Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / transformers /pr_33913 /it /run_scripts.html
download
raw
69.9 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Addestramento con script&quot;,&quot;local&quot;:&quot;addestramento-con-script&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Installazione&quot;,&quot;local&quot;:&quot;installazione&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script&quot;,&quot;local&quot;:&quot;esegui-uno-script&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Addestramento distribuito e precisione mista&quot;,&quot;local&quot;:&quot;addestramento-distribuito-e-precisione-mista&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script su TPU&quot;,&quot;local&quot;:&quot;esegui-uno-script-su-tpu&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script con 🤗 Accelerate&quot;,&quot;local&quot;:&quot;esegui-uno-script-con--accelerate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Uso di un dataset personalizzato&quot;,&quot;local&quot;:&quot;uso-di-un-dataset-personalizzato&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Testare uno script&quot;,&quot;local&quot;:&quot;testare-uno-script&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Riavviare addestramento da un checkpoint&quot;,&quot;local&quot;:&quot;riavviare-addestramento-da-un-checkpoint&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Condividi il tuo modello&quot;,&quot;local&quot;:&quot;condividi-il-tuo-modello&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
 <link href="/docs/transformers/pr_33913/it/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/entry/start.cef8f86b.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/scheduler.36a0863c.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/singletons.39cd3b38.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/index.733708bb.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/paths.64312358.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/entry/app.d399d34f.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/index.9c13489a.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/nodes/0.7859fb6f.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/each.e59479a4.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/nodes/30.49d8f7d8.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/CodeBlock.05d8ec32.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/Markdown.88297c0b.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/globals.7f7f1b26.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/IconTensorflow.e7b84ee9.js">
 <link rel="modulepreload" href="/docs/transformers/pr_33913/it/_app/immutable/chunks/EditOnGithub.e88f2b7b.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Addestramento con script&quot;,&quot;local&quot;:&quot;addestramento-con-script&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Installazione&quot;,&quot;local&quot;:&quot;installazione&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script&quot;,&quot;local&quot;:&quot;esegui-uno-script&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Addestramento distribuito e precisione mista&quot;,&quot;local&quot;:&quot;addestramento-distribuito-e-precisione-mista&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script su TPU&quot;,&quot;local&quot;:&quot;esegui-uno-script-su-tpu&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Esegui uno script con 🤗 Accelerate&quot;,&quot;local&quot;:&quot;esegui-uno-script-con--accelerate&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Uso di un dataset personalizzato&quot;,&quot;local&quot;:&quot;uso-di-un-dataset-personalizzato&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Testare uno script&quot;,&quot;local&quot;:&quot;testare-uno-script&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Riavviare addestramento da un checkpoint&quot;,&quot;local&quot;:&quot;riavviare-addestramento-da-un-checkpoint&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;Condividi il tuo modello&quot;,&quot;local&quot;:&quot;condividi-il-tuo-modello&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="addestramento-con-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="#addestramento-con-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>Addestramento con script</span></h1> <p data-svelte-h="svelte-1q55lrk">Insieme ai <a href="./notebooks">notebooks</a> 🤗 Transformers, ci sono anche esempi di script che dimostrano come addestrare un modello per un task con <a href="https://github.com/huggingface/transformers/tree/main/examples/pytorch" rel="nofollow">PyTorch</a>, <a href="https://github.com/huggingface/transformers/tree/main/examples/tensorflow" rel="nofollow">TensorFlow</a>, o <a href="https://github.com/huggingface/transformers/tree/main/examples/flax" rel="nofollow">JAX/Flax</a>.