Buckets:

hf-doc-build
/

doc-dev

hf-doc-build/doc-dev / transformers /pr_30690 /ja /_app /immutable /nodes /149.efd0cd6d.js

rtrm's picture

about 2 months ago

60.5 kB

	import{s as Yt,o as St,n as ts}from"../chunks/scheduler.9bc65507.js";import{S as Lt,i as Dt,g as y,s as i,r as f,A as Pt,h as J,f as a,c as o,j as Ht,u,x as T,k as Qt,y as Kt,a as n,v as j,d as w,t as h,w as d}from"../chunks/index.707bf1b6.js";import{T as Tt}from"../chunks/Tip.c2ecdbf4.js";import{Y as qt}from"../chunks/Youtube.e1129c6f.js";import{C as R}from"../chunks/CodeBlock.54a9f38d.js";import{D as Ot}from"../chunks/DocNotebookDropdown.41f65cb5.js";import{F as gt,M as Gs}from"../chunks/Markdown.8ab98a13.js";import{H as As,E as se}from"../chunks/EditOnGithub.922df6ba.js";function te(I){let l,m='このタスクと互換性のあるすべてのアーキテクチャとチェックポイントを確認するには、<a href="https://huggingface.co/tasks/question-answering" rel="nofollow">タスクページ</a> を確認することをお勧めします。';return{c(){l=y("p"),l.innerHTML=m},l(e){l=J(e,"P",{"data-svelte-h":!0}),T(l)!=="svelte-c4xwg1"&&(l.innerHTML=m)},m(e,M){n(e,l,M)},p:ts,d(e){e&&a(l)}}}function ee(I){let l,m;return l=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMERlZmF1bHREYXRhQ29sbGF0b3IlMEElMEFkYXRhX2NvbGxhdG9yJTIwJTNEJTIwRGVmYXVsdERhdGFDb2xsYXRvcigp",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> DefaultDataCollator

	<span class="hljs-meta">>>> </span>data_collator = DefaultDataCollator()`,wrap:!1}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p:ts,i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function le(I){let l,m;return l=new Gs({props:{$$slots:{default:[ee]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function ae(I){let l,m;return l=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMERlZmF1bHREYXRhQ29sbGF0b3IlMEElMEFkYXRhX2NvbGxhdG9yJTIwJTNEJTIwRGVmYXVsdERhdGFDb2xsYXRvcihyZXR1cm5fdGVuc29ycyUzRCUyMnRmJTIyKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> DefaultDataCollator

	<span class="hljs-meta">>>> </span>data_collator = DefaultDataCollator(return_tensors=<span class="hljs-string">"tf"</span>)`,wrap:!1}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p:ts,i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function ne(I){let l,m;return l=new Gs({props:{$$slots:{default:[ae]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function pe(I){let l,m='<a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.Trainer">Trainer</a> を使用したモデルの微調整に慣れていない場合は、<a href="../training#train-with-pytorch-trainer">ここ</a> の基本的なチュートリアルをご覧ください。';return{c(){l=y("p"),l.innerHTML=m},l(e){l=J(e,"P",{"data-svelte-h":!0}),T(l)!=="svelte-1oyh4os"&&(l.innerHTML=m)},m(e,M){n(e,l,M)},p:ts,d(e){e&&a(l)}}}function re(I){let l,m,e,M='これでモデルのトレーニングを開始する準備が整いました。 <a href="/docs/transformers/pr_30690/ja/model_doc/auto#transformers.AutoModelForQuestionAnswering">AutoModelForQuestionAnswering</a> を使用して DitilBERT をロードします。',g,U,v,k,X="この時点で残っている手順は次の 3 つだけです。",G,b,E='<li><a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.TrainingArguments">TrainingArguments</a> でトレーニングハイパーパラメータを定義します。唯一の必須パラメータは、モデルの保存場所を指定する <code>output_dir</code> です。 <code>push_to_hub=True</code>を設定して、このモデルをハブにプッシュします (モデルをアップロードするには、Hugging Face にサインインする必要があります)。</li> <li>トレーニング引数をモデル、データセット、トークナイザー、データ照合器とともに <a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.Trainer">Trainer</a> に渡します。</li> <li><a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.Trainer.train">train()</a> を呼び出してモデルを微調整します。</li>',x,C,W,_,V='トレーニングが完了したら、 <a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.Trainer.push_to_hub">push_to_hub()</a> メソッドを使用してモデルをハブに共有し、誰もがモデルを使用できるようにします。',A,Z,B;return l=new Tt({props:{$$slots:{default:[pe]},$$scope:{ctx:I}}}),U=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMEF1dG9Nb2RlbEZvclF1ZXN0aW9uQW5zd2VyaW5nJTJDJTIwVHJhaW5pbmdBcmd1bWVudHMlMkMlMjBUcmFpbmVyJTBBJTBBbW9kZWwlMjAlM0QlMjBBdXRvTW9kZWxGb3JRdWVzdGlvbkFuc3dlcmluZy5mcm9tX3ByZXRyYWluZWQoJTIyZGlzdGlsYmVydCUyRmRpc3RpbGJlcnQtYmFzZS11bmNhc2VkJTIyKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForQuestionAnswering, TrainingArguments, Trainer

