course/pr_1021/it/_app/immutable/nodes/12.fa1f10bd.js

import{s as D,n as N,o as U}from"../chunks/scheduler.37c15a92.js";import{S as j,i as k,g as u,s as o,r as L,A as F,h as f,f as a,c as r,j as I,u as P,x,k as S,y as O,a as i,v as A,d as w,t as B,w as M}from"../chunks/index.2bf4358c.js";import{C as X}from"../chunks/CourseFloatingBanner.9ff4c771.js";import{H as J,E as K}from"../chunks/getInferenceSnippets.ebf8be91.js";function Q(y){let n,b,h,T,s,_,l,v,d,H="In questo capitolo, hai scoperto come approcciare diversi compiti di NLP utilizzando la funzione di alto livello <code>pipeline()</code> degli 🤗 Transformer. Abbiamo anche visto come cercare e utilizzare i modelli dell’Hub, nonché come usare l’Inference API per testare i modelli direttamente nel tuo browser.",g,m,G="Abbiamo discusso di come i modelli Transformer lavorino a livello alto, e parlato dell’importanza del transfer learning e dell’affinamento. Un aspetto chiave è che è possibile utilizzare l’architettuta completa oppure solo l’encoder o il decoder, dipendentemente dal compito a cui desideri lavorare. La tabella seguente riordina questi concetti:",E,c,q="<thead><tr><th>Modello</th> <th>Esempi</th> <th>Compiti</th></tr></thead> <tbody><tr><td>Encoder</td> <td>ALBERT, BERT, DistilBERT, ELECTRA, RoBERTa</td> <td>Classificazione frasale, riconoscimento delle entità nominate, estrazione di risposte a domande</td></tr> <tr><td>Decoder</td> <td>CTRL, GPT, GPT-2, Transformer XL</td> <td>Generazione di testi</td></tr> <tr><td>Encoder-decoder</td> <td>BART, T5, Marian, mBART</td> <td>Riassunti, traduzione, generazione di risposte a domande</td></tr></tbody>",z,p,R,$,C;return s=new J({props:{title:"Riassunto",local:"riassunto",headingTag:"h1"}}),l=new X({props:{chapter:1,classNames:"absolute z-10 right-0 top-0"}}),p=new K({props:{source:"https://github.com/huggingface/course/blob/main/chapters/it/chapter1/9.mdx"}}),{c(){n=u("meta"),b=o(),h=u("p"),T=o(),L(s.$$.fragment),_=o(),L(l.$$.fragment),v=o(),d=u("p"),d.innerHTML=H,g=o(),m=u("p"),m.textContent=G,E=o(),c=u("table"),c.innerHTML=q,z=o(),L(p.$$.fragment),R=o(),$=u("p"),this.h()},l(e){const t=F("svelte-u9bgzb",document.head);n=f(t,"META",{name:!0,content:!0}),t.forEach(a),b=r(e),h=f(e,"P",{}),I(h).forEach(a),T=r(e),P(s.$$.fragment,e),_=r(e),P(l.$$.fragment,e),v=r(e),d=f(e,"P",{"data-svelte-h":!0}),x(d)!=="svelte-rcp72y"&&(d.innerHTML=H),g=r(e),m=f(e,"P",{"data-svelte-h":!0}),x(m)!=="svelte-1xp5tfc"&&(m.textContent=G),E=r(e),c=f(e,"TABLE",{"data-svelte-h":!0}),x(c)!=="svelte-9glssv"&&(c.innerHTML=q),z=r(e),P(p.$$.fragment,e),R=r(e),$=f(e,"P",{}),I($).forEach(a),this.h()},h(){S(n,"name","hf:doc:metadata"),S(n,"content",V)},m(e,t){O(document.head,n),i(e,b,t),i(e,h,t),i(e,T,t),A(s,e,t),i(e,_,t),A(l,e,t),i(e,v,t),i(e,d,t),i(e,g,t),i(e,m,t),i(e,E,t),i(e,c,t),i(e,z,t),A(p,e,t),i(e,R,t),i(e,$,t),C=!0},p:N,i(e){C||(w(s.$$.fragment,e),w(l.$$.fragment,e),w(p.$$.fragment,e),C=!0)},o(e){B(s.$$.fragment,e),B(l.$$.fragment,e),B(p.$$.fragment,e),C=!1},d(e){e&&(a(b),a(h),a(T),a(_),a(v),a(d),a(g),a(m),a(E),a(c),a(z),a(R),a($)),a(n),M(s,e),M(l,e),M(p,e)}}}const V='{"title":"Riassunto","local":"riassunto","sections":[],"depth":1}';function W(y){return U(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class ae extends j{constructor(n){super(),k(this,n,W,Q,D,{})}}export{ae as component};