</p> <p data-svelte-h="svelte-p3vgrn">Troverai anche script che abbiamo usato nei nostri <a href="https://github.com/huggingface/transformers/tree/main/examples/research_projects" rel="nofollow">progetti di ricerca</a> e <a href="https://github.com/huggingface/transformers/tree/main/examples/legacy" rel="nofollow">precedenti esempi</a> a cui contribuisce per lo più la comunità. Questi script non sono attivamente mantenuti e richiedono una specifica versione di 🤗 Transformers che sarà molto probabilmente incompatibile con l’ultima versione della libreria.</p> <p data-svelte-h="svelte-3iprul">Non è dato per scontato che gli script di esempio funzionino senza apportare modifiche per ogni problema, bensì potrebbe essere necessario adattare lo script al tuo caso specifico. Per aiutarti in ciò, la maggioranza degli script espone le modalità di pre-processamento dei dati, consentendoti di modificare lo script come preferisci.</p> <p data-svelte-h="svelte-skpcg4">Per qualsiasi feature che vorresti implementare in uno script d’esempio, per favore discutine nel <a href="https://discuss.huggingface.co/" rel="nofollow">forum</a> o in un’<a href="https://github.com/huggingface/transformers/issues" rel="nofollow">issue</a> prima di inviare una Pull Request. Mentre accogliamo con piacere la correzione di bug, è più improbabile che faremo la stessa con una PR che aggiunge funzionalità sacrificando la leggibilità.</p> <p data-svelte-h="svelte-1dty626">Questa guida ti mostrerà come eseguire uno script di esempio relativo al task di summarization in <a href="https://github.com/huggingface/transformers/tree/main/examples/pytorch/summarization" rel="nofollow">PyTorch</a> e <a href="https://github.com/huggingface/transformers/tree/main/examples/tensorflow/summarization" rel="nofollow">TensorFlow</a>. Tutti gli esempi funzioneranno con entrambi i framework a meno che non sia specificato altrimenti.</p> <h2 class="relative group"><a id="installazione" 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="#installazione"><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>Installazione</span></h2> <p data-svelte-h="svelte-1miqfm2">Per eseguire con successo l’ultima versione degli script di esempio, devi <strong>installare 🤗 Transformers dalla fonte</strong> in un nuovo ambiente virtuale:</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 -->git <span class="hljs-built_in">clone</span> https://github.com/huggingface/transformers
<span class="hljs-built_in">cd</span> transformers
pip install .<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-b2rskv">Per le precedenti versioni degli script di esempio, clicca sul pulsante di seguito:</p> <details data-svelte-h="svelte-ytnfkk"><summary>Esempi per versioni precedenti di 🤗 Transformers</summary> <ul><li><a href="https://github.com/huggingface/transformers/tree/v4.5.1/examples">v4.5.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v4.4.2/examples">v4.4.2</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v4.3.3/examples">v4.3.3</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v4.2.2/examples">v4.2.2</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v4.1.1/examples">v4.1.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v4.0.1/examples">v4.0.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.5.1/examples">v3.5.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.4.0/examples">v3.4.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.3.1/examples">v3.3.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.2.0/examples">v3.2.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.1.0/examples">v3.1.