TechPulseTelegram
← PrevFri, May 1Next β†’

πŸ€– AI Agent9 items

Latest AI agent research papers, analysis, and improvement insights

semantic_scholarresearch

ClawTrace: Cost-Aware Tracing for LLM Agent Skill Distillation

μ—μ΄μ „νŠΈμ˜ μŠ€ν…λ³„ λΉ„μš© 좔적 및 μŠ€ν‚¬ 증λ₯˜ ν”Œλž«νΌ

semantic_scholarresearch

Revisable by Design: A Theory of Streaming LLM Agent Execution

μ‚¬μš©μž κ°œμž… κ°€λŠ₯ν•œ 슀트리밍 기반 μ—μ΄μ „νŠΈ μ‹€ν–‰ νŒ¨λŸ¬λ‹€μž„

semantic_scholarresearch

Seeing the Whole Elephant: A Benchmark for Failure Attribution in LLM-based Multi-Agent Systems

닀쀑 μ—μ΄μ „νŠΈ μ‹œμŠ€ν…œμ˜ μž₯μ•  원인 뢄석 및 μ±…μž„ νŒŒμ•… 벀치마크

semantic_scholarresearch

BenchGuard: Who Guards the Benchmarks? Automated Auditing of LLM Agent Benchmarks

LLM이 벀치마크 자체 였λ₯˜λ₯Ό κ°μ‹œν•˜λŠ” μžλ™ κ°μ‹œ ν”„λ ˆμž„μ›Œν¬

semantic_scholarresearch

ITAS: A Multi-Agent Architecture for LLM-Based Intelligent Tutoring

λŒ€ν•™ κ°•μ’Œμ— μ‹€μ œ 배포된 μ „λ¬Έκ°€ μ—μ΄μ „νŠΈ 기반 νŠœν„°λ§ μ‹œμŠ€ν…œ

semantic_scholarresearch

CUJBench: Benchmarking LLM-Agent on Cross-Modal Failure Diagnosis from Browser to Backend

λΈŒλΌμš°μ €-λ°±μ—”λ“œ 크둜슀λͺ¨λ‹¬ μž₯μ•  진단 μ—μ΄μ „νŠΈ 벀치마크

semantic_scholarresearch

LLM-Powered Multi-Agent Framework for Automated PPV Prediction in Tunnel Blasting

닀쀑 μ—μ΄μ „νŠΈ ν˜‘λ ₯ 기반 μžλ™ν™”λœ ML νŒŒμ΄ν”„λΌμΈ ꡬ좕

semantic_scholarresearch

Cutscene Agent: An LLM Agent Framework for Automated 3D Cutscene Generation

κ²Œμž„ 3D μ»·μ‹  μžλ™ 생성을 μœ„ν•œ LLM μ—μ΄μ „νŠΈ ν”„λ ˆμž„μ›Œν¬

semantic_scholarresearch

How Personal Characteristics Shape User Exploration of Diverse Movie Recommendations with a LLM-Based Multi-Agent System

개인 νŠΉμ„±μ— λ”°λ₯Έ 닀쀑 μ—μ΄μ „νŠΈ μΆ”μ²œ μ‹œμŠ€ν…œμ˜ μ‚¬μš©μž 탐색 뢄석