Hugging Face's logo

Buckets:
hf-doc-build
/
doc

Files
xet
hf-doc-build/doc / diffusers /v0.43.0 /zh /optimization /pruna.html
mishig's picture
mishig
download
raw
30.5 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Pruna&quot;,&quot;local&quot;:&quot;pruna&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;安装&quot;,&quot;local&quot;:&quot;installation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;优化 Diffusers 模型&quot;,&quot;local&quot;:&quot;optimize-diffusers-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;评估和基准测试 Diffusers 模型&quot;,&quot;local&quot;:&quot;evaluate-and-benchmark-diffusers-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;参考&quot;,&quot;local&quot;:&quot;reference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/>
<link href="/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/I9eooN77.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/Bupkjr9f.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/CyvZ3FS5.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/l0QnnOED.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DnVIJfC5.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DsnmJJEf.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/nodes/0.DteTvmQZ.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/CoSrdQax.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/nodes/345.RIgY2DFL.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/DSPDl1gz.js" rel="modulepreload">
<link href="/docs/diffusers/main/en/_app/immutable/chunks/D5eijdnr.js" rel="modulepreload">
<!--15jkk7f--><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Pruna&quot;,&quot;local&quot;:&quot;pruna&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Installation&quot;,&quot;local&quot;:&quot;installation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Optimize Diffusers models&quot;,&quot;local&quot;:&quot;optimize-diffusers-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Evaluate and benchmark Diffusers models&quot;,&quot;local&quot;:&quot;evaluate-and-benchmark-diffusers-models&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Reference&quot;,&quot;local&quot;:&quot;reference&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"/><!---->
<link href="/docs/diffusers/main/en/_app/immutable/assets/0.tn0RQdqM.css" rel="modulepreload"> <!--[--><!--[0--><!--[--><!--[0--><!--[--><p></p> <div class="items-center shrink-0 min-w-[100px] max-sm:min-w-[50px] justify-end ml-auto flex" style="float: inline-end; margin-inline-start: 10px; display: inline-flex; position: relative; z-index: 10;"><div dir="ltr" class="inline-flex rounded-md max-sm:rounded-sm"><button class="inline-flex items-center gap-1 h-7 max-sm:h-7 px-2 max-sm:px-1.5 text-sm font-medium text-gray-800 border border-r-0 rounded-l-md max-sm:rounded-l-sm border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-live="polite"><span class="inline-flex items-center justify-center rounded-md p-0.5 max-sm:p-0 hover:text-gray-800 dark:hover:text-gray-200"><svg class="sm:size-3.5 size-3" 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><!----></span> <span dir="auto">Copy page</span></button> <button class="inline-flex items-center justify-center w-6 max-sm:w-5 h-7 max-sm:h-7 disabled:pointer-events-none text-sm text-gray-500 hover:text-gray-700 dark:hover:text-white rounded-r-md max-sm:rounded-r-sm border border-l transition border-gray-200 bg-white hover:shadow-inner dark:border-gray-850 dark:bg-gray-950 dark:text-gray-200 dark:hover:bg-gray-800" aria-haspopup="menu" aria-expanded="false" aria-label="Open copy menu"><svg class="transition-transform text-gray-400 overflow-visible sm:size-3.5 size-3 rotate-0" width="1em" height="1em" viewBox="0 0 12 7" fill="none" xmlns="http://www.w3.org/2000/svg"><path d="M1 1L6 6L11 1" stroke="currentColor"></path></svg><!----></button></div> <!--[-1--><!--]--></div><!----> <!--[0--><h1 class="relative group"><a id="pruna" 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="#pruna"><span><svg 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><!---->Pruna<!----></span></h1><!--]--><!----> <p><a href="https://github.com/PrunaAI/pruna" rel="nofollow">Pruna</a> 是一个模型优化框架，提供多种优化方法——量化、剪枝、缓存、编译——以加速推理并降低内存占用。下图展示了这些优化方法的总体概览。</p> <table><thead><tr><th>技术</th><th>描述</th><th align="center">速度</th><th align="center">内存</th><th align="center">质量</th></tr></thead><tbody><tr><td><code>batcher</code></td><td>将多个输入组合在一起以同时处理，从而提高计算效率并缩短处理时间。