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v3.0.2/examples">v3.0.2</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.11.0/examples">v2.11.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.10.0/examples">v2.10.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.9.1/examples">v2.9.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.8.0/examples">v2.8.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.7.0/examples">v2.7.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.6.0/examples">v2.6.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.5.1/examples">v2.5.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.4.0/examples">v2.4.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.3.0/examples">v2.3.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.2.0/examples">v2.2.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.1.0/examples">v2.1.1</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v2.0.0/examples">v2.0.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v1.2.0/examples">v1.2.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v1.1.0/examples">v1.1.0</a></li> <li><a href="https://github.com/huggingface/transformers/tree/v1.0.0/examples">v1.0.0</a></li></ul></details> <p data-svelte-h="svelte-cm69sz">Successivamente, cambia la tua attuale copia di 🤗 Transformers specificandone la versione, ad esempio v3.5.1:</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 -->git checkout tags/v3.5.1<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-2jautu">Dopo aver configurato correttamente la versione della libreria, naviga nella cartella degli esempi di tua scelta e installa i requisiti:</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 -->pip install -r requirements.txt<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="esegui-uno-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="#esegui-uno-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>Esegui uno script</span></h2> <div class="space-y-10 py-6 2xl:py-8 2xl:-mx-4"> <div class="border border-gray-200 rounded-xl px-4 relative"><div class="flex h-[22px] mt-[-12.5px] justify-between leading-none"><div class="flex px-1 items-center space-x-1 bg-white dark:bg-gray-950"><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><defs><clipPath id="a"><rect x="3.05" y="0.5" width="25.73" height="31" fill="none"></rect></clipPath></defs><g clip-path="url(#a)"><path d="M24.94,9.51a12.81,12.81,0,0,1,0,18.16,12.68,12.68,0,0,1-18,0,12.81,12.81,0,0,1,0-18.16l9-9V5l-.84.83-6,6a9.58,9.58,0,1,0,13.55,0ZM20.44,9a1.68,1.68,0,1,1,1.67-1.67A1.68,1.68,0,0,1,20.44,9Z" fill="#ee4c2c"></path></g></svg> <span>Pytorch</span></div> <div class="cursor-pointer flex items-center justify-center space-x-1 text-sm px-2 bg-white dark:bg-gray-950 hover:underline leading-none"><svg class="" width="0.9em" height="0.9em" viewBox="0 0 10 9" fill="currentColor" xmlns="http://www.w3.org/2000/svg"><path d="M1.39125 1.9725L0.0883333 0.669997L0.677917 0.0804138L8.9275 8.33041L8.33792 8.91958L6.95875 7.54041C6.22592 8.00523 5.37572 8.25138 4.50792 8.25C2.26125 8.25 0.392083 6.63333 0 4.5C0.179179 3.52946 0.667345 2.64287 1.39167 1.9725H1.39125ZM5.65667 6.23833L5.04667 5.62833C4.81335 5.73996 4.55116 5.77647 4.29622 5.73282C4.04129 5.68918 3.80617 5.56752 3.62328 5.38463C3.44039 5.20175 3.31874 4.96663 3.27509 4.71169C3.23144 4.45676 3.26795 4.19456 3.37958 3.96125L2.76958 3.35125C2.50447 3.75187 2.38595 4.2318 2.4341 4.70978C2.48225 5.18777 2.6941 5.63442 3.0338 5.97411C3.37349 6.31381 3.82015 6.52567 4.29813 6.57382C4.77611 6.62197 5.25605 6.50345 5.65667 6.23833ZM2.83042 1.06666C3.35 0.862497 3.91625 0.749997 4.50792 0.749997C6.75458 0.749997 8.62375 2.36666 9.01583 4.5C8.88816 5.19404 8.60119 5.84899 8.1775 6.41333L6.56917 4.805C6.61694 4.48317 6.58868 4.15463 6.48664 3.84569C6.3846 3.53675 6.21162 3.256 5.98156 3.02594C5.7515 2.79588 5.47075 2.6229 5.16181 2.52086C4.85287 2.41882 4.52433 2.39056 4.2025 2.43833L2.83042 1.06708V1.06666Z" fill="currentColor"></path></svg> <span>Hide Pytorch content</span></div></div> <div class="framework-content"> <p data-svelte-h="svelte-1c47yw4">Lo script di esempio scarica e pre-processa un dataset dalla libreria 🤗 <a href="https://huggingface.co/docs/datasets/" rel="nofollow">Datasets</a>. Successivamente, lo