Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"GGUF 和 Transformers 的交互","local":"gguf-和-transformers-的交互","sections":[{"title":"在 Transformers 中的支持","local":"在-transformers-中的支持","sections":[{"title":"支持的量化类型","local":"支持的量化类型","sections":[],"depth":3},{"title":"支持的模型架构","local":"支持的模型架构","sections":[],"depth":3}],"depth":2},{"title":"使用示例","local":"使用示例","sections":[],"depth":2}],"depth":1}"> | |
| <link href="/docs/transformers/pr_33913/zh/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/entry/start.11693184.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/scheduler.9991993c.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/singletons.d9873ff3.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/index.02cfeb18.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/paths.7d3a2ffb.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/entry/app.080f849f.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/index.7fc9a5e7.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/nodes/0.59361ae6.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/each.e59479a4.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/nodes/13.c11155af.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/CodeBlock.e11cba92.js"> | |
| <link rel="modulepreload" href="/docs/transformers/pr_33913/zh/_app/immutable/chunks/EditOnGithub.84ab7f0e.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{"title":"GGUF 和 Transformers 的交互","local":"gguf-和-transformers-的交互","sections":[{"title":"在 Transformers 中的支持","local":"在-transformers-中的支持","sections":[{"title":"支持的量化类型","local":"支持的量化类型","sections":[],"depth":3},{"title":"支持的模型架构","local":"支持的模型架构","sections":[],"depth":3}],"depth":2},{"title":"使用示例","local":"使用示例","sections":[],"depth":2}],"depth":1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="gguf-和-transformers-的交互" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#gguf-和-transformers-的交互"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>GGUF 和 Transformers 的交互</span></h1> <p data-svelte-h="svelte-mdnd46">GGUF文件格式用于存储模型，以便通过<a href="https://github.com/ggerganov/ggml" rel="nofollow">GGML</a>和其他依赖它的库进行推理，例如非常流行的<a href="https://github.com/ggerganov/llama.cpp" rel="nofollow">llama.cpp</a>或<a href="https://github.com/ggerganov/whisper.cpp" rel="nofollow">whisper.cpp</a>。</p> <p data-svelte-h="svelte-twtju8">该文件格式<a href="https://huggingface.co/docs/hub/en/gguf" rel="nofollow">由抱抱脸支持</a>，可用于快速检查文件中张量和元数据。</p> <p data-svelte-h="svelte-1ccuwz8">该文件格式是一种“单文件格式”，通常单个文件就包含了配置属性、分词器词汇表和其他属性，同时还有模型中要加载的所有张量。这些文件根据文件的量化类型有不同的格式。我们在<a href="https://huggingface.co/docs/hub/en/gguf#quantization-types" rel="nofollow">这里</a>进行了简要介绍。</p> <h2 class="relative group"><a id="在-transformers-中的支持" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#在-transformers-中的支持"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>在 Transformers 中的支持</span></h2> <p data-svelte-h="svelte-1olmn47">我们在 transformers 中添加了加载 gguf 文件的功能，这样可以对 GGUF 模型进行进一步的训练或微调，然后再将模型转换回 GGUF 格式，以便在 ggml 生态系统中使用。加载模型时，我们首先将其反量化为 FP32，然后再加载权重以在 PyTorch 中使用。</p> <blockquote data-svelte-h="svelte-1hw6ivd"><p>[!注意] | |
| 目前这个功能还处于探索阶段，欢迎大家贡献力量，以便在不同量化类型和模型架构之间更好地完善这一功能。</p></blockquote> <p data-svelte-h="svelte-5gud0f">目前，支持的模型架构和量化类型如下：</p> <h3 class="relative group"><a id="支持的量化类型" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#支持的量化类型"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>支持的量化类型</span></h3> <p data-svelte-h="svelte-7f6b8c">根据分享在 Hub 上的较为热门的量化文件，初步支持以下量化类型：</p> <ul data-svelte-h="svelte-1f29qgp"><li>F32</li> <li>F16</li> <li>BF16</li> <li>Q4_0</li> <li>Q4_1</li> <li>Q5_0</li> <li>Q5_1</li> <li>Q8_0</li> <li>Q2_K</li> <li>Q3_K</li> <li>Q4_K</li> <li>Q5_K</li> <li>Q6_K</li> <li>IQ1_S</li> <li>IQ1_M</li> <li>IQ2_XXS</li> <li>IQ2_XS</li> <li>IQ2_S</li> <li>IQ3_XXS</li> <li>IQ3_S</li> <li>IQ4_XS</li> <li>IQ4_NL</li></ul> <blockquote data-svelte-h="svelte-el52vx"><p>[!注意] | |
