SPACE (EMNLP 2026): learning chunk boundaries from trajectories cuts agent LLM rounds by up to 78.9%
rohanpaul_ai · x · 2026-09-04
The arXiv paper behind rohanpaul's thread (EMNLP 2026 camera-ready) details SPACE. ReAct-style one-action-per-round wastes LLM rounds on long-horizon tasks, while naive RL for variable-length action chunks collapses to single actions or over-commits — both from failing to learn chunk boundaries.
SPACE induces two-level programmatic skills from successful trajectories, uses subskill boundaries as chunk-boundary supervision, and distills the temporal structure into a primitive-chunk policy via hybrid on-/off-policy optimization with chunk-aware credit assignment. On ALFWorld and ScienceWorld it beats the strongest baselines by 7.0%-31.3% success and cuts decision rounds by up to 78.9%.
Related event: SPACE Cuts Agent LLM Calls by 78.9% While Boosting Success(3 posts)→
More from coding & agent
- Burkov: if you don't interrogate Codex line-by-line at handoff, you get 80%-done work — burkov · 2026-10-21
- Mintlify launches Autopilot, an always-on agent that updates docs as you ship — stuffyokodraws · 2026-10-21
- Skills vs MCP vs RAG vs Memory: the four knowledge layers production agents need — MaryamMiradi · 2026-10-21
- When customer service bans AI agents, one goes incognito to fight Xfinity — MoonL88537 · 2026-10-21
- Every.to publishes raw interviews so readers can agent-check quotes in its Codex story — every · 2026-10-21
- One-prompt Blender: Voyager + Claude Opus 5.5 auto-connects to Blender and Unity — ravisparikh · 2026-10-21