Reproduce “HeLM: Hierarchical Explicit Language Memory for Non-Markovian Manipulation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Gyeonghun Kim, Donghoon Kim, Minsoo Choi, Suyun Lee
HeLM gives robots a high-level policy that records predefined task-relevant events in text and updates progress when events or instruction changes occur. It uses that memory to select commands, while a low-level policy executes them. Across five real-world intra-episode tasks, mean subtask success was 85.7% versus 72.9% for Keyframe-HLP; this metric excludes order and termination.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.