| 为了支持 gguf 反量化，需要安装 <code>gguf>=0.10.0</code>。</p></blockquote> <h3 class="relative group"><a id="支持的模型架构" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#支持的模型架构"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>支持的模型架构</span></h3> <p data-svelte-h="svelte-5qsmy1">目前支持以下在 Hub 上非常热门的模型架构：</p> <ul data-svelte-h="svelte-1bkwewe"><li>LLaMa</li> <li>Mistral</li> <li>Qwen2</li> <li>Qwen2Moe</li> <li>Phi3</li> <li>Bloom</li> <li>Falcon</li> <li>StableLM</li> <li>GPT2</li> <li>Starcoder2</li></ul> <h2 class="relative group"><a id="使用示例" class="header-link block pr-1.5 text-lg no-hover:hidden with-hover:absolute with-hover:p-1.5 with-hover:opacity-0 with-hover:group-hover:opacity-100 with-hover:right-full" href="#使用示例"><span><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 256"><path d="M167.594 88.393a8.001 8.001 0 0 1 0 11.314l-67.882 67.882a8 8 0 1 1-11.314-11.315l67.882-67.881a8.003 8.003 0 0 1 11.314 0zm-28.287 84.86l-28.284 28.284a40 40 0 0 1-56.567-56.567l28.284-28.284a8 8 0 0 0-11.315-11.315l-28.284 28.284a56 56 0 0 0 79.196 79.197l28.285-28.285a8 8 0 1 0-11.315-11.314zM212.852 43.14a56.002 56.002 0 0 0-79.196 0l-28.284 28.284a8 8 0 1 0 11.314 11.314l28.284-28.284a40 40 0 0 1 56.568 56.567l-28.285 28.285a8 8 0 0 0 11.315 11.314l28.284-28.284a56.065 56.065 0 0 0 0-79.196z" fill="currentColor"></path></svg></span></a> <span>使用示例</span></h2> <p data-svelte-h="svelte-1lqi9da">为了在<code>transformers</code>中加载<code>gguf</code>文件，你需要在 <code>from_pretrained</code>方法中为分词器和模型指定 <code>gguf_file</code>参数。下面是从同一个文件中加载分词器和模型的示例：</p> <div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START --><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer, AutoModelForCausalLM | |
| model_id = <span class="hljs-string">"TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF"</span> | |
| filename = <span class="hljs-string">"tinyllama-1.1b-chat-v1.0.Q6_K.gguf"</span> | |
| tokenizer = AutoTokenizer.from_pretrained(model_id, gguf_file=filename) | |
| model = AutoModelForCausalLM.from_pretrained(model_id, gguf_file=filename)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-5jxp91">现在，你就已经可以结合 PyTorch 生态系统中的一系列其他工具，来使用完整的、未量化的模型了。</p> <p data-svelte-h="svelte-v2zjkj">为了将模型转换回<code>gguf</code>文件，我们建议使用<code>llama.cpp</code>中的<a href="https://github.com/ggerganov/llama.cpp/blob/master/convert_hf_to_gguf.py" rel="nofollow"><code>convert-hf-to-gguf.py</code>文件</a>。</p> <p data-svelte-h="svelte-1yxp62b">以下是如何补充上面的脚本，以保存模型并将其导出回 <code>gguf</code>的示例：</p> <div class="code-block relative"><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START -->tokenizer.save_pretrained(<span class="hljs-string">'directory'</span>) | |
| model.save_pretrained(<span class="hljs-string">'directory'</span>) | |
| !python ${path_to_llama_cpp}/convert-hf-to-gguf.py ${directory}<!-- HTML_TAG_END --></pre></div> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/transformers/blob/main/docs/source/zh/gguf.md" target="_blank"><span data-svelte-h="svelte-1kd6by1"><</span> <span data-svelte-h="svelte-x0xyl0">></span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a> <p></p> | |
| <script> | |
| { | |
| __sveltekit_zfhg1q = { | |
| assets: "/docs/transformers/pr_33913/zh", | |
| base: "/docs/transformers/pr_33913/zh", | |
| env: {} | |
| }; | |
| const element = document.currentScript.parentElement; | |
| const data = [null,null]; | |
| Promise.all([ | |
| import("/docs/transformers/pr_33913/zh/_app/immutable/entry/start.11693184.js"), | |
| import("/docs/transformers/pr_33913/zh/_app/immutable/entry/app.080f849f.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 13], | |
| data, | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 16.1 kB
- Xet hash:
- 7ffae936d1ffcd92c43f04f4b7064337e61488938654fa4fb2506a00dc96f55d
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.