π€ AI Agent9 items
Latest AI agent research papers, analysis, and improvement insights
EPC: A Standardized Protocol for Measuring Evaluator Preference Dynamics in LLM Agent Systems
μμ΄μ νΈμ νΌλλ°± 루νμμ νκ°μ νΈν₯μ΄ μ νλλ νμμ μΈ‘μ νλ νμ€νλ νλ‘ν μ½
Self-Evolving World Models for LLM Agent Planning
λ°°ν¬ μμ μ μλ μ§ννλ©° μμ΄μ νΈμ μ₯κΈ° κ³ν μ νμ±μ ν₯μνλ μΈκ³λͺ¨λΈ νλ μμν¬
Calibrating the Evaluator: Does Probability Calibration Mitigate Preference Coupling in LLM Agent Feedback Loops?
νλ₯ 보μ κΈ°λ²μΌλ‘ μμ΄μ νΈ νμ΅ μ νκ°μμ 체κ³μ νΈν₯μ μ€μΌ μ μλμ§ μ€μ¦
Capability Gates Are Not Authorization: Confused-Deputy Failures in LLM Agent Frameworks
μ£Όμ μμ΄μ νΈ νλ μμν¬μμ λꡬ νΈμΆ μ μ¬μΈμ¦ λΆμ¬λ‘ μΈν 보μ μν λΆμ
A Systematic Approach to Multi-Agent AI from Advanced Regulatory Control Theory: Safe and Auditable LLM Operator Agents for Process Control
μ μ΄ μ΄λ‘ μ리λ₯Ό λ©ν°μμ΄μ νΈ LLM μ€κ³μ μ μ©νμ¬ μμ μ±κ³Ό κ°μ¬μ± ν보
SmoothAgent: Efficient Long-Horizon LLM-Based Agent Serving with Lookahead Context Engineering
KV μΊμ ν¨μ¨μ±μ λμ΄λ 컨ν μ€νΈ κ΄λ¦¬λ‘ λ©ν°ν΄ μμ΄μ νΈ μν¬νλ‘μ° μ±λ₯ μ΅μ ν
When Latent Agents Lie: KV-Cache Integrity in Multi-Agent LLM Collaboration
λ©ν°μμ΄μ νΈ νμ μμ KV μΊμ λ³μ‘°λ₯Ό ν΅ν μ¨κ²¨μ§ μν 곡격 λ¬Έμ λΆμ
TO-Master: an LLM-agent framework for automated topology optimization
μ νμμ κΈ°λ° μμ μ΅μ νμ μ 체 μν¬νλ‘μ°λ₯Ό LLM μμ΄μ νΈλ‘ μλννλ νλ μμν¬
An LLM-agent-based framework for calculating nodal carbon intensity in regional power systems.
LLM μμ΄μ νΈλ₯Ό ν΅ν΄ μ λ ₯λ§μ λ°μ΄ν° ν΅ν© λ° νμκ°λ λΆμμ μλννλ μμ€ν