https://whylove.top/tags/%E6%95%88%E7%8E%87%E6%8F%90%E5%8D%87/ Recent content in 效率提升 on why｜在行动中靠近自己 Hugo zh-cn why Mon, 15 Jun 2026 21:25:37 +0800 https://whylove.top/p/%E7%94%A8%E6%9C%AC%E5%9C%B0-ai-%E8%87%AA%E5%8A%A8%E7%94%9F%E6%88%90%E6%96%87%E7%AB%A0%E7%AE%80%E4%BB%8B%E5%92%8C%E6%A0%87%E7%AD%BElm-studio-%E5%AE%9E%E8%B7%B5/ Mon, 15 Jun 2026 21:25:37 +0800 https://whylove.top/p/%E7%94%A8%E6%9C%AC%E5%9C%B0-ai-%E8%87%AA%E5%8A%A8%E7%94%9F%E6%88%90%E6%96%87%E7%AB%A0%E7%AE%80%E4%BB%8B%E5%92%8C%E6%A0%87%E7%AD%BElm-studio-%E5%AE%9E%E8%B7%B5/ <p>在编写文章的过程中，我常常会遇到怎么写简介可以让人一眼看出你想讲什么，以及给文章打上什么标签和分类，这些繁琐的工作常常消耗了我大量的精力（特别是文章多的时候），那么为了专注于文章内容本身，我们可以借助<strong>ai</strong>来帮助我们在发布之前就进行补全，在开始先介绍一下我的电脑配置，有助于如何选择合适的<strong>大模型</strong>，如下图<picture><source type="image/avif" srcset="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-320.avif 320w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-640.avif 640w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-960.avif 960w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-1440.avif 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)"><source type="image/webp" srcset="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-320.webp 320w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-640.webp 640w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-960.webp 960w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-1440.webp 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)"><img src="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-320.jpg" width="320" height="187" srcset="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-320.jpg 320w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-640.jpg 640w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-960.jpg 960w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-1440.jpg 1440w" data-native-src="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-320.jpg" data-native-srcset="https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-320.jpg 320w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-640.jpg 640w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-960.jpg 960w, https://whylove.top/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-1440.jpg 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)" loading="eager" decoding="async" fetchpriority="high" alt="alt text" data-image-state="ready" class="gallery-image" data-lightbox-enabled="true" data-lightbox-strategy="eager" data-lightbox-avif="/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-320.avif 320w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-640.avif 640w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-960.avif 960w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-1440.avif 1440w" data-lightbox-webp="/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-320.webp 320w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-640.webp 640w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-960.webp 960w, /generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-1440.webp 1440w" data-lightbox-fallback="/generated/image-variants/9f8203b40418/f80cd6ea1c60/responsive_content-fallback-1440.jpg" data-flex-basis="410px" data-flex-grow="170"></picture>主要看显卡显存和内存大小，8G显存这一级别下，最优解应该是qwen3.6 35b a3b模型，专家架构可以显著提高输出速度，减少性能需求，选择了大模型后，接下来就是选择管理大模型的工具了，最主流的应该是<strong>LM studio</strong>和<strong>ollama</strong>，在这里我推荐LM studio，因为它有图形化界面，配置方便，下面是针对本机的参数调整界面<span class="stack-lazy-picture" data-lazy-state="preview"><picture><source type="image/avif" data-srcset="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-320.avif 320w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-640.avif 640w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-960.avif 960w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-1440.avif 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)"><source type="image/webp" data-srcset="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-320.webp 320w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-640.webp 640w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-960.webp 960w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-1440.webp 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)"><img src="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-placeholder-64.jpg" width="320" height="191" data-srcset="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-320.jpg 320w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-640.jpg 640w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-960.jpg 960w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-1440.jpg 1440w" data-src="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-320.jpg" data-native-src="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-320.jpg" data-native-srcset="https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-320.jpg 320w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-640.jpg 640w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-960.jpg 960w, https://whylove.top/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-1440.jpg 1440w" sizes="(min-width: 1024px) 864px, (min-width: 768px) calc(100vw - 64px), calc(100vw - 32px)" loading="lazy" decoding="async" fetchpriority="low" alt="alt text" data-image-state="loading" class="gallery-image stack-lazy-image" data-lazy-image="true" data-lazy-state="preview" data-lightbox-enabled="true" data-lightbox-strategy="eager" data-lightbox-avif="/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-320.avif 320w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-640.avif 640w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-960.avif 960w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-1440.avif 1440w" data-lightbox-webp="/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-320.webp 320w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-640.webp 640w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-960.webp 960w, /generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-1440.webp 1440w" data-lightbox-fallback="/generated/image-variants/81554606fd6f/6edd2eb5894c/responsive_content-fallback-1440.jpg" data-flex-basis="402px" data-flex-grow="167"></picture></span>实测处理单篇博文可以达到30token/s，接下来遇到的难题是怎么连到原有的部署流程，这涉及到程序之间端口的通讯，恰好LM studio有方便的本地端口1234，具体实现可以用ai写</p>