让游戏角色既像人又会做事，解决大模型对话过度表演的问题
Deflanderization for Game Dialogue: Balancing Character Authenticity with Task Execution in LLM-based NPCs
- 用轻量提示词控制大模型，减少角色过度扮演
- 在任务执行上获API赛道第2、4名，兼顾真实与效率
- 适合想提升游戏角色自然度的开发者和研究者
大型语言模型（LLMs）为游戏中的动态非玩家角色（NPC）带来了新可能，可同时实现任务执行与符合人物设定的对话生成。本文报告了团队Tu_Character_lab参与2025年常识性人格基底对话挑战赛（CPDC）第二轮的经历，该比赛评估三个方向：任务导向对话、上下文感知对话及其融合。我们的方法结合两种互补策略：（i）在API赛道采用轻量提示技术，包括一种名为Deflanderization的提示方法，用于抑制过度角色扮演，提升任务完成度；（ii）在GPU赛道使用微调的大模型，基于Qwen3-14B，通过监督微调（SFT）与低秩适应（LoRA）进行优化。最佳提交结果在任务1中位列第2，在任务3（API）中排名第2，在任务3（GPU）中排名第4。
原文摘要 · Abstract (English)
The emergence of large language models (LLMs) has opened new opportunities for creating dynamic non-player characters (NPCs) in gaming environments, enabling both functional task execution and persona-consistent dialogue generation. In this paper, we (Tu_Character_lab) report our participation in the Commonsense Persona-Grounded Dialogue Challenge (CPDC) 2025 Round 2, which evaluates agents across three tracks: task-oriented dialogue, context-aware dialogue, and their integration. Our approach combines two complementary strategies: (i) lightweight prompting techniques in the API track, including a Deflanderization prompting method to suppress excessive role-play and improve task fidelity, and (ii) fine-tuned large models in the GPU track, leveraging Qwen3-14B with supervisedfinetuning (SFT) and Low-Rank Adaptation(LoRA). Our best submissions ranked 2nd on Task 1, 2nd on Task 3 (API track), and 4th on Task 3 (GPU track).
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。