相关推荐
AnimeGamer
<div class="post-content leading-8 text-[#242424] post-content text-lg space-y-7 mt-8">AnimeGamer 是基于多模态大型语言模型（MLLM）构建的，可以生成动态动画镜头和角色状态更新，为用户提供无尽的动漫生活体验。它允许用户通过开放式语言指令与动漫角色互动，创建独特的冒险故事。该产品的主要优点包括：动态生成与角色交互的动画，能够在不同动漫之间创建交互，丰富的游戏状态预测等。</div> <div class="mt-8"> <div class="flex flex-col gap-2" dir="ltr" data-orientation="horizontal" data-slot="tabs"> <div class="bg-muted text-muted-foreground w-fit items-center justify-center rounded-lg p-[3px] flex md:flex-row flex-wrap h-auto px-2" tabindex="0" role="tablist" aria-orientation="horizontal" data-slot="tabs-list" data-orientation="horizontal"> </div> <div id="radix-«R2sfetlb»-content-website-screenshot" class="flex-1 outline-none" tabindex="0" role="tabpanel" data-state="active" data-orientation="horizontal" aria-labelledby="radix-«R2sfetlb»-trigger-website-screenshot" data-slot="tabs-content"> <div class="relative group"> <div class="p-5 z-10 bg-gradient-img rounded-lg mb-8 md:h-[480px] h-64"> <div class="relative overflow-hidden object-cover w-full h-full rounded-lg border-surface-100 border transition-transform duration-300 ease-in-out transform group-hover:scale-101"><img style="display: block; margin-left: auto; margin-right: auto;" src="https://img.medsci.cn/aisite/tinymce/2025/05/21/1747867190_7koZiw8gKa.jpg"></div> <div class="relative overflow-hidden object-cover w-full h-full rounded-lg border-surface-100 border transition-transform duration-300 ease-in-out transform group-hover:scale-101"> </div> <div class="relative overflow-hidden object-cover w-full h-full rounded-lg border-surface-100 border transition-transform duration-300 ease-in-out transform group-hover:scale-101"> <div class="markdown-heading" dir="auto"> <h2 class="heading-element" dir="auto" tabindex="-1">快速入门</h2> <a id="user-content--quick-start" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-quick-start" aria-label="永久链接：🚀 快速入门"></a></div> <div class="markdown-heading" dir="auto"> <h3 class="heading-element" dir="auto" tabindex="-1">🔮 环境设置</h3> <a id="user-content--environment-setup" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-environment-setup" aria-label="永久链接：🔮 环境设置"></a></div> <p>要设置推理环境，您可以运行以下命令：</p> <div class="highlight highlight-source-shell notranslate position-relative overflow-auto" dir="auto"> <pre>git clone https://github.com/TencentARC/AnimeGamer.git cd AnimeGamer conda create -n animegamer python==3.10 -y conda activate animegamer pip install -r requirements.txt</pre> <div class="zeroclipboard-container"> </div> </div> <p><a href="https://huggingface.co/TencentARC/AnimeGamer" rel="nofollow">请先下载AnimeGamer</a>和<a href="https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.1" rel="nofollow">Mistral-7B</a>的存档，并保存在 文件夹下<code>./checkpoints</code>。然后下载 CogvideoX 的 3D-VAE：</p> <div class="highlight highlight-source-shell notranslate position-relative overflow-auto" dir="auto"> <pre>cd checkpoints wget https://cloud.tsinghua.edu.cn/f/fdba7608a49c463ba754/?dl=1 mv 'index.html?dl=1' vae.zip unzip vae.zip</pre> <div class="zeroclipboard-container"> </div> </div> <div class="markdown-heading" dir="auto"> <h3 class="heading-element" dir="auto" tabindex="-1">🧸 格拉迪奥</h3> <a id="user-content--gradio" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-gradio" aria-label="永久链接：🧸 Gradio"></a></div> <p>运行本地 gradio 演示：</p> <div class="highlight highlight-source-shell notranslate position-relative overflow-auto" dir="auto"> <pre>python app.py </pre> <div class="zeroclipboard-container"> </div> </div> <p>此 Gradio 演示专为低显存环境设计，其中 MLLM 和 VDM 解码器部署在两块 GPU 上（每块至少配备 24G 显存）。如果您希望部署在单块 GPU 上（至少配备 60G 显存），请设置<code>LOW_VRAM_VERSION = False</code>。</p> <div class="markdown-heading" dir="auto"> <h3 class="heading-element" dir="auto" tabindex="-1">🪄 推理</h3> <a id="user-content--inference" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-inference" aria-label="永久链接：🪄 推理"></a></div> <p>要生成动作感知的多模态表示并更新角色状态，您可以运行：</p> <div class="highlight highlight-source-shell notranslate position-relative overflow-auto" dir="auto"> <pre>python inference_MLLM.py </pre> <div class="zeroclipboard-container"> </div> </div> <p>要将表示解码为动画镜头，您可以运行：</p> <div class="highlight highlight-source-shell notranslate position-relative overflow-auto" dir="auto"> <pre>python inference_Decoder.py </pre> <div class="zeroclipboard-container"> </div> </div> <p>更改说明以<code>./game_demo</code>定制您的游戏。</p> <div class="markdown-heading" dir="auto"> <h2 class="heading-element" dir="auto" tabindex="-1">🤗 致谢</h2> <a id="user-content--acknowledgements" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-acknowledgements" aria-label="永久链接：🤗 致谢"></a></div> <p>我们参考了<a href="https://github.com/XLabs-AI/x-flux">CogvideoX</a>和<a href="https://github.com/AILab-CVC/SEED-X/tree/main">SEED-X</a>来构建我们的代码库。感谢他们出色的项目。</p> <div class="markdown-heading" dir="auto"> <h2 class="heading-element" dir="auto" tabindex="-1">📜 引用</h2> <a id="user-content--citation" class="anchor" href="https://github.com/TencentARC/AnimeGamer#-citation" aria-label="永久链接：📜 引用"></a></div> <p>如果您发现这项工作有帮助，请考虑引用：</p> <div class="snippet-clipboard-content notranslate position-relative overflow-auto"> <pre class="notranslate" lang="BibTeXw"><code>@article{cheng2025animegamer, title={AnimeGamer: Infinite Anime Life Simulation with Next Game State Prediction}, author={Cheng, Junhao and Ge, Yuying and Ge, Yixiao and Liao, Jing and Shan, Ying}, journal={arXiv preprint arXiv:2504.01014}, year={2025} }</code></pre> </div> </div> </div> </div> </div> </div> </div>
YouCam AI & AR Business Solutions
AI & AR solutions for beauty, fashion, and skincare