Buckets:

hf-doc-build
/

doc

Files

xet

hf-doc-build/doc / datasets /v2.2.2 /en /dataset_script.html

rtrm

about 2 months ago

download

raw

60.2 kB

	<meta charset="utf-8" /><meta http-equiv="content-security-policy" content=""><meta name="hf:doc:metadata" content="{"local":"create-a-dataset-loading-script","sections":[{"local":"add-dataset-attributes","sections":[{"local":"multiple-configurations","title":"Multiple configurations"},{"local":"default-configurations","title":"Default configurations"}],"title":"Add dataset attributes"},{"local":"download-data-files-and-organize-splits","title":"Download data files and organize splits"},{"local":"generate-samples","title":"Generate samples"},{"local":"testing-data-and-checksum-metadata","sections":[{"local":"dataset-metadata","title":"Dataset metadata"},{"local":"optional-dummy-data","sections":[{"local":"automatic","title":"Automatic"},{"local":"manual","title":"Manual"},{"local":"run-the-tests","title":"Run the tests"}],"title":"(Optional) Dummy data"}],"title":"Testing data and checksum metadata"}],"title":"Create a dataset loading script"}" data-svelte="svelte-1phssyn">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/assets/pages/__layout.svelte-efc77dbd.css">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/start-0f8c1da7.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/vendor-8138ceec.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/paths-4b3c6e7e.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/pages/__layout.svelte-efb8e839.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/pages/dataset_script.mdx-23e0aeb8.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/Tip-12722dfc.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/IconCopyLink-2dd3a6ac.js">
	<link rel="modulepreload" href="/docs/datasets/v2.2.2/en/_app/chunks/CodeBlock-fc89709f.js">





	<h1 class="relative group"><a id="create-a-dataset-loading-script" 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="#create-a-dataset-loading-script"><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>Create a dataset loading script
	</span></h1>

	<p>Write a dataset script to load and share your own datasets. It is a Python file that defines the different configurations and splits of your dataset, as well as how to download and process the data.</p>
	<p>The script can download data files from any website, or from the same dataset repository.</p>
	<p>Any dataset script, for example <code>my_dataset.py</code>, can be placed in a folder or a repository named <code>my_dataset</code> and be loaded with:</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><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset
	<span class="hljs-meta">>>> </span>load_dataset(<span class="hljs-string">"path/to/my_dataset"</span>)<!-- HTML_TAG_END --></pre></div>
	<p>The following guide includes instructions for dataset scripts for how to:</p>
	<ul><li>Add dataset metadata.</li>
	<li>Download data files.</li>
	<li>Generate samples.</li>
	<li>Test if your dataset was generated correctly.</li>
	<li>Create a Dataset card.</li>
	<li>Upload a dataset to the Hugging Face Hub or GitHub.</li></ul>
	<p>Open the <a href="https://github.com/huggingface/datasets/blob/master/datasets/squad/squad.py" rel="nofollow">SQuAD dataset loading script</a> template to follow along on how to share a dataset.</p>


	<div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p>To help you get started, try beginning with the dataset loading script <a href="https://github.com/huggingface/datasets/blob/master/templates/new_dataset_script.py" rel="nofollow">template</a>!</p></div>
	<h2 class="relative group"><a id="add-dataset-attributes" 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="#add-dataset-attributes"><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>Add dataset attributes
	</span></h2>

	<p>The first step is to add some information, or attributes, about your dataset in <code>DatasetBuilder._info()</code>. The most important attributes you should specify are:</p>
	<ol><li><p><code>DatasetInfo.description</code> provides a concise description of your dataset. The description informs the user what’s in the dataset, how it was collected, and how it can be used for a NLP task.</p></li>
	<li><p><code>DatasetInfo.features</code> defines the name and type of each column in your dataset. This will also provide the structure for each example, so it is possible to create nested subfields in a column if you want. Take a look at <a href="/docs/datasets/v2.2.2/en/package_reference/main_classes#datasets.Features">Features</a> for a full list of feature types you can use.</p></li></ol>