script esegue il fine-tuning su un dataset usando il <a href="https://huggingface.co/docs/transformers/main_classes/trainer" rel="nofollow">Trainer</a> su un’architettura che supporta la summarization. Il seguente esempio mostra come eseguire il fine-tuning di <a href="https://huggingface.co/google-t5/t5-small" rel="nofollow">T5-small</a> sul dataset <a href="https://huggingface.co/datasets/cnn_dailymail" rel="nofollow">CNN/DailyMail</a>. Il modello T5 richiede un parametro addizionale <code>source_prefix</code> a causa del modo in cui è stato addestrato. Questo prefisso permette a T5 di sapere che si tratta di un task di summarization.</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 -->python examples/pytorch/summarization/run_summarization.py \
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div></div></div> <div class="border border-gray-200 rounded-xl px-4 relative"><div class="flex h-[22px] mt-[-12.5px] justify-between leading-none"><div class="flex px-1 items-center space-x-1 bg-white dark:bg-gray-950"><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="0.94em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 274"><path d="M145.726 42.065v42.07l72.861 42.07v-42.07l-72.86-42.07zM0 84.135v42.07l36.43 21.03V105.17L0 84.135zm109.291 21.035l-36.43 21.034v126.2l36.43 21.035v-84.135l36.435 21.035v-42.07l-36.435-21.034V105.17z" fill="#E55B2D"></path><path d="M145.726 42.065L36.43 105.17v42.065l72.861-42.065v42.065l36.435-21.03v-84.14zM255.022 63.1l-36.435 21.035v42.07l36.435-21.035V63.1zm-72.865 84.135l-36.43 21.035v42.07l36.43-21.036v-42.07zm-36.43 63.104l-36.436-21.035v84.135l36.435-21.035V210.34z" fill="#ED8E24"></path><path d="M145.726 0L0 84.135l36.43 21.035l109.296-63.105l72.861 42.07L255.022 63.1L145.726 0zm0 126.204l-36.435 21.03l36.435 21.036l36.43-21.035l-36.43-21.03z" fill="#F8BF3C"></path></svg> <span>TensorFlow</span></div> <div class="cursor-pointer flex items-center justify-center space-x-1 text-sm px-2 bg-white dark:bg-gray-950 hover:underline leading-none"><svg class="" width="0.9em" height="0.9em" viewBox="0 0 10 9" fill="currentColor" xmlns="http://www.w3.org/2000/svg"><path d="M1.39125 1.9725L0.0883333 0.669997L0.677917 0.0804138L8.9275 8.33041L8.33792 8.91958L6.95875 7.54041C6.22592 8.00523 5.37572 8.25138 4.50792 8.25C2.26125 8.25 0.392083 6.63333 0 4.5C0.179179 3.52946 0.667345 2.64287 1.39167 1.9725H1.39125ZM5.65667 6.23833L5.04667 5.62833C4.81335 5.73996 4.55116 5.77647 4.29622 5.73282C4.04129 5.68918 3.80617 5.56752 3.62328 5.38463C3.44039 5.20175 3.31874 4.96663 3.27509 4.71169C3.23144 4.45676 3.26795 4.19456 3.37958 3.96125L2.76958 3.35125C2.50447 3.75187 2.38595 4.2318 2.4341 4.70978C2.48225 5.18777 2.6941 5.63442 3.0338 5.97411C3.37349 6.31381 3.82015 6.52567 4.29813 6.57382C4.77611 6.62197 5.25605 6.50345 5.65667 6.23833ZM2.83042 1.06666C3.35 0.862497 3.91625 0.749997 4.50792 0.749997C6.75458 0.749997 8.62375 2.36666 9.01583 4.5C8.88816 5.19404 8.60119 5.84899 8.1775 6.41333L6.56917 4.805C6.61694 4.48317 6.58868 4.15463 6.48664 3.84569C6.3846 3.53675 6.21162 3.256 5.98156 3.02594C5.7515 2.79588 5.47075 2.6229 5.16181 2.52086C4.85287 2.41882 4.52433 2.39056 4.2025 2.43833L2.83042 1.06708V1.06666Z" fill="currentColor"></path></svg> <span>Hide TensorFlow content</span></div></div> <div class="framework-content"> <p data-svelte-h="svelte-11ga8js">Lo script di esempio scarica e pre-processa un dataset dalla libreria 🤗 <a href="https://huggingface.co/docs/datasets/" rel="nofollow">Datasets</a>. Successivamente, lo script esegue il fine-tuning su un dataset usando Keras su un’architettura che supporta la summarization. Il seguente esempio mostra come eseguire il fine-tuning di <a href="https://huggingface.co/google-t5/t5-small" rel="nofollow">T5-small</a> sul dataset <a href="https://huggingface.co/datasets/cnn_dailymail" rel="nofollow">CNN/DailyMail</a>. Il modello T5 richiede un parametro addizionale <code>source_prefix</code> a causa del modo in cui è stato addestrato. Questo prefisso permette a T5 di sapere che si tratta di un task di summarization.</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 -->python examples/tensorflow/summarization/run_summarization.py \
