterexample to multimodality and MIP≈Flow claims in Much Ado About Noising.
Instrumented LeWorldModel + PushT lab for latent world-model learning, CEM planning, and MPC visualization.
To associate your repository with the pusht topic, visit your repo's landing page and select "manage topics."