	<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><!-- HTML_TAG_START -->datasets.Features(
	{
	<span class="hljs-string">"id"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"title"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"context"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"question"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"answers"</span>: datasets.<span class="hljs-type">Sequence</span>(
	{
	<span class="hljs-string">"text"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"answer_start"</span>: datasets.Value(<span class="hljs-string">"int32"</span>),
	}
	),
	}
	)<!-- HTML_TAG_END --></pre></div>
	<ol start="3"><li><p><code>DatasetInfo.homepage</code> contains the URL to the dataset homepage so users can find more details about the dataset.</p></li>
	<li><p><code>DatasetInfo.citation</code> contains a BibTeX citation for the dataset.</p></li></ol>
	<p>After you’ve filled out all these fields in the template, it should look like the following example from the SQuAD loading script:</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><!-- HTML_TAG_START --><span class="hljs-keyword">def</span> <span class="hljs-title function_">_info</span>(<span class="hljs-params">self</span>):
	<span class="hljs-keyword">return</span> datasets.DatasetInfo(
	description=_DESCRIPTION,
	features=datasets.Features(
	{
	<span class="hljs-string">"id"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"title"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"context"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"question"</span>: datasets.Value(<span class="hljs-string">"string"</span>),
	<span class="hljs-string">"answers"</span>: datasets.features.<span class="hljs-type">Sequence</span>(
	{<span class="hljs-string">"text"</span>: datasets.Value(<span class="hljs-string">"string"</span>), <span class="hljs-string">"answer_start"</span>: datasets.Value(<span class="hljs-string">"int32"</span>),}
	),
	}
	),
	<span class="hljs-comment"># No default supervised_keys (as we have to pass both question</span>
	<span class="hljs-comment"># and context as input).</span>
	supervised_keys=<span class="hljs-literal">None</span>,
	homepage=<span class="hljs-string">"https://rajpurkar.github.io/SQuAD-explorer/"</span>,
	citation=_CITATION,
	)<!-- HTML_TAG_END --></pre></div>
	<h3 class="relative group"><a id="multiple-configurations" 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="#multiple-configurations"><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>Multiple configurations
	</span></h3>

	<p>In some cases, your dataset may have multiple configurations. For example, the <a href="https://huggingface.co/datasets/super_glue" rel="nofollow">SuperGLUE</a> dataset is a collection of 5 datasets designed to evaluate language understanding tasks. 🤗 Datasets provides <a href="/docs/datasets/v2.2.2/en/package_reference/builder_classes#datasets.BuilderConfig">BuilderConfig</a> which allows you to create different configurations for the user to select from.</p>
	<p>Let’s study the <a href="https://github.com/huggingface/datasets/blob/master/datasets/super_glue/super_glue.py" rel="nofollow">SuperGLUE loading script</a> to see how you can define several configurations.</p>
	<ol><li>Create a <a href="/docs/datasets/v2.2.2/en/package_reference/builder_classes#datasets.BuilderConfig">BuilderConfig</a> subclass with attributes about your dataset. These attributes can be the features of your dataset, label classes, and a URL to the data files.</li></ol>

	<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><!-- HTML_TAG_START --><span class="hljs-keyword">class</span> <span class="hljs-title class_">SuperGlueConfig</span>(datasets.BuilderConfig):
	<span class="hljs-string">"""BuilderConfig for SuperGLUE."""</span>

	<span class="hljs-keyword">def</span> <span class="hljs-title function_">__init__</span>(<span class="hljs-params">self, features, data_url, citation, url, label_classes=(<span class="hljs-params"><span class="hljs-string">"False"</span>, <span class="hljs-string">"True"</span></span>), **kwargs</span>):
	<span class="hljs-string">"""BuilderConfig for SuperGLUE.

	Args:
	features: list[string], list of the features that will appear in the
	feature dict. Should not include "label".
	data_url: string, url to download the zip file from.
	citation: string, citation for the data set.
	url: string, url for information about the data set.
	label_classes: list[string], the list of classes for the label if the
	label is present as a string. Non-string labels will be cast to either
	'False' or 'True'.
	**kwargs: keyword arguments forwarded to super.
	"""</span>
	<span class="hljs-comment"># Version history:</span>
	<span class="hljs-comment"># 1.0.2: Fixed non-nondeterminism in ReCoRD.</span>
	<span class="hljs-comment"># 1.0.1: Change from the pre-release trial version of SuperGLUE (v1.9) to</span>
	<span class="hljs-comment"># the full release (v2.0).</span>
	<span class="hljs-comment"># 1.0.0: S3 (new shuffling, sharding and slicing mechanism).</span>
	<span class="hljs-comment"># 0.0.2: Initial version.</span>
	<span class="hljs-built_in">super</span>(SuperGlueConfig, self).__init__(version=datasets.Version(<span class="hljs-string">"1.0.2"</span>), **kwargs)
	self.features = features
	self.label_classes = label_classes
	self.data_url = data_url
	self.citation = citation
	self.url = url<!-- HTML_TAG_END --></pre></div>
	<ol start="2"><li>Create instances of your config to specify the values of the attributes of each configuration. This gives you the flexibility to specify all the name and description of each configuration. These sub-class instances should be listed under <code>DatasetBuilder.BUILDER_CONFIGS</code>:</li></ol>

