mlsanityData to 14 Oct 2026

Exploring and Controlling Diversity in LLM-Agent Conversation

arXiv 2412.21102 ↗First posted 30 Dec 2024cs.CL cs.AIcited by 12

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

Abstract (arXiv)

Controlling diversity in LLM-agent simulations is essential for balancing stability in structured tasks with variability in open-ended interactions. However, we observe that dialogue diversity tends to degrade over long-term simulations. To explore the role of prompt design in this phenomenon, we modularized the utterance generation prompt and found that reducing contextual information leads to more diverse outputs. Based on this insight, we propose Adaptive Prompt Pruning (APP), a novel method that allows users to control diversity via a single parameter, lambda. APP dynamically prunes prompt segments based on attention scores and is compatible with existing diversity control methods. We demonstrate that APP effectively modulates diversity through extensive experiments and propose a method to balance the control trade-offs. Our analysis reveals that all prompt components impose constraints on diversity, with the Memory being the most influential. Additionally, high-attention contents consistently suppress output diversity.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. Builds on 26 of its 36 references that are in the map. First posted on arXiv 30 Dec 2024.

Builds on

26 of 36 references.

Closest ancestors · 5

Same direction, newest first

  1. 2024Cohesive Conversations: Enhancing Authenticity in Multi-Agent Simulated Dialogues13 Jul
  2. 2024Exploring Large Language Model based Intelligent Agents: Definitions, Methods, and Prospects7 Jan
  3. 2023SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents18 Oct
  4. 2023A Dynamic LLM-Powered Agent Network for Task-Oriented Agent Collaboration3 Oct
  5. 2023The Rise and Potential of Large Language Model Based Agents: A Survey14 Sep

Other references · 21

  1. 2024The Llama 3 Herd of Modelscited by 14,715
  2. 2019Sentence-BERT: Sentence Embeddings using Siamese BERT-Networkscited by 9,170
  3. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  4. 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,568
  5. 2023Lost in the Middle: How Language Models Use Long Contextscited by 3,455
  6. 2019The Curious Case of Neural Text Degenerationcited by 2,916
  7. 2015A Diversity-Promoting Objective Function for Neural Conversation Modelscited by 1,077
  8. 2024LLM Evaluators Recognize and Favor Their Own Generationscited by 686
  9. 2021Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpuscited by 492
  10. 2022Self-critiquing models for assisting human evaluatorscited by 380
  11. 2024LLM Agents Grounded in Self-Reports Enable General-Purpose Simulation of Individualscited by 249
  12. 2020MiniLMv2: Multi-Head Self-Attention Relation Distillation for Compressing Pretrained Transformerscited by 245
9 more
  1. 2023Automatically Correcting Large Language Models: Surveying the landscape of diverse self-correction strategiescited by 238
  2. 2024Judging the Judges: Evaluating Alignment and Vulnerabilities in LLMs-as-Judgescited by 205
  3. 2022Factuality Enhanced Language Models for Open-Ended Text Generationcited by 202
  4. 2023Embers of Autoregression: Understanding Large Language Models Through the Problem They are Trained to Solvecited by 178
  5. 2023Increasing Diversity While Maintaining Accuracy: Text Data Generation with Large Language Models and Human Interventionscited by 125
  6. 2023System 2 Attention (is something you might need too)cited by 89
  7. 2023Humanoid Agents: Platform for Simulating Human-like Generative Agentscited by 73
  8. 2023SimOAP: Improve Coherence and Consistency in Persona-based Dialogue Generation via Over-sampling and Post-evaluationcited by 11
  9. 2024On the Benchmarking of LLMs for Open-Domain Dialogue Evaluationcited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 5.

12 papers.

  1. 2025Debate or Vote: Which Yields Better Decisions in Multi-Agent Large Language Models?cited by 81
  2. 2025The Homogenizing Effect of Large Language Models on Human Expression and Thoughtcited by 35
  3. 2025When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoningcited by 21
  4. 2025Localized Cultural Knowledge is Conserved and Controllable in Large Language Modelscited by 20
  5. 2025Bayesian Social Deduction with Graph-Informed Language Modelscited by 9
  6. 2025What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversitycited by 7
  7. 2026Breaking the Martingale Curse: Multi-Agent Debate via Asymmetric Cognitive Potential Energycited by 3
  8. 2026AdaExplore: Failure-Driven Adaptation and Diversity-Preserving Search for Efficient Kernel Generationcited by 3
  9. 2026M3MAD-Bench: Multi-Dimensional Evaluation of Multi-Agent Debate Across Domains and Modalitiescited by 1
  10. 2025Convergence of Outputs When Two Large Language Models Interact in a Multi-Agentic Setupcited by 0
  11. 2026Dial HEALTHDIAL for Advice: A Multilingual and Multi-Parallel Spoken Dialogue Dataset for Knowledge-Grounded Information Seekingcited by 0
  12. 2026BOUNDARY_SYNC: Measuring Communication-Induced Representational Coupling in Multi-Agent LLM Systemscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Influencing LLM Multi-Agent Dialogue via Policy-Parameterized Prompts
  2. 2025Algorithmic Prompt Generation for Diverse Human-like Teaming and Communication with Large Language Models
  3. 2025Prompt Optimization Across Multiple Agents for Representing Diverse Human Populations
  4. 2025Agent-Centric Projection of Prompting Techniques and Implications for Synthetic Training Data for Large Language Models
  5. 2026From Static Context to Calibrated Interactive RL: Mitigating Distribution Shift in Multi-turn Dialogue with Aligned Simulator
  6. 2025Tournament of Prompts: Evolving LLM Instructions Through Structured Debates and Elo Ratings
  7. 2025The Meta-Prompting Protocol: Orchestrating LLMs via Adversarial Feedback Loops
  8. 2025Fairness Evaluation and Inference Level Mitigation in LLMs