Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"Using mlx-image at Hugging Face","local":"using-mlx-image-at-hugging-face","sections":[{"title":"Exploring mlx-image on the Hub","local":"exploring-mlx-image-on-the-hub","sections":[],"depth":2},{"title":"Installation","local":"installation","sections":[],"depth":2},{"title":"Models","local":"models","sections":[],"depth":2},{"title":"ImageNet-1K Results","local":"imagenet-1k-results","sections":[],"depth":2},{"title":"Similarity to PyTorch and other familiar tools","local":"similarity-to-pytorch-and-other-familiar-tools","sections":[],"depth":2},{"title":"Training","local":"training","sections":[],"depth":2},{"title":"Additional Resources","local":"additional-resources","sections":[],"depth":2},{"title":"Contact","local":"contact","sections":[],"depth":2}],"depth":1}"> | |
| <link href="/docs/hub/main/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/entry/start.d0cd5065.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/scheduler.d6170356.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/singletons.d032f1eb.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/paths.752f1c6b.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/entry/app.b6abe3c1.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/index.fcd4cc08.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/nodes/0.f045427f.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/nodes/63.35434383.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/Tip.b09c67cf.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/CodeBlock.7b16bdef.js"> | |
| <link rel="modulepreload" href="/docs/hub/main/en/_app/immutable/chunks/EditOnGithub.da2b595c.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{"title":"Using mlx-image at Hugging Face","local":"using-mlx-image-at-hugging-face","sections":[{"title":"Exploring mlx-image on the Hub","local":"exploring-mlx-image-on-the-hub","sections":[],"depth":2},{"title":"Installation","local":"installation","sections":[],"depth":2},{"title":"Models","local":"models","sections":[],"depth":2},{"title":"ImageNet-1K Results","local":"imagenet-1k-results","sections":[],"depth":2},{"title":"Similarity to PyTorch and other familiar tools","local":"similarity-to-pytorch-and-other-familiar-tools","sections":[],"depth":2},{"title":"Training","local":"training","sections":[],"depth":2},{"title":"Additional Resources","local":"additional-resources","sections":[],"depth":2},{"title":"Contact","local":"contact","sections":[],"depth":2}],"depth":1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="using-mlx-image-at-hugging-face" 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="#using-mlx-image-at-hugging-face"><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>Using mlx-image at Hugging Face</span></h1> <p data-svelte-h="svelte-1c0kah6"><a href="https://github.com/riccardomusmeci/mlx-image" rel="nofollow"><code>mlx-image</code></a> is an image models library developed by <a href="https://github.com/riccardomusmeci" rel="nofollow">Riccardo Musmeci</a> built on Apple <a href="https://github.com/ml-explore/mlx" rel="nofollow">MLX</a>. It tries to replicate the great <a href="https://github.com/huggingface/pytorch-image-models" rel="nofollow">timm</a>, but for MLX models.</p> <h2 class="relative group"><a id="exploring-mlx-image-on-the-hub" 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="#exploring-mlx-image-on-the-hub"><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>Exploring mlx-image on the Hub</span></h2> <p data-svelte-h="svelte-15ir2pb">You can find <code>mlx-image</code> models by filtering using the <code>mlx-image</code> library name, like in <a href="https://huggingface.co/models?library=mlx-image&sort=trending" rel="nofollow">this query</a>. | |
| There’s also an open <a href="https://huggingface.co/mlx-vision" rel="nofollow">mlx-vision</a> community for contributors converting and publishing weights for MLX format.</p> <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 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>Installation</span></h2> <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 -->pip install mlx-image<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="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="#models"><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>Models</span></h2> <p data-svelte-h="svelte-13m84xx">Model weights are available on the <a href="https://huggingface.co/mlx-vision" rel="nofollow"><code>mlx-vision</code></a> community on HuggingFace.</p> <p data-svelte-h="svelte-rlp6qh">To load a model with pre-trained weights:</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> mlxim.model <span class="hljs-keyword">import</span> create_model | |
| <span class="hljs-comment"># loading weights from HuggingFace (https://huggingface.co/mlx-vision/resnet18-mlxim)</span> | |
| model = create_model(<span class="hljs-string">"resnet18"</span>) <span class="hljs-comment"># pretrained weights loaded from HF</span> | |
| <span class="hljs-comment"># loading weights from local file</span> | |
| model = create_model(<span class="hljs-string">"resnet18"</span>, weights=<span class="hljs-string">"path/to/resnet18/model.safetensors"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1xjx8g7">To list all available models:</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> mlxim.model <span class="hljs-keyword">import</span> list_models | |