	<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><!-- HTML_TAG_START --><span class="hljs-keyword">class</span> <span class="hljs-title class_">SuperGlue</span>(datasets.GeneratorBasedBuilder):
	<span class="hljs-string">"""The SuperGLUE benchmark."""</span>

	BUILDER_CONFIGS = [
	SuperGlueConfig(
	name=<span class="hljs-string">"boolq"</span>,
	description=_BOOLQ_DESCRIPTION,
	features=[<span class="hljs-string">"question"</span>, <span class="hljs-string">"passage"</span>],
	data_url=<span class="hljs-string">"https://dl.fbaipublicfiles.com/glue/superglue/data/v2/BoolQ.zip"</span>,
	citation=_BOOLQ_CITATION,
	url=<span class="hljs-string">"https://github.com/google-research-datasets/boolean-questions"</span>,
	),
	...
	...
	SuperGlueConfig(
	name=<span class="hljs-string">"axg"</span>,
	description=_AXG_DESCRIPTION,
	features=[<span class="hljs-string">"premise"</span>, <span class="hljs-string">"hypothesis"</span>],
	label_classes=[<span class="hljs-string">"entailment"</span>, <span class="hljs-string">"not_entailment"</span>],
	data_url=<span class="hljs-string">"https://dl.fbaipublicfiles.com/glue/superglue/data/v2/AX-g.zip"</span>,
	citation=_AXG_CITATION,
	url=<span class="hljs-string">"https://github.com/rudinger/winogender-schemas"</span>,
	),<!-- HTML_TAG_END --></pre></div>
	<ol start="3"><li>Now, users can load a specific configuration of the dataset with the configuration <code>name</code>:</li></ol>

	<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><!-- HTML_TAG_START --><span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset
	<span class="hljs-meta">>>> </span>dataset = load_dataset(<span class="hljs-string">'super_glue'</span>, <span class="hljs-string">'boolq'</span>)<!-- HTML_TAG_END --></pre></div>
	<h3 class="relative group"><a id="default-configurations" 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="#default-configurations"><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>Default configurations
	</span></h3>

	<p>Users must specify a configuration name when they load a dataset with multiple configurations. Otherwise, 🤗 Datasets will raise a <code>ValueError</code>, and prompt the user to select a configuration name. You can avoid this by setting a default dataset configuration with the <code>DEFAULT_CONFIG_NAME</code> attribute:</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><!-- HTML_TAG_START --><span class="hljs-keyword">class</span> <span class="hljs-title class_">NewDataset</span>(datasets.GeneratorBasedBuilder):

	VERSION = datasets.Version(<span class="hljs-string">"1.1.0"</span>)

	BUILDER_CONFIGS = [
	datasets.BuilderConfig(name=<span class="hljs-string">"first_domain"</span>, version=VERSION, description=<span class="hljs-string">"This part of my dataset covers a first domain"</span>),
	datasets.BuilderConfig(name=<span class="hljs-string">"second_domain"</span>, version=VERSION, description=<span class="hljs-string">"This part of my dataset covers a second domain"</span>),
	]

	DEFAULT_CONFIG_NAME = <span class="hljs-string">"first_domain"</span><!-- 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>Only use a default configuration when it makes sense. Don’t set one because it may be more convenient for the user to not specify a configuration when they load your dataset. For example, multi-lingual datasets often have a separate configuration for each language. An appropriate default may be an aggregated configuration that loads all the languages of the dataset if the user doesn’t request a particular one.</p></div>
	<h2 class="relative group"><a id="download-data-files-and-organize-splits" 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="#download-data-files-and-organize-splits"><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>Download data files and organize splits
	</span></h2>

	<p>After you’ve defined the attributes of your dataset, the next step is to download the data files and organize them according to their splits. </p>
	<ol><li>Create a dictionary of URLs in the loading script that point to the original SQuAD data files:</li></ol>

