Buckets:

hf-doc-build/doc-dev / datasets /main /en /use_with_spark.html
rtrm's picture
download
raw
18.2 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Use with Spark&quot;,&quot;local&quot;:&quot;use-with-spark&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Load from Spark&quot;,&quot;local&quot;:&quot;load-from-spark&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Caching&quot;,&quot;local&quot;:&quot;caching&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Feature types&quot;,&quot;local&quot;:&quot;feature-types&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
<link href="/docs/datasets/main/en/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/entry/start.4d44eea4.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/scheduler.bdbef820.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/singletons.36b689ad.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/index.8a885b74.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/paths.27092e28.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/entry/app.d83067e8.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/index.c0aea24a.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/nodes/0.bfb01985.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/each.e59479a4.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/nodes/50.f6e0733d.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/Tip.31005f7d.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/CodeBlock.6ccca92e.js">
<link rel="modulepreload" href="/docs/datasets/main/en/_app/immutable/chunks/EditOnGithub.725ee0c1.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Use with Spark&quot;,&quot;local&quot;:&quot;use-with-spark&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Load from Spark&quot;,&quot;local&quot;:&quot;load-from-spark&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Caching&quot;,&quot;local&quot;:&quot;caching&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Feature types&quot;,&quot;local&quot;:&quot;feature-types&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="use-with-spark" 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="#use-with-spark"><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>Use with Spark</span></h1> <p data-svelte-h="svelte-n01qzp">This document is a quick introduction to using 🤗 Datasets with Spark, with a particular focus on how to load a Spark DataFrame into a <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset">Dataset</a> object.</p> <p data-svelte-h="svelte-1pvlf1d">From there, you have fast access to any element and you can use it as a data loader to train models.</p> <h2 class="relative group"><a id="load-from-spark" 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="#load-from-spark"><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>Load from Spark</span></h2> <p data-svelte-h="svelte-1qkrgz7">A <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset">Dataset</a> object is a wrapper of an Arrow table, which allows fast reads from arrays in the dataset to PyTorch, TensorFlow and JAX tensors.
The Arrow table is memory mapped from disk, which can load datasets bigger than your available RAM.</p> <p data-svelte-h="svelte-1r572x2">You can get a <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset">Dataset</a> from a Spark DataFrame using <code>Dataset.from_spark()</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 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-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> Dataset
<span class="hljs-meta">&gt;&gt;&gt; </span>df = spark.createDataFrame(
<span class="hljs-meta">... </span> data=[[<span class="hljs-number">1</span>, <span class="hljs-string">&quot;Elia&quot;</span>], [<span class="hljs-number">2</span>, <span class="hljs-string">&quot;Teo&quot;</span>], [<span class="hljs-number">3</span>, <span class="hljs-string">&quot;Fang&quot;</span>]],
<span class="hljs-meta">... </span> columns=[<span class="hljs-string">&quot;id&quot;</span>, <span class="hljs-string">&quot;name&quot;</span>],
<span class="hljs-meta">... </span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>ds = Dataset.from_spark(df)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-kc65ui">The Spark workers write the dataset on disk in a cache directory as Arrow files, and the <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset">Dataset</a> is loaded from there.</p> <p data-svelte-h="svelte-eu4ytu">Alternatively, you can skip materialization by using <code>IterableDataset.from_spark()</code>, which returns an <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.IterableDataset">IterableDataset</a>:</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-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> IterableDataset
<span class="hljs-meta">&gt;&gt;&gt; </span>df = spark.createDataFrame(
<span class="hljs-meta">... </span> data=[[<span class="hljs-number">1</span>, <span class="hljs-string">&quot;Elia&quot;</span>], [<span class="hljs-number">2</span>, <span class="hljs-string">&quot;Teo&quot;</span>], [<span class="hljs-number">3</span>, <span class="hljs-string">&quot;Fang&quot;</span>]],