 --model_name_or_path google-t5/t5-small \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size 8 \
 --per_device_eval_batch_size 16 \
 --num_train_epochs 3 \
 --do_train \
 --do_eval<!-- HTML_TAG_END --></pre></div> </div></div> </div> <h2 class="relative group"><a id="addestramento-distribuito-e-precisione-mista" 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="#addestramento-distribuito-e-precisione-mista"><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>Addestramento distribuito e precisione mista</span></h2> <p data-svelte-h="svelte-143nmbm">Il <a href="https://huggingface.co/docs/transformers/main_classes/trainer" rel="nofollow">Trainer</a> supporta l’addestramento distribuito e la precisione mista, che significa che puoi anche usarla in uno script. Per abilitare entrambe le funzionalità:</p> <ul data-svelte-h="svelte-1ld4sku"><li>Aggiunto l’argomento <code>fp16</code> per abilitare la precisione mista.</li> <li>Imposta un numero di GPU da usare con l’argomento <code>nproc_per_node</code>.</li></ul> <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 -->torchrun \
 --nproc_per_node 8 pytorch/summarization/run_summarization.py \
 --fp16 \
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-3o4ent">Gli script TensorFlow utilizzano una <a href="https://www.tensorflow.org/guide/distributed_training#mirroredstrategy" rel="nofollow"><code>MirroredStrategy</code></a> per il training distribuito e non devi aggiungere alcun argomento addizionale allo script di training. Lo script TensorFlow userà multiple GPU in modo predefinito se quest’ultime sono disponibili:</p> <h2 class="relative group"><a id="esegui-uno-script-su-tpu" 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="#esegui-uno-script-su-tpu"><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>Esegui uno script su TPU</span></h2> <div class="space-y-10 py-6 2xl:py-8 2xl:-mx-4"> <div class="border border-gray-200 rounded-xl px-4 relative"><div class="flex h-[22px] mt-[-12.5px] justify-between leading-none"><div class="flex px-1 items-center space-x-1 bg-white dark:bg-gray-950"><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><defs><clipPath id="a"><rect x="3.05" y="0.5" width="25.73" height="31" fill="none"></rect></clipPath></defs><g clip-path="url(#a)"><path d="M24.94,9.51a12.81,12.81,0,0,1,0,18.16,12.68,12.68,0,0,1-18,0,12.81,12.81,0,0,1,0-18.16l9-9V5l-.84.83-6,6a9.58,9.58,0,1,0,13.55,0ZM20.44,9a1.68,1.68,0,1,1,1.67-1.67A1.68,1.68,0,0,1,20.44,9Z" fill="#ee4c2c"></path></g></svg> <span>Pytorch</span></div> <div class="cursor-pointer flex items-center justify-center space-x-1 text-sm px-2 bg-white dark:bg-gray-950 hover:underline leading-none"><svg class="" width="0.9em" height="0.9em" viewBox="0 0 10 9" fill="currentColor" xmlns="http://www.w3.org/2000/svg"><path d="M1.39125 1.9725L0.0883333 0.669997L0.677917 0.0804138L8.9275 8.33041L8.33792 8.91958L6.95875 7.54041C6.22592 8.00523 5.37572 8.25138 4.50792 8.25C2.26125 8.25 0.392083 6.63333 0 4.5C0.179179 3.52946 0.667345 2.64287 1.39167 1.9725H1.39125ZM5.65667 6.23833L5.04667 5.62833C4.81335 5.73996 4.55116 5.77647 4.29622 5.73282C4.04129 5.68918 3.80617 5.56752 3.62328 5.38463C3.44039 5.20175 3.31874 4.96663 3.27509 4.71169C3.23144 4.45676 3.26795 4.19456 3.37958 3.96125L2.76958 3.35125C2.50447 3.75187 2.38595 4.2318 2.4341 4.70978C2.48225 5.18777 2.6941 5.63442 3.0338 5.97411C3.37349 6.31381 3.82015 6.52567 4.29813 6.57382C4.77611 6.62197 5.25605 6.50345 5.65667 6.23833ZM2.83042 1.06666C3.35 0.862497 3.91625 0.749997 4.50792 0.749997C6.75458 0.749997 8.62375 2.36666 9.01583 4.5C8.88816 5.19404 8.60119 5.84899 8.1775 6.41333L6.56917 4.805C6.61694 4.48317 6.58868 4.15463 6.48664 3.84569C6.3846 3.53675 6.21162 3.256 5.98156 3.02594C5.7515 2.79588 5.47075 2.6229 5.16181 2.52086C4.85287 2.41882 4.52433 2.39056 4.2025 2.43833L2.83042 1.06708V1.06666Z" fill="currentColor"></path></svg> <span>Hide Pytorch content</span></div></div> <div class="framework-content"> <p data-svelte-h="svelte-1ndgg9f">Le Tensor Processing Units (TPU) sono state progettate per migliorare le prestazioni. PyTorch supporta le TPU con il compilatore per deep learning <a href="https://www.tensorflow.org/xla" rel="nofollow">XLA</a> (guarda <a href="https://github.com/pytorch/xla/blob/master/README.md" rel="nofollow">questo link</a> per maggiori dettagli). Per usare una TPU, avvia lo script <code>xla_spawn.py</code> e usa l’argomento <code>num_cores</code> per impostare il numero di core TPU che intendi usare.</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 -->python xla_spawn.py --num_cores 8 \
