CLEAR trains a context augmentation model via agentic contrastive reflection, SFT, and RL to generate tailored context for LLM agents, improving task completion from 72.62% to 81.15% on AppWorld and rewards from 0.68 to 0.74 on WebShop.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CLEAR: Context Augmentation from Contrastive Learning of Experience via Agentic Reflection
CLEAR trains a context augmentation model via agentic contrastive reflection, SFT, and RL to generate tailored context for LLM agents, improving task completion from 72.62% to 81.15% on AppWorld and rewards from 0.68 to 0.74 on WebShop.