<span class="hljs-meta">... </span> columns=[<span class="hljs-string">&quot;id&quot;</span>, <span class="hljs-string">&quot;name&quot;</span>],
<span class="hljs-meta">... </span>)
<span class="hljs-meta">&gt;&gt;&gt; </span>ds = IterableDataset.from_spark(df)
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-built_in">print</span>(<span class="hljs-built_in">next</span>(<span class="hljs-built_in">iter</span>(ds)))
{<span class="hljs-string">&quot;id&quot;</span>: <span class="hljs-number">1</span>, <span class="hljs-string">&quot;name&quot;</span>: <span class="hljs-string">&quot;Elia&quot;</span>}<!-- HTML_TAG_END --></pre></div> <h3 class="relative group"><a id="caching" 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="#caching"><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>Caching</span></h3> <p data-svelte-h="svelte-176k11x">When using <code>Dataset.from_spark()</code>, the resulting <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Dataset">Dataset</a> is cached; if you call <code>Dataset.from_spark()</code> multiple
times on the same DataFrame it won’t re-run the Spark job that writes the dataset as Arrow files on disk.</p> <p data-svelte-h="svelte-n6wfk7">You can set the cache location by passing <code>cache_dir=</code> to <code>Dataset.from_spark()</code>.
Make sure to use a disk that is available to both your workers and your current machine (the driver).</p> <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-1m6gjz4">In a different session, a Spark DataFrame doesn’t have the same <a href="https://spark.apache.org/docs/3.2.0/api/python/reference/api/pyspark.sql.DataFrame.semanticHash.html" rel="nofollow">semantic hash</a>, and it will rerun a Spark job and store it in a new cache.</p></div> <h3 class="relative group"><a id="feature-types" 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="#feature-types"><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>Feature types</span></h3> <p data-svelte-h="svelte-9b4dmt">If your dataset is made of images, audio data or N-dimensional arrays, you can specify the <code>features=</code> argument in
<code>Dataset.from_spark()</code> (or <code>IterableDataset.from_spark()</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 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-meta">&gt;&gt;&gt; </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> Dataset, Features, Image, Value
<span class="hljs-meta">&gt;&gt;&gt; </span>data = [(<span class="hljs-number">0</span>, <span class="hljs-built_in">open</span>(<span class="hljs-string">&quot;image.png&quot;</span>, <span class="hljs-string">&quot;rb&quot;</span>).read())]
<span class="hljs-meta">&gt;&gt;&gt; </span>df = spark.createDataFrame(data, <span class="hljs-string">&quot;idx: int, image: binary&quot;</span>)
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-comment"># Also works if you have arrays</span>
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-comment"># data = [(0, np.zeros(shape=(32, 32, 3), dtype=np.int32).tolist())]</span>
<span class="hljs-meta">&gt;&gt;&gt; </span><span class="hljs-comment"># df = spark.createDataFrame(data, &quot;idx: int, image: array&lt;array&lt;array&lt;int&gt;&gt;&gt;&quot;)</span>
<span class="hljs-meta">&gt;&gt;&gt; </span>features = Features({<span class="hljs-string">&quot;idx&quot;</span>: Value(<span class="hljs-string">&quot;int64&quot;</span>), <span class="hljs-string">&quot;image&quot;</span>: Image()})
<span class="hljs-meta">&gt;&gt;&gt; </span>dataset = Dataset.from_spark(df, features=features)
<span class="hljs-meta">&gt;&gt;&gt; </span>dataset[<span class="hljs-number">0</span>]
{<span class="hljs-string">&#x27;idx&#x27;</span>: <span class="hljs-number">0</span>, <span class="hljs-string">&#x27;image&#x27;</span>: &lt;PIL.PngImagePlugin.PngImageFile image mode=RGB size=32x32&gt;}<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-iwja1a">You can check the <a href="/docs/datasets/main/en/package_reference/main_classes#datasets.Features">Features</a> documentation to know about all the feature types available.</p> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/datasets/blob/main/docs/source/use_with_spark.mdx" target="_blank"><span data-svelte-h="svelte-1kd6by1">&lt;</span> <span data-svelte-h="svelte-x0xyl0">&gt;</span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a> <p></p>
<script>
{
__sveltekit_w3org2 = {
assets: "/docs/datasets/main/en",
base: "/docs/datasets/main/en",
env: {}
};
const element = document.currentScript.parentElement;
const data = [null,null];
Promise.all([
import("/docs/datasets/main/en/_app/immutable/entry/start.4d44eea4.js"),
import("/docs/datasets/main/en/_app/immutable/entry/app.d83067e8.js")
]).then(([kit, app]) => {
kit.start(app, element, {
node_ids: [0, 50],
data,
form: null,
error: null
});
});
}
</script>

Xet Storage Details

Size:
18.2 kB
·
Xet hash:
02b7392b2f4f49a867128e68f804a13686f569ff17ce2f113448e9970774a9f9

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