130万参数模型用人类示范数据学会实时玩DOOM，胜过9.2万倍大的大模型。
Playing DOOM with 1.3M Parameters: Specialized Small Models vs Large Language Models for Real-Time Game Control
- 用小模型+特定编码器处理游戏帧，31毫秒决策一次
- 仅3.1万次人类示范就达成每局17.8杀，远超所有大模型总和
- 适合资源有限的实时控制场景，可部署在消费级硬件
我们提出SauerkrautLM-Doom-MultiVec，一个130万参数的小模型，能实时操控经典第一人称射击游戏DOOM。该模型结合ModernBERT编码器、哈希嵌入、深度感知标记表示和注意力池化分类头，从ASCII帧中每31毫秒做出一次动作决策。仅用3.1万条人类游戏示范训练，其在defend_the_center场景中10局累计获得178杀（每局17.8杀），超过所有测试的大语言模型（共13杀）总和。所有智能体输入相同：ASCII帧与深度图。尽管参数量仅为Nemotron-120B的1/92,000，该模型是唯一主动攻击敌人的代理，而非仅躲避。结果表明，基于领域适配数据训练的小型专用模型，在实时控制任务中可显著超越通用大语言模型，且推理成本极低，可在消费级硬件上部署。
原文摘要 · Abstract (English)
We present SauerkrautLM-Doom-MultiVec, a 1.3 million parameter model that plays the classic first-person shooter DOOM in real time, outperforming large language models up to 92,000x its size, including Nemotron-120B, Qwen3.5-27B, and GPT-4o-mini. Our model combines a ModernBERT encoder with hash embeddings, depth-aware token representations, and an attention pooling classification head to select game actions from ASCII frame representations at 31ms per decision. Trained on just 31,000 human gameplay demonstrations, it achieves 178 frags in 10 episodes (17.8 per episode) in the defend_the_center scenario, more than all tested LLMs combined (13 frags total). All agents receive equivalent input: ASCII frames and depth maps. Despite having 92,000x fewer parameters than Nemotron-120B, our model is the only agent that actively engages enemies rather than purely evading them. These results demonstrate that small, task-specific models trained on domain-appropriate data can decisively outperform general-purpose LLMs at real-time control tasks, at a fraction of the inference cost, with deployment capability on consumer hardware.
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。