 summarization/run_summarization.py \
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div></div></div> <div class="border border-gray-200 rounded-xl px-4 relative"><div class="flex h-[22px] mt-[-12.5px] justify-between leading-none"><div class="flex px-1 items-center space-x-1 bg-white dark:bg-gray-950"><svg class="" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="0.94em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 274"><path d="M145.726 42.065v42.07l72.861 42.07v-42.07l-72.86-42.07zM0 84.135v42.07l36.43 21.03V105.17L0 84.135zm109.291 21.035l-36.43 21.034v126.2l36.43 21.035v-84.135l36.435 21.035v-42.07l-36.435-21.034V105.17z" fill="#E55B2D"></path><path d="M145.726 42.065L36.43 105.17v42.065l72.861-42.065v42.065l36.435-21.03v-84.14zM255.022 63.1l-36.435 21.035v42.07l36.435-21.035V63.1zm-72.865 84.135l-36.43 21.035v42.07l36.43-21.036v-42.07zm-36.43 63.104l-36.436-21.035v84.135l36.435-21.035V210.34z" fill="#ED8E24"></path><path d="M145.726 0L0 84.135l36.43 21.035l109.296-63.105l72.861 42.07L255.022 63.1L145.726 0zm0 126.204l-36.435 21.03l36.435 21.036l36.43-21.035l-36.43-21.03z" fill="#F8BF3C"></path></svg> <span>TensorFlow</span></div> <div class="cursor-pointer flex items-center justify-center space-x-1 text-sm px-2 bg-white dark:bg-gray-950 hover:underline leading-none"><svg class="" width="0.9em" height="0.9em" viewBox="0 0 10 9" fill="currentColor" xmlns="http://www.w3.org/2000/svg"><path d="M1.39125 1.9725L0.0883333 0.669997L0.677917 0.0804138L8.9275 8.33041L8.33792 8.91958L6.95875 7.54041C6.22592 8.00523 5.37572 8.25138 4.50792 8.25C2.26125 8.25 0.392083 6.63333 0 4.5C0.179179 3.52946 0.667345 2.64287 1.39167 1.9725H1.39125ZM5.65667 6.23833L5.04667 5.62833C4.81335 5.73996 4.55116 5.77647 4.29622 5.73282C4.04129 5.68918 3.80617 5.56752 3.62328 5.38463C3.44039 5.20175 3.31874 4.96663 3.27509 4.71169C3.23144 4.45676 3.26795 4.19456 3.37958 3.96125L2.76958 3.35125C2.50447 3.75187 2.38595 4.2318 2.4341 4.70978C2.48225 5.18777 2.6941 5.63442 3.0338 5.97411C3.37349 6.31381 3.82015 6.52567 4.29813 6.57382C4.77611 6.62197 5.25605 6.50345 5.65667 6.23833ZM2.83042 1.06666C3.35 0.862497 3.91625 0.749997 4.50792 0.749997C6.75458 0.749997 8.62375 2.36666 9.01583 4.5C8.88816 5.19404 8.60119 5.84899 8.1775 6.41333L6.56917 4.805C6.61694 4.48317 6.58868 4.15463 6.48664 3.84569C6.3846 3.53675 6.21162 3.256 5.98156 3.02594C5.7515 2.79588 5.47075 2.6229 5.16181 2.52086C4.85287 2.41882 4.52433 2.39056 4.2025 2.43833L2.83042 1.06708V1.06666Z" fill="currentColor"></path></svg> <span>Hide TensorFlow content</span></div></div> <div class="framework-content"> <p data-svelte-h="svelte-11qd6ng">Le Tensor Processing Units (TPU) sono state progettate per migliorare le prestazioni. Gli script TensorFlow utilizzano una <a href="https://www.tensorflow.org/guide/distributed_training#tpustrategy" rel="nofollow"><code>TPUStrategy</code></a> per eseguire l’addestramento su TPU. Per usare una TPU, passa il nome della risorsa TPU all’argomento <code>tpu</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 -->python run_summarization.py \
 --tpu name_of_tpu_resource \
 --model_name_or_path google-t5/t5-small \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size 8 \
 --per_device_eval_batch_size 16 \
 --num_train_epochs 3 \
 --do_train \
