/LLM-Agent-Papers-Collection

A collection of papers on Large Language Model Agents

LLM-Agent-Papers-Collection

A collection of papers on Large Language Model Agents

Our collection of papers focuses on various aspects of LLM-Agent Design Patterns. The design patterns can be abstracted and illustrated as shown in the figure below. The collection is categorized into actions, memory, reasoning, agents, and multi-agent collaboration.

agent_design_pattern

Besides the aspects shown in the figure, this repository will also include: overview of LLM-Agent, benchmarks, prompt design.

Agent Design Patterns

Actions

  1. Tulip Agent -- Enabling LLM-Based Agents to Solve Tasks Using Large Tool Libraries 2024.7
  2. Reason for Future, Act for Now: A Principled Architecture for Autonomous LLM Agents ICML2024
  3. Multi-Programming Language Sandbox for LLMs 2024.10

Memory

  1. 360° REA: Towards A Reusable Experience Accumulation with Assessment for Multi-Agent System ACL2024
  2. Symbolic Working Memory Enhances Language Models for Complex Rule Application 2024.8
  3. MemoRAG: Moving towards Next-Gen RAG Via Memory-Inspired Knowledge Discovery 2024.9
  4. AGENT WORKFLOW MEMORY 2024.9

Reasoning

  1. AgentGen: Enhancing Planning Abilities for Large Language Model based Agent via Environment and Task Generation 2024.8
  2. Optimus-1: Hybrid Multimodal Memory Empowered Agents Excel in Long-Horizon Tasks 2024.8
  3. Position: LLMs Can’t Plan, But Can Help Planning in LLM-Modulo Frameworks ICML2024
  4. Rethinking the Bounds of LLM Reasoning: Are Multi-Agent Discussions the Key? ACL2024
  5. Strategic Chain-of-Thought: Guiding Accurate Reasoning in LLMs through Strategy Elicitation 2024.9
  6. Self-Harmonized Chain of Thought 2024.9
  7. Iteration of Thought: Leveraging Inner Dialogue for Autonomous Large Language Model Reasoning 2024.9
  8. Textualized Agent-Style Reasoning for Complex Tasks by Multiple Round LLM Generation 2024.9
  9. Logic-of-Thought: Injecting Logic into Contexts for Full Reasoning in Large Language Models 2024.9
  10. BEATS: OPTIMIZING LLM MATHEMATICAL CAPABILITIES WITH BACKVERIFY AND ADAPTIVE DISAMBIGUATE BASED EFFICIENT TREE SEARCH 2024.9
  11. HDFLOW: ENHANCING LLM COMPLEX PROBLEMSOLVING WITH HYBRID THINKING AND DYNAMIC WORKFLOWS 2024.9
  12. CoMAT: Chain of Mathematically Annotated Thought Improves Mathematical Reasoning 2024.10
  13. Optimizing Chain-of-Thought Reasoning: Tackling Arranging Bottleneck via Plan Augmentation 2024.10

Agent

  1. From LLMs to LLM-based Agents for Software Engineering: A Survey of Current, Challenges and Future 2024.8
  2. MathLearner: A Large Language Model Agent Framework for Learning to Solve Mathematical Problems 2024.8
  3. Executable Code Actions Elicit Better LLM Agents ICML2024
  4. Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents 2024.8
  5. SciAgents: Automating scientific discovery through multi-agent intelligent graph reasoning 2024.9

Multi-agent Collaboration

  1. Experiential Co-Learning ACL2024
  2. Should we be going MAD? A Look at Multi-Agent Debate Strategies for LLMs ICML2024
  3. Improving Factuality and Reasoning in Language Models through Multiagent Debate ICML2024
  4. Synergistic Simulations: Multi-Agent Problem Solving with Large Language Models

Additional Topics Covered

Overview of LLM-Agent

Benchmarks

  1. OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problems. ACL2024

  2. CFBench: A Comprehensive Constraints-Following Benchmark for LLMs 2024.8

  3. Evaluating and Enhancing LLMs Agent based on Theory of Mind in Guandan: A Multi-Player Cooperative Game under Imperfect Information 2024.8

  4. TravelPlanner: A Benchmark for Real-World Planning with Language Agents ICML 2024

  5. LLMs Still Can't Plan; Can LRMs? A Preliminary Evaluation of OpenAI's o1 on PlanBench 2024.9

  6. Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models

Prompt Design

  1. Minstrel: Structural Prompt Generation with Multi-Agents Coordination for Non-AI Experts 2024.9

Continuous Updates

This repository will be continuously updated as new research on LLM Agents is conducted. Stay tuned for the latest papers, benchmarks, and innovations in the field.