SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation

SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation

SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation --> --> SARM2: Multi-Task Stage Aware Reward Modeling for Self Improving Robotic Manipulation Qianzhong Chen 1 , Hau Zheng 1 , Justin Yu 2 , Suning Huang 1 , Jiankai Sun 1 , Ken Goldberg 2 , Pieter Abbeel 2 , Chuan Wen 3 , Yide Shentu 2, 4 , Philipp Wu 4 Mac Schwager 1 , 1 Stanford University, 2 UC Berkeley, 3 Shanghai Jiao Tong University, 4 xdof.ai Accepted to ICLR 2026 --> Try SARM, now native to LeRobot ! Thanks to Hugging Face team for the support! --> Paper --> SARM (prev. Work) arXiv twitter Code (comi

saved by

• Freeman Irabaruta

related reading

• A Unified Model for Motion-Conditioned Robot Co-designtransformer-transformer.github.io
• A Steerable Model with Emergent Capabilitiespi.website
• [2103.06326] S4RL: Surprisingly Simple Self-Supervision for Offline Reinforcement Learningarxiv.org
• A Careful Examination of Large Behavior Models for Multitask Dexterous Manipulationtoyotaresearchinstitute.github.io
• Reward Isn’t Free: Supervising Robot Learning with Language and Video from the Web | SAIL Blogai.stanford.edu
• [2606.09758] Difference-Aware Retrieval Policies for Imitation Learningarxiv.org
• RT-2: Vision-Language-Action Modelsrobotics-transformer2.github.io
• 2310.12921.pdfarxiv.org
• MemER: Scaling Up Memory for Robot Control via Experience Retrievalarxiv.org
• e5b5c402bb7bd5e60bede6961d6fe39e-Paper-Conference.pdfproceedings.iclr.cc
• Interpretable Preferences via Multi-Objective Reward Modeling and Mixture-of-Experts | RLHFlowrlhflow.github.io
• Emergent Dexterity via Diverse Resets and Large-Scale Reinforcement Learningarxiv.org