| list_models()<!-- HTML_TAG_END --></pre></div> <div class="course-tip course-tip-orange bg-gradient-to-br dark:bg-gradient-to-r before:border-orange-500 dark:before:border-orange-800 from-orange-50 dark:from-gray-900 to-white dark:to-gray-950 border border-orange-50 text-orange-700 dark:text-gray-400"><p data-svelte-h="svelte-1ayt94f">As of today (2024-03-15) mlx does not support <code>group</code> param for nn.Conv2d. Therefore, architectures such as <code>resnext</code>, <code>regnet</code> or <code>efficientnet</code> are not yet supported in <code>mlx-image</code>.</p></div> <h2 class="relative group"><a id="imagenet-1k-results" 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="#imagenet-1k-results"><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>ImageNet-1K Results</span></h2> <p data-svelte-h="svelte-kcleo1">Go to <a href="https://github.com/riccardomusmeci/mlx-image/blob/main/results/results-imagenet-1k.csv" rel="nofollow">results-imagenet-1k.csv</a> to check every model converted to <code>mlx-image</code> and its performance on ImageNet-1K with different settings.</p> <blockquote data-svelte-h="svelte-1xz4blu"><p><strong>TL;DR</strong> performance is comparable to the original models from PyTorch implementations.</p></blockquote> <h2 class="relative group"><a id="similarity-to-pytorch-and-other-familiar-tools" 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="#similarity-to-pytorch-and-other-familiar-tools"><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>Similarity to PyTorch and other familiar tools</span></h2> <p data-svelte-h="svelte-11sjtis"><code>mlx-image</code> tries to be as close as possible to PyTorch:</p> <ul data-svelte-h="svelte-jcjia8"><li><p><code>DataLoader</code> -> you can define your own <code>collate_fn</code> and also use <code>num_workers</code> to speed up data loading</p></li> <li><p><code>Dataset</code> -> <code>mlx-image</code> already supports <code>LabelFolderDataset</code> (the good and old PyTorch <code>ImageFolder</code>) and <code>FolderDataset</code> (a generic folder with images in it)</p></li> <li><p><code>ModelCheckpoint</code> -> keeps track of the best model and saves it to disk (similar to PyTorchLightning). It also suggests early stopping</p></li></ul> <h2 class="relative group"><a id="training" 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="#training"><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>Training</span></h2> <p data-svelte-h="svelte-1cvxtzq">Training is similar to PyTorch. Here’s an example of how to train a model:</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">import</span> mlx.nn <span class="hljs-keyword">as</span> nn | |
| <span class="hljs-keyword">import</span> mlx.optimizers <span class="hljs-keyword">as</span> optim | |
| <span class="hljs-keyword">from</span> mlxim.model <span class="hljs-keyword">import</span> create_model | |
| <span class="hljs-keyword">from</span> mlxim.data <span class="hljs-keyword">import</span> LabelFolderDataset, DataLoader | |
| train_dataset = LabelFolderDataset( | |
| root_dir=<span class="hljs-string">"path/to/train"</span>, | |
| class_map={<span class="hljs-number">0</span>: <span class="hljs-string">"class_0"</span>, <span class="hljs-number">1</span>: <span class="hljs-string">"class_1"</span>, <span class="hljs-number">2</span>: [<span class="hljs-string">"class_2"</span>, <span class="hljs-string">"class_3"</span>]} | |
| ) | |
| train_loader = DataLoader( | |
| dataset=train_dataset, | |
| batch_size=<span class="hljs-number">32</span>, | |
| shuffle=<span class="hljs-literal">True</span>, | |
| num_workers=<span class="hljs-number">4</span> | |
| ) | |
| model = create_model(<span class="hljs-string">"resnet18"</span>) <span class="hljs-comment"># pretrained weights loaded from HF</span> | |
| optimizer = optim.Adam(learning_rate=<span class="hljs-number">1e-3</span>) | |
| <span class="hljs-keyword">def</span> <span class="hljs-title function_">train_step</span>(<span class="hljs-params">model, inputs, targets</span>): | |
| logits = model(inputs) | |
| loss = mx.mean(nn.losses.cross_entropy(logits, target)) | |
| <span class="hljs-keyword">return</span> loss | |
| model.train() | |
| <span class="hljs-keyword">for</span> epoch <span class="hljs-keyword">in</span> <span class="hljs-built_in">range</span>(<span class="hljs-number">10</span>): | |
| <span class="hljs-keyword">for</span> batch <span class="hljs-keyword">in</span> train_loader: | |
| x, target = batch | |
| train_step_fn = nn.value_and_grad(model, train_step) | |
| loss, grads = train_step_fn(x, target) | |
| optimizer.update(model, grads) | |
| mx.<span class="hljs-built_in">eval</span>(model.state, optimizer.state)<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="additional-resources" 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="#additional-resources"><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>Additional Resources</span></h2> <ul data-svelte-h="svelte-k2vhlo"><li><a href="https://github.com/riccardomusmeci/mlx-image" rel="nofollow">mlx-image repository</a></li> <li><a href="https://huggingface.co/mlx-vision" rel="nofollow">mlx-vision community</a></li></ul> <h2 class="relative group"><a id="contact" 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="#contact"><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>Contact</span></h2> <p data-svelte-h="svelte-12c9hfs">If you have any questions, please email <code>riccardomusmeci92@gmail.com</code>.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/hub-docs/blob/main/docs/hub/mlx-image.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_1vatp3t = { | |
| assets: "/docs/hub/main/en", | |
| base: "/docs/hub/main/en", | |
| env: {} | |
| }; | |
| const element = document.currentScript.parentElement; | |
| const data = [null,null]; | |
| Promise.all([ | |
| import("/docs/hub/main/en/_app/immutable/entry/start.d0cd5065.js"), | |
| import("/docs/hub/main/en/_app/immutable/entry/app.b6abe3c1.js") | |
| ]).then(([kit, app]) => { | |
| kit.start(app, element, { | |
| node_ids: [0, 63], | |
| data, | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 25.8 kB
- Xet hash:
- 703f03130ee715ae746f83f69971afdf91d8825644169748d5b55fa8f1ff04b4
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.