</td><td align="center">✅</td><td align="center">❌</td><td align="center">➖</td></tr><tr><td><code>cacher</code></td><td>存储计算的中间结果，以加速后续操作。</td><td align="center">✅</td><td align="center">➖</td><td align="center">➖</td></tr><tr><td><code>compiler</code></td><td>针对特定硬件优化模型。</td><td align="center">✅</td><td align="center">➖</td><td align="center">➖</td></tr><tr><td><code>distiller</code></td><td>训练一个更小、更简单的模型，以模拟一个更大、更复杂的模型。</td><td align="center">✅</td><td align="center">✅</td><td align="center">❌</td></tr><tr><td><code>quantizer</code></td><td>降低权重和激活值的精度，从而减少内存需求。</td><td align="center">✅</td><td align="center">✅</td><td align="center">❌</td></tr><tr><td><code>pruner</code></td><td>移除不太重要或冗余的连接和神经元，从而得到一个更稀疏、更高效的网络。</td><td align="center">✅</td><td align="center">✅</td><td align="center">❌</td></tr><tr><td><code>recoverer</code></td><td>恢复模型压缩后的性能。</td><td align="center">➖</td><td align="center">➖</td><td align="center">✅</td></tr><tr><td><code>factorizer</code></td><td>Factorization 将多个小型矩阵乘法合并为一个大型融合操作。</td><td align="center">✅</td><td align="center">➖</td><td align="center">➖</td></tr><tr><td><code>enhancer</code></td><td>通过应用去噪或放大等后处理算法来增强模型输出。</td><td align="center">❌</td><td align="center">-</td><td align="center">✅</td></tr></tbody></table> <p>✅（改善），➖（大致相同），❌（恶化）</p> <p>在 <a href="https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html#configure-algorithms" rel="nofollow">Pruna 文档</a>中了解完整的优化方法。</p> <!--[1--><h2 class="relative group"><a id="installation" 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="#installation"><span><svg 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><!---->安装<!----></span></h2><!--]--><!----> <p>使用以下命令安装 Pruna。</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 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> 已复制</div><!----></button><!----></div> <pre class="language-bash "><!---->pip install pruna<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="optimize-diffusers-models" 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="#optimize-diffusers-models"><span><svg 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><!---->优化 Diffusers 模型<!----></span></h2><!--]--><!----> <p>如下所示，Diffusers 模型支持多种优化算法。</p> <div class="flex justify-center"><img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/diffusers_combinations.png" alt="Overview of the supported optimization algorithms for diffusers models"/></div> <p>以下示例通过组合使用 factorizer、compiler 和 cacher 算法来优化 <a href="https://huggingface.co/black-forest-labs/FLUX.1-dev" rel="nofollow">black-forest-labs/FLUX.1-dev</a>。该组合可将推理速度提升最高 4.2 倍，并将 GPU 峰值内存占用从 34.7GB 降低至 28.0GB，同时几乎保持相同的输出质量。</p> <blockquote class="tip"><p>请参阅 <a href="https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html" rel="nofollow">Pruna 优化</a> 文档，以了解本示例中使用的优化技术。</p></blockquote> <div class="flex justify-center"><img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/flux_combination.png" alt="Optimization techniques used for FLUX.1-dev showing the combination of factorizer, compiler, and cacher algorithms"/></div> <p>首先，定义一个 <code>SmashConfig</code>，指定要使用的优化算法。要优化模型，请使用 <code>smash</code> 将 pipeline 和 <code>SmashConfig</code> 包装起来，然后像往常一样使用 pipeline 进行推理。</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 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> 已复制</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> FluxPipeline
<span class="hljs-keyword">from</span> pruna <span class="hljs-keyword">import</span> PrunaModel, SmashConfig, smash
<span class="hljs-comment"># load the model</span>
<span class="hljs-comment"># Try segmind/Segmind-Vega or black-forest-labs/FLUX.1-schnell with a small GPU memory</span>
pipe = FluxPipeline.from_pretrained(
<span class="hljs-string">&quot;black-forest-labs/FLUX.1-dev&quot;</span>,
dtype=torch.bfloat16
).to(<span class="hljs-string">&quot;cuda&quot;</span>) <span class="hljs-comment"># or &quot;mps&quot;, &quot;xpu&quot;, &quot;cpu&quot;</span>
<span class="hljs-comment"># define the configuration</span>
smash_config = SmashConfig()
smash_config[<span class="hljs-string">&quot;factorizer&quot;</span>] = <span class="hljs-string">&quot;qkv_diffusers&quot;</span>
smash_config[<span class="hljs-string">&quot;compiler&quot;</span>] = <span class="hljs-string">&quot;torch_compile&quot;</span>
smash_config[<span class="hljs-string">&quot;torch_compile_target&quot;</span>] = <span class="hljs-string">&quot;module_list&quot;</span>