	<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><!-- HTML_TAG_START -->_URL = <span class="hljs-string">"https://rajpurkar.github.io/SQuAD-explorer/dataset/"</span>
	_URLS = {
	<span class="hljs-string">"train"</span>: _URL + <span class="hljs-string">"train-v1.1.json"</span>,
	<span class="hljs-string">"dev"</span>: _URL + <span class="hljs-string">"dev-v1.1.json"</span>,
	}<!-- HTML_TAG_END --></pre></div>


	<div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p>If the data files live in the same folder or repository of the dataset script, you can just pass the relative paths to the files instead of URLs.</p></div>
	<ol start="2"><li><p><a href="/docs/datasets/v2.2.2/en/package_reference/builder_classes#datasets.DownloadManager.download_and_extract">DownloadManager.download_and_extract()</a> takes this dictionary and downloads the data files. Once the files are downloaded, use <a href="/docs/datasets/v2.2.2/en/package_reference/builder_classes#datasets.SplitGenerator">SplitGenerator</a> to organize each split in the dataset. This is a simple class that contains:</p>
	<ul><li><p>The <code>name</code> of each split. You should use the standard split names: <code>Split.TRAIN</code>, <code>Split.TEST</code>, and <code>Split.VALIDATION</code>.</p></li>
	<li><p><code>gen_kwargs</code> provides the file paths to the data files to load for each split.</p></li></ul></li></ol>
	<p>Your <code>DatasetBuilder._split_generator()</code> should look like this now:</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><!-- HTML_TAG_START --><span class="hljs-keyword">def</span> <span class="hljs-title function_">_split_generators</span>(<span class="hljs-params">self, dl_manager: datasets.DownloadManager</span>) -> <span class="hljs-type">List</span>[datasets.SplitGenerator]:
	urls_to_download = self._URLS
	downloaded_files = dl_manager.download_and_extract(urls_to_download)

	<span class="hljs-keyword">return</span> [
	datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={<span class="hljs-string">"filepath"</span>: downloaded_files[<span class="hljs-string">"train"</span>]}),
	datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={<span class="hljs-string">"filepath"</span>: downloaded_files[<span class="hljs-string">"dev"</span>]}),
	]<!-- HTML_TAG_END --></pre></div>
	<h2 class="relative group"><a id="generate-samples" 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="#generate-samples"><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>Generate samples
	</span></h2>

	<p>At this point, you have:</p>
	<ul><li>Added the dataset attributes.</li>
	<li>Provided instructions for how to download the data files.</li>
	<li>Organized the splits.</li></ul>
	<p>The next step is to actually generate the samples in each split. </p>
	<ol><li><p><code>DatasetBuilder._generate_examples</code> takes the file path provided by <code>gen_kwargs</code> to read and parse the data files. You need to write a function that loads the data files and extracts the columns.</p></li>
	<li><p>Your function should yield a tuple of an <code>id_</code>, and an example from the dataset.</p></li></ol>

	<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><!-- HTML_TAG_START --><span class="hljs-keyword">def</span> <span class="hljs-title function_">_generate_examples</span>(<span class="hljs-params">self, filepath</span>):
	<span class="hljs-string">"""This function returns the examples in the raw (text) form."""</span>
	logger.info(<span class="hljs-string">"generating examples from = %s"</span>, filepath)
	<span class="hljs-keyword">with</span> <span class="hljs-built_in">open</span>(filepath) <span class="hljs-keyword">as</span> f:
	squad = json.load(f)
	<span class="hljs-keyword">for</span> article <span class="hljs-keyword">in</span> squad[<span class="hljs-string">"data"</span>]:
	title = article.get(<span class="hljs-string">"title"</span>, <span class="hljs-string">""</span>).strip()
	<span class="hljs-keyword">for</span> paragraph <span class="hljs-keyword">in</span> article[<span class="hljs-string">"paragraphs"</span>]:
	context = paragraph[<span class="hljs-string">"context"</span>].strip()
	<span class="hljs-keyword">for</span> qa <span class="hljs-keyword">in</span> paragraph[<span class="hljs-string">"qas"</span>]:
	question = qa[<span class="hljs-string">"question"</span>].strip()
	id_ = qa[<span class="hljs-string">"id"</span>]

	answer_starts = [answer[<span class="hljs-string">"answer_start"</span>] <span class="hljs-keyword">for</span> answer <span class="hljs-keyword">in</span> qa[<span class="hljs-string">"answers"</span>]]
	answers = [answer[<span class="hljs-string">"text"</span>].strip() <span class="hljs-keyword">for</span> answer <span class="hljs-keyword">in</span> qa[<span class="hljs-string">"answers"</span>]]

