Submission checkpoints for the RoboSynChallenge. The runnable code and evaluation commands are in the GitHub repository. The checkpoint path in the second column is the weight used for the corresponding row. Success rates: official random configuration, official scripts/evalpolicy.py (expert pre-check on), RTX PRO 6000, five seed streams x 100 episodes. sampleloading has two policies: π0.5 (policy/fmc3no1) and policy/perceptionclosedloop (image estimators plus a closed-loop scripted expert; its two networks are the files under sampleloading/perceptionclosedloop/v3, it reads no object state from the simulator). The manipulatepipette entry is a 10-stream validation (1000 episodes) of the same…
Independent publisher
Puheliang
puheliang
Embodied AI, Vision-Language-Action (VLA) Models, Robot Learning, Reinforcement Learning, Imitation Learning, Multimodal Learning, Robot Manipulation, Tactile Perception, Vision-Language Models, Humanoid Robotics
Models in Library1
Datasets in Library0
Models on Hugging Face5
Followers3
Compare the published evidence
This profile connects the 1 model and dataset record shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.