smash_config[<span class="hljs-string">&quot;cacher&quot;</span>] = <span class="hljs-string">&quot;fora&quot;</span>
smash_config[<span class="hljs-string">&quot;fora_interval&quot;</span>] = <span class="hljs-number">2</span>
<span class="hljs-comment"># for the best results in terms of speed you can add these configs</span>
<span class="hljs-comment"># however they will increase your warmup time from 1.5 min to 10 min</span>
<span class="hljs-comment"># smash_config[&quot;torch_compile_mode&quot;] = &quot;max-autotune-no-cudagraphs&quot;</span>
<span class="hljs-comment"># smash_config[&quot;quantizer&quot;] = &quot;torchao&quot;</span>
<span class="hljs-comment"># smash_config[&quot;torchao_quant_type&quot;] = &quot;fp8dq&quot;</span>
<span class="hljs-comment"># smash_config[&quot;torchao_excluded_modules&quot;] = &quot;norm+embedding&quot;</span>
<span class="hljs-comment"># optimize the model</span>
smashed_pipe = smash(pipe, smash_config)
<span class="hljs-comment"># run the model</span>
smashed_pipe(<span class="hljs-string">&quot;a knitted purple prune&quot;</span>).images[<span class="hljs-number">0</span>]<!----></pre></div><!----> <div class="flex justify-center"><img src="https://huggingface.co/datasets/PrunaAI/documentation-images/resolve/main/diffusers/flux_smashed_comparison.png"/></div> <p>优化完成后，我们可以使用 Hugging Face Hub 来共享和加载优化后的模型。</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 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> 已复制</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-comment"># save the model</span>
smashed_pipe.save_to_hub(<span class="hljs-string">&quot;&lt;username&gt;/FLUX.1-dev-smashed&quot;</span>)
<span class="hljs-comment"># load the model</span>
smashed_pipe = PrunaModel.from_hub(<span class="hljs-string">&quot;&lt;username&gt;/FLUX.1-dev-smashed&quot;</span>)<!----></pre></div><!----> <!--[1--><h2 class="relative group"><a id="evaluate-and-benchmark-diffusers-models" 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="#evaluate-and-benchmark-diffusers-models"><span><svg 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><!---->评估和基准测试 Diffusers 模型<!----></span></h2><!--]--><!----> <p>Pruna 提供了 <a href="https://docs.pruna.ai/en/stable/docs_pruna/user_manual/evaluate.html" rel="nofollow">EvaluationAgent</a>，用于评估您优化后模型的质量。</p> <p>我们可以关注诸如总耗时和吞吐量等指标，以及用于评估的数据集。我们可以定义一个模型并将其传递给 <code>EvaluationAgent</code>。</p> <div class="flex space-x-2 items-center my-1.5 mr-8 h-7 !pl-0 -mx-3 md:mx-0"><!--[--><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd border-gray-800 bg-black dark:bg-gray-700 text-white">优化后的模型</div><div class="flex items-center border rounded-lg px-1.5 py-1 leading-none select-none text-smd text-gray-500 cursor-pointer opacity-90 hover:text-gray-700 dark:hover:text-gray-200 hover:shadow-sm">独立模型</div><!--]--></div> <div class="language-select"><div><p>我们可以使用 <code>EvaluationAgent</code> 加载并评估优化后的模型，并将其传递给 <code>Task</code>。</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 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> 已复制</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> FluxPipeline
<span class="hljs-keyword">from</span> pruna <span class="hljs-keyword">import</span> PrunaModel
<span class="hljs-keyword">from</span> pruna.data.pruna_datamodule <span class="hljs-keyword">import</span> PrunaDataModule
<span class="hljs-keyword">from</span> pruna.evaluation.evaluation_agent <span class="hljs-keyword">import</span> EvaluationAgent
<span class="hljs-keyword">from</span> pruna.evaluation.metrics <span class="hljs-keyword">import</span> (
ThroughputMetric,
TorchMetricWrapper,
TotalTimeMetric,
)
<span class="hljs-keyword">from</span> pruna.evaluation.task <span class="hljs-keyword">import</span> Task
<span class="hljs-comment"># define the device</span>
device = <span class="hljs-string">&quot;cuda&quot;</span> <span class="hljs-keyword">if</span> torch.cuda.is_available() <span class="hljs-keyword">else</span> <span class="hljs-string">&quot;mps&quot;</span> <span class="hljs-keyword">if</span> torch.backends.mps.is_available() <span class="hljs-keyword">else</span> <span class="hljs-string">&quot;cpu&quot;</span>
<span class="hljs-comment"># load the model</span>
<span class="hljs-comment"># Try PrunaAI/Segmind-Vega-smashed or PrunaAI/FLUX.1-dev-smashed with a small GPU memory</span>
smashed_pipe = PrunaModel.from_hub(<span class="hljs-string">&quot;PrunaAI/FLUX.1-dev-smashed&quot;</span>)