	<span class="hljs-comment"># Features currently used are "context", "question", and "answers".</span>
	<span class="hljs-comment"># Others are extracted here for the ease of future expansions.</span>
	<span class="hljs-keyword">yield</span> id_, {
	<span class="hljs-string">"title"</span>: title,
	<span class="hljs-string">"context"</span>: context,
	<span class="hljs-string">"question"</span>: question,
	<span class="hljs-string">"id"</span>: id_,
	<span class="hljs-string">"answers"</span>: {<span class="hljs-string">"answer_start"</span>: answer_starts, <span class="hljs-string">"text"</span>: answers,},
	}<!-- HTML_TAG_END --></pre></div>
	<h2 class="relative group"><a id="testing-data-and-checksum-metadata" 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="#testing-data-and-checksum-metadata"><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>Testing data and checksum metadata
	</span></h2>

	<p>We strongly recommend adding testing data and checksum metadata to your dataset to verify and test its behavior. This ensures the generated dataset matches your expectations.
	Testing data and checksum metadata are mandatory for datasets stored in the GitHub repository of the 🤗 Datasets library.</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>Make sure you run all of the following commands <strong>from the root</strong> of your local <code>datasets</code> repository.</p></div>
	<h3 class="relative group"><a id="dataset-metadata" 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="#dataset-metadata"><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>Dataset metadata
	</span></h3>

	<ol><li>Run the following command to create the metadata file, <code>dataset_infos.json</code>. This will also test your new dataset loading script and make sure it works correctly.</li></ol>

	<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><!-- HTML_TAG_START --><span class="hljs-comment">datasets</span><span class="hljs-literal">-</span><span class="hljs-comment">cli</span> <span class="hljs-comment">test</span> <span class="hljs-comment">datasets/</span><<span class="hljs-comment">your</span><span class="hljs-literal">-</span><span class="hljs-comment">dataset</span><span class="hljs-literal">-</span><span class="hljs-comment">folder</span>> --<span class="hljs-comment">save_infos</span> --<span class="hljs-comment">all_configs</span><!-- HTML_TAG_END --></pre></div>
	<ol start="2"><li>If your dataset loading script passed the test, you should now have a <code>dataset_infos.json</code> file in your dataset folder. This file contains information about the dataset, like its <code>features</code> and <code>download_size</code>.</li></ol>
	<h3 class="relative group"><a id="optional-dummy-data" 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="#optional-dummy-data"><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>(Optional) Dummy data
	</span></h3>

	<p>If you want to be able to test your dataset script without downloading the full dataset, you need to create some dummy data for automated testing. There are two methods for generating dummy data: automatically and manually. </p>
	<h4 class="relative group"><a id="automatic" 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="#automatic"><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>Automatic
	</span></h4>

	<p>If your data file is one of the following formats, then you can automatically generate the dummy data:</p>
	<ul><li>txt</li>
	<li>csv</li>
	<li>tsv</li>
	<li>jsonl</li>
	<li>json</li>
	<li>xml</li></ul>
	<p>Run the command below to generate the dummy data:</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><!-- HTML_TAG_START -->datasets-cli dummy_data datasets/<your-dataset-<span class="hljs-built_in">folder</span>> <span class="hljs-comment">--auto_generate</span><!-- HTML_TAG_END --></pre></div>
	<h4 class="relative group"><a id="manual" 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="#manual"><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>Manual
	</span></h4>

	<p>If your data files are not among the supported formats, you will need to generate your dummy data manually. Run the command below to output detailed instructions on how to create the dummy data:</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><!-- HTML_TAG_START -->datasets-cli dummy_data datasets/<span class="hljs-symbol"><your-dataset-folder></span>

	==============================DUMMY DATA INSTRUCTIONS==============================
	- In order <span class="hljs-keyword">to</span> create the dummy data <span class="hljs-keyword">for</span> my-dataset, please <span class="hljs-keyword">go</span> into the folder <span class="hljs-string">'./datasets/my-dataset/dummy/1.1.0'</span> with <span class="hljs-keyword">cd</span> ./datasets/my-dataset/dummy/<span class="hljs-number">1.1</span>.<span class="hljs-number">0</span> .

