HASTE reaches 77.3% medal rate on MLE-Bench Lite (22 competitions) and 100% in an 8-competition ablation by loading a fixed 159-skill inventory in tiered fashion rather than flatly.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.AI 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Persistent notebook memory enables reliable language emergence in LLM agents across channel capacities, outperforming stateless and rolling-context setups in Lewis signaling games.
citing papers explorer
-
Why Solve It Twice? Hierarchical Accumulation of Skills for Transfer-Efficient ML Engineering
HASTE reaches 77.3% medal rate on MLE-Bench Lite (22 competitions) and 100% in an 8-competition ablation by loading a fixed 159-skill inventory in tiered fashion rather than flatly.
-
From Signals to Structure: How Memory Architecture Drives Language Emergence in LLM Agents
Persistent notebook memory enables reliable language emergence in LLM agents across channel capacities, outperforming stateless and rolling-context setups in Lewis signaling games.