GameScaling
介绍
**GameScaling** 是近期在公开渠道被持续讨论的产品/模型/工具条目。雷达侧原始摘录如下： > "Tsinghua × miHoYo are open-sourcing GameScaling -- the first open-source generalist game-playing LLM series, with 8 models ranging from 0.8B to 27B parameters." (first-author post 2026-10-22 09:17Z ≈ 17:17 BJT; Chinese recap by @huang18825 7.5k views/99 likes; still circulating through 23:55Z). miHoYo Honkai AI R&D + Tsinghua continued pre-training of Qwen3.5 VLMs on keyboard/mouse gameplay; outputs the next 200 ms of actions; plays Minecraft, Doom, escape rooms; S1 (Minecraft) / S3 (multi-game) checkpoints, Apache-2.0 weights + inference code, no training data; research-only. 信息来源标注为 `X (search_posts_all) + Hugging Face THUSI-Lab + gamescaling.github.io`；若需核对原文，可访问公开链接。
解释
根据摘录中的表述，可将 **GameScaling** 理解为围绕该名称发布或曝光的具体能力与定位说明。下文仅复述摘录已给出的事实，不补充未出现的融资额、参数规模或榜单分数。 摘录中可确认的要点包括： - "Tsinghua × miHoYo are open-sourcing GameScaling -- the first open-source generalist game-playing LLM series, with 8 models ranging from 0.8B to 27B parameters." (first-author post 2026-10-22 09:17Z ≈ 17:17 BJT; - Chinese recap by @huang18825 7.5k views/99 likes; - still circulating through 23:55Z). - miHoYo Honkai AI R&D + Tsinghua continued pre-training of Qwen3.5 VLMs on keyboard/mouse gameplay; - outputs the next 200 ms of actions; - plays Minecraft, Doom, escape rooms; 本词条在雷达中的 score 为 `0.86`（该分数来自雷达排序字段，不是第三方评测榜单分）。
用法
若你想进一步了解或试用 **GameScaling**，建议优先从公开入口核实： 1. 打开来源链接：<https://huggingface.co/THUSI-Lab/GameScaling>，对照官方说明与当前可用性。 2. 阅读摘录中提到的访问方式（如 API、开放权重、waitlist、平台内置入口等），按官方指引申请或下载。 3. 落地前请自行核对许可协议、地区可用性与安全限制；摘录未写明的配额、价格或评测结果请勿默认成立。