<span class="hljs-comment"># Define the metrics</span>
metrics = [
TotalTimeMetric(n_iterations=<span class="hljs-number">20</span>, n_warmup_iterations=<span class="hljs-number">5</span>),
ThroughputMetric(n_iterations=<span class="hljs-number">20</span>, n_warmup_iterations=<span class="hljs-number">5</span>),
TorchMetricWrapper(<span class="hljs-string">&quot;clip&quot;</span>),
]
<span class="hljs-comment"># Define the datamodule</span>
datamodule = PrunaDataModule.from_string(<span class="hljs-string">&quot;LAION256&quot;</span>)
datamodule.limit_datasets(<span class="hljs-number">10</span>)
<span class="hljs-comment"># Define the task and evaluation agent</span>
task = Task(metrics, datamodule=datamodule, device=device)
eval_agent = EvaluationAgent(task)
<span class="hljs-comment"># Evaluate smashed model and offload it to CPU</span>
smashed_pipe.move_to_device(device)
smashed_pipe_results = eval_agent.evaluate(smashed_pipe)
smashed_pipe.move_to_device(<span class="hljs-string">&quot;cpu&quot;</span>)<!----></pre></div><!----><!----></div><!----> <div hidden=""><p>除了将优化后的模型与基础模型进行比较，您还可以单独评估 <code>diffusers</code> 模型。如果您希望在未应用优化的情况下评估模型性能，这将非常有用。我们可以使用 <code>PrunaModel</code> 包装器，并在其上运行 <code>EvaluationAgent</code> 来实现这一点。</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 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> 已复制</div><!----></button><!----></div> <pre class="language-python "><!----><span class="hljs-keyword">import</span> torch
<span class="hljs-keyword">from</span> diffusers <span class="hljs-keyword">import</span> FluxPipeline
<span class="hljs-keyword">from</span> pruna <span class="hljs-keyword">import</span> PrunaModel
<span class="hljs-comment"># load the model</span>
<span class="hljs-comment"># Try PrunaAI/Segmind-Vega-smashed or PrunaAI/FLUX.1-dev-smashed with a small GPU memory</span>
pipe = FluxPipeline.from_pretrained(
<span class="hljs-string">&quot;black-forest-labs/FLUX.1-dev&quot;</span>,
dtype=torch.bfloat16
).to(<span class="hljs-string">&quot;cpu&quot;</span>)
wrapped_pipe = PrunaModel(model=pipe)<!----></pre></div><!----><!----></div><!----><!----></div><!----> <p>现在你已经了解了如何优化和评估模型，可以开始使用 Pruna 来优化自己的模型了。幸运的是，我们提供了许多示例来帮助你入门。</p> <blockquote class="tip"><p>有关 Flux 基准测试的更多详细信息，请参阅 <a href="https://huggingface.co/blog/PrunaAI/flux-fastest-image-generation-endpoint" rel="nofollow">Announcing FLUX-Juiced: The Fastest Image Generation Endpoint (2.6 times faster)!</a> 博客文章以及 <a href="https://huggingface.co/spaces/PrunaAI/InferBench" rel="nofollow">InferBench</a> Space。</p></blockquote> <!--[1--><h2 class="relative group"><a id="reference" 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="#reference"><span><svg 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><!---->参考<!----></span></h2><!--]--><!----> <ul><li><a href="https://github.com/PrunaAI/pruna" rel="nofollow">Pruna</a></li> <li><a href="https://docs.pruna.ai/en/stable/docs_pruna/user_manual/configure.html#configure-algorithms" rel="nofollow">Pruna 优化</a></li> <li><a href="https://docs.pruna.ai/en/stable/docs_pruna/user_manual/evaluate.html" rel="nofollow">Pruna 评估</a></li> <li><a href="https://docs.pruna.ai/en/stable/docs_pruna/tutorials/index.html" rel="nofollow">Pruna 教程</a></li></ul> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/diffusers/blob/main/docs/source/en/optimization/pruna.md" target="_blank"><svg class="mr-1" 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="M31,16l-7,7l-1.41-1.41L28.17,16l-5.58-5.59L24,9l7,7z"></path><path d="M1,16l7-7l1.41,1.41L3.83,16l5.58,5.59L8,23l-7-7z"></path><path d="M12.419,25.484L17.639,6.552l1.932,0.518L14.351,26.002z"></path></svg><!----> <span><span class="underline">Update</span> on GitHub</span></a><!----> <p></p><!--]--><!----><!--]--><!--]--><!--]--> <!--[-1--><!--]--><!--]-->
<script>
{
__sveltekit_1qsdbwf = {
base: "/docs/diffusers/main/en",
assets: "/docs/diffusers/main/en"
};
const element = document.currentScript.parentElement;
Promise.all([
import("/docs/diffusers/main/en/_app/immutable/entry/start.0S0_iEoG.js"),
import("/docs/diffusers/main/en/_app/immutable/entry/app.CPdOxCbw.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 345],
data: [null,null],
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
30.5 kB
·
Xet hash:
ac952b2ce2c388d1662210350a034cb1467bf2a6c1944c867cd197dac1b12a64

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.