	- Please create the following dummy data <span class="hljs-keyword">files</span> <span class="hljs-string">'dummy_data/TREC_10.label, dummy_data/train_5500.label'</span> from the folder <span class="hljs-string">'./datasets/my-dataset/dummy/1.1.0'</span>

	- For each of the splits <span class="hljs-string">'train, test'</span>, <span class="hljs-keyword">make</span> sure that one <span class="hljs-built_in">or</span> more of the dummy data <span class="hljs-keyword">files</span> provide at least one example

	- If the method _generate_examples(...) includes multiple <span class="hljs-keyword">open</span>() statements, you might have <span class="hljs-keyword">to</span> create other <span class="hljs-keyword">files</span> in addition <span class="hljs-keyword">to</span> <span class="hljs-string">'dummy_data/TREC_10.label, dummy_data/train_5500.label'</span>. In this case please refer <span class="hljs-keyword">to</span> the _generate_examples(...) method

	- After <span class="hljs-keyword">all</span> dummy data <span class="hljs-keyword">files</span> are created, they should <span class="hljs-keyword">be</span> zipped recursively <span class="hljs-keyword">to</span> <span class="hljs-string">'dummy_data.zip'</span> with the <span class="hljs-keyword">command</span> zip -r dummy_data.zip dummy_data/

	- You can now <span class="hljs-keyword">delete</span> the folder <span class="hljs-string">'dummy_data'</span> with the <span class="hljs-keyword">command</span> rm -r dummy_data

	- To <span class="hljs-built_in">get</span> the folder <span class="hljs-string">'dummy_data'</span> back <span class="hljs-keyword">for</span> further <span class="hljs-keyword">changes</span> <span class="hljs-keyword">to</span> the dummy data, simply unzip dummy_data.zip with the <span class="hljs-keyword">command</span> unzip dummy_data.zip

	- Make sure you have created the <span class="hljs-keyword">file</span> <span class="hljs-string">'dummy_data.zip'</span> in <span class="hljs-string">'./datasets/my-dataset/dummy/1.1.0'</span>
	===================================================================================<!-- HTML_TAG_END --></pre></div>


	<div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p>Manually creating dummy data can be tricky. Make sure you follow the instructions from the command <code>datasets-cli dummy_data datasets/<your-dataset-folder></code>. If you are still unable to successfully generate dummy data, open a <a href="https://github.com/huggingface/datasets/pulls" rel="nofollow">Pull Request</a> and we will be happy to help you out!</p></div>
	<p>There should be two new files in your dataset folder:</p>
	<ul><li><p><code>dataset_infos.json</code> stores the dataset metadata including the data file checksums, and the number of examples required to confirm the dataset was generated properly.</p></li>
	<li><p><code>dummy_data.zip</code> is a file used to test the behavior of the loading script without having to download the full dataset.</p></li></ul>
	<h4 class="relative group"><a id="run-the-tests" 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="#run-the-tests"><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>Run the tests
	</span></h4>

	<p>The last step is to actually test dataset generation with the real and dummy data. Clone the <code>huggingface/datasets</code> repository and run the following command to test the real data:</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><!-- HTML_TAG_START -->RUN_SLOW=<span class="hljs-number">1</span> pytest tests/test_dataset_common.py<span class="hljs-number">::</span>LocalDatasetTest<span class="hljs-number">::</span>test_load_real_dataset_<your_dataset_name><!-- HTML_TAG_END --></pre></div>
	<p>Test the dummy data:</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><!-- HTML_TAG_START -->RUN_SLOW=<span class="hljs-number">1</span> pytest tests/test_dataset_common.py<span class="hljs-number">::</span>LocalDatasetTest<span class="hljs-number">::</span>test_load_dataset_all_configs_<your_dataset_name><!-- HTML_TAG_END --></pre></div>
	<p>If both tests pass, your dataset was generated correctly!</p>


	<script type="module" data-hydrate="qx88zx">
	import { start } from "/docs/datasets/v2.2.2/en/_app/start-0f8c1da7.js";
	start({
	target: document.querySelector('[data-hydrate="qx88zx"]').parentNode,
	paths: {"base":"/docs/datasets/v2.2.2/en","assets":"/docs/datasets/v2.2.2/en"},
	session: {},
	route: false,
	spa: false,
	trailing_slash: "never",
	hydrate: {
	status: 200,
	error: null,
	nodes: [
	import("/docs/datasets/v2.2.2/en/_app/pages/__layout.svelte-efb8e839.js"),
	import("/docs/datasets/v2.2.2/en/_app/pages/dataset_script.mdx-23e0aeb8.js")
	],
	params: {}
	}
	});
	</script>

Xet Storage Details

Size:: 60.2 kB
Xet hash:: d5d91e418be91d75633942e8c196140e266087104276781ecf434d223d1e0987

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