 --do_eval<!-- HTML_TAG_END --></pre></div> </div></div> </div> <h2 class="relative group"><a id="esegui-uno-script-con--accelerate" 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="#esegui-uno-script-con--accelerate"><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>Esegui uno script con 🤗 Accelerate</span></h2> <p data-svelte-h="svelte-10zit0a">🤗 <a href="https://huggingface.co/docs/accelerate" rel="nofollow">Accelerate</a> è una libreria compatibile solo con PyTorch che offre un metodo unificato per addestrare modelli su diverse tipologie di configurazioni (CPU, multiple GPU, TPU) mantenendo una completa visibilità rispetto al ciclo di training di PyTorch. Assicurati di aver effettuato l’installazione di 🤗 Accelerate, nel caso non lo avessi fatto:</p> <blockquote><p data-svelte-h="svelte-1ehqze5">Nota: dato che Accelerate è in rapido sviluppo, è necessario installare la versione proveniente da git per eseguire gli 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 class=""><!-- HTML_TAG_START -->pip install git+https://github.com/huggingface/accelerate<!-- HTML_TAG_END --></pre></div></blockquote> <p data-svelte-h="svelte-1b94v6a">Invece che usare lo script <code>run_summarization.py</code>, devi usare lo script <code>run_summarization_no_trainer.py</code>. Gli script supportati in 🤗 Accelerate avranno un file chiamato <code>task_no_trainer.py</code> nella rispettiva cartella. Per iniziare, esegui il seguente comando per creare e salvare un file di configurazione:</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 -->accelerate config<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-13dqlnm">Testa la tua configurazione per assicurarti della sua correttezza:</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 -->accelerate <span class="hljs-built_in">test</span><!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-azupwy">Ora sei pronto per avviare l’addestramento:</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 -->accelerate launch run_summarization_no_trainer.py \
 --model_name_or_path google-t5/t5-small \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir ~/tmp/tst-summarization<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="uso-di-un-dataset-personalizzato" 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="#uso-di-un-dataset-personalizzato"><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>Uso di un dataset personalizzato</span></h2> <p data-svelte-h="svelte-3m0es0">Lo script di summarization supporta dataset personalizzati purché siano file CSV o JSON Line. Quando usi il tuo dataset, devi specificare diversi argomenti aggiuntivi:</p> <ul data-svelte-h="svelte-1msqo4o"><li><code>train_file</code> e <code>validation_file</code> specificano dove si trovano i file di addestramento e validazione.</li> <li><code>text_column</code> è il file di input da riassumere.</li> <li><code>summary_column</code> è il file di destinazione per l’output.</li></ul> <p data-svelte-h="svelte-1cj93yk">Uno script di summarization usando un dataset personalizzato sarebbe simile a questo:</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 -->python examples/pytorch/summarization/run_summarization.py \
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --train_file path_to_csv_or_jsonlines_file \
 --validation_file path_to_csv_or_jsonlines_file \
 --text_column text_column_name \
 --summary_column summary_column_name \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --overwrite_output_dir \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="testare-uno-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="#testare-uno-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>Testare uno script</span></h2> <p data-svelte-h="svelte-1sqi69s">È spesso una buona idea avviare il tuo script su un numero inferiore di esempi tratti dal dataset, per assicurarti che tutto funzioni come previsto prima di eseguire lo script sull’intero dataset, che potrebbe necessitare di ore. Usa i seguenti argomenti per limitare il dataset ad un massimo numero di esempi:</p> <ul data-svelte-h="svelte-14g5f8g"><li><code>max_train_samples</code></li> <li><code>max_eval_samples</code></li> <li><code>max_predict_samples</code></li></ul> <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 -->python examples/pytorch/summarization/run_summarization.py \
 --model_name_or_path google-t5/t5-small \
 --max_train_samples 50 \
 --max_eval_samples 50 \
 --max_predict_samples 50 \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-lpiiy4">Non tutti gli esempi di script supportano l’argomento <code>max_predict_samples</code>. Se non sei sicuro circa il supporto di questo argomento da parte del tuo script, aggiungi l’argomento <code>-h</code> per controllare:</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 -->examples/pytorch/summarization/run_summarization.py -h<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="riavviare-addestramento-da-un-checkpoint" 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="#riavviare-addestramento-da-un-checkpoint"><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>Riavviare addestramento da un checkpoint</span></h2> <p data-svelte-h="svelte-1ojfq0o">Un’altra utile opzione è riavviare un addestramento da un checkpoint precedente. Questo garantirà che tu possa riprendere da dove hai interrotto senza ricominciare se l’addestramento viene interrotto. Ci sono due metodi per riavviare l’addestramento da un checkpoint:</p> <p data-svelte-h="svelte-er6rqd">Il primo metodo usa l’argomento <code>output_dir previous_output_dir</code> per riavviare l’addestramento dall’ultima versione del checkpoint contenuto in <code>output_dir</code>. In questo caso, dovresti rimuovere <code>overwrite_output_dir</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 -->python examples/pytorch/summarization/run_summarization.py
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --output_dir previous_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-vzkarm">Il secondo metodo usa l’argomento <code>resume_from_checkpoint path_to_specific_checkpoint</code> per riavviare un addestramento da una specifica cartella di checkpoint.</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 -->python examples/pytorch/summarization/run_summarization.py
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --resume_from_checkpoint path_to_specific_checkpoint \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <h2 class="relative group"><a id="condividi-il-tuo-modello" 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="#condividi-il-tuo-modello"><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>Condividi il tuo modello</span></h2> <p data-svelte-h="svelte-5tj2pz">Tutti gli script possono caricare il tuo modello finale al <a href="https://huggingface.co/models" rel="nofollow">Model Hub</a>. Prima di iniziare, assicurati di aver effettuato l’accesso su Hugging Face:</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 -->huggingface-cli login<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-gmt2gy">Poi, aggiungi l’argomento <code>push_to_hub</code> allo script. Questo argomento consentirà di creare un repository con il tuo username Hugging Face e la cartella specificata in <code>output_dir</code>.</p> <p data-svelte-h="svelte-3hq370">Per dare uno specifico nome al repository, usa l’argomento <code>push_to_hub_model_id</code>. Il repository verrà automaticamente elencata sotto al tuo namespace.</p> <p data-svelte-h="svelte-1doeu81">Il seguente esempio mostra come caricare un modello specificando il nome del repository:</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 -->python examples/pytorch/summarization/run_summarization.py
 --model_name_or_path google-t5/t5-small \
 --do_train \
 --do_eval \
 --dataset_name cnn_dailymail \
 --dataset_config <span class="hljs-string">&quot;3.0.0&quot;</span> \
 --source_prefix <span class="hljs-string">&quot;summarize: &quot;</span> \
 --push_to_hub \
 --push_to_hub_model_id finetuned-t5-cnn_dailymail \
 --output_dir /tmp/tst-summarization \
 --per_device_train_batch_size=4 \
 --per_device_eval_batch_size=4 \
 --overwrite_output_dir \
 --predict_with_generate<!-- HTML_TAG_END --></pre></div> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/transformers/blob/main/docs/source/it/run_scripts.md" 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_728xau = {
 assets: "/docs/transformers/pr_33913/it",
 base: "/docs/transformers/pr_33913/it",
 env: {}
 };
 const element = document.currentScript.parentElement;
 const data = [null,null];
 Promise.all([
 import("/docs/transformers/pr_33913/it/_app/immutable/entry/start.cef8f86b.js"),
 import("/docs/transformers/pr_33913/it/_app/immutable/entry/app.d399d34f.js")
 ]).then(([kit, app]) => {
 kit.start(app, element, {
 node_ids: [0, 30],
 data,
 form: null,
 error: null
 });
 });
 }
 </script>

Xet Storage Details

Size:
69.9 kB
·
Xet hash:
b0e9b7688b7da200f2759d5f9f3de31f929b1b0614b0724ad252cff5c94db85a

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