	<span class="hljs-meta">>>> </span>model = AutoModelForQuestionAnswering.from_pretrained(<span class="hljs-string">"distilbert/distilbert-base-uncased"</span>)`,wrap:!1}}),C=new R({props:{code:"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",highlighted:`<span class="hljs-meta">>>> </span>training_args = TrainingArguments(
	<span class="hljs-meta">... </span> output_dir=<span class="hljs-string">"my_awesome_qa_model"</span>,
	<span class="hljs-meta">... </span> eval_strategy=<span class="hljs-string">"epoch"</span>,
	<span class="hljs-meta">... </span> learning_rate=<span class="hljs-number">2e-5</span>,
	<span class="hljs-meta">... </span> per_device_train_batch_size=<span class="hljs-number">16</span>,
	<span class="hljs-meta">... </span> per_device_eval_batch_size=<span class="hljs-number">16</span>,
	<span class="hljs-meta">... </span> num_train_epochs=<span class="hljs-number">3</span>,
	<span class="hljs-meta">... </span> weight_decay=<span class="hljs-number">0.01</span>,
	<span class="hljs-meta">... </span> push_to_hub=<span class="hljs-literal">True</span>,
	<span class="hljs-meta">... </span>)

	<span class="hljs-meta">>>> </span>trainer = Trainer(
	<span class="hljs-meta">... </span> model=model,
	<span class="hljs-meta">... </span> args=training_args,
	<span class="hljs-meta">... </span> train_dataset=tokenized_squad[<span class="hljs-string">"train"</span>],
	<span class="hljs-meta">... </span> eval_dataset=tokenized_squad[<span class="hljs-string">"test"</span>],
	<span class="hljs-meta">... </span> processing_class=tokenizer,
	<span class="hljs-meta">... </span> data_collator=data_collator,
	<span class="hljs-meta">... </span>)

	<span class="hljs-meta">>>> </span>trainer.train()`,wrap:!1}}),Z=new R({props:{code:"dHJhaW5lci5wdXNoX3RvX2h1Yigp",highlighted:'<span class="hljs-meta">>>> </span>trainer.push_to_hub()',wrap:!1}}),{c(){f(l.$$.fragment),m=i(),e=y("p"),e.innerHTML=M,g=i(),f(U.$$.fragment),v=i(),k=y("p"),k.textContent=X,G=i(),b=y("ol"),b.innerHTML=E,x=i(),f(C.$$.fragment),W=i(),_=y("p"),_.innerHTML=V,A=i(),f(Z.$$.fragment)},l(t){u(l.$$.fragment,t),m=o(t),e=J(t,"P",{"data-svelte-h":!0}),T(e)!=="svelte-lalunw"&&(e.innerHTML=M),g=o(t),u(U.$$.fragment,t),v=o(t),k=J(t,"P",{"data-svelte-h":!0}),T(k)!=="svelte-1j8bgyv"&&(k.textContent=X),G=o(t),b=J(t,"OL",{"data-svelte-h":!0}),T(b)!=="svelte-4l72ow"&&(b.innerHTML=E),x=o(t),u(C.$$.fragment,t),W=o(t),_=J(t,"P",{"data-svelte-h":!0}),T(_)!=="svelte-eob4i9"&&(_.innerHTML=V),A=o(t),u(Z.$$.fragment,t)},m(t,c){j(l,t,c),n(t,m,c),n(t,e,c),n(t,g,c),j(U,t,c),n(t,v,c),n(t,k,c),n(t,G,c),n(t,b,c),n(t,x,c),j(C,t,c),n(t,W,c),n(t,_,c),n(t,A,c),j(Z,t,c),B=!0},p(t,c){const z={};c&2&&(z.$$scope={dirty:c,ctx:t}),l.$set(z)},i(t){B\|\|(w(l.$$.fragment,t),w(U.$$.fragment,t),w(C.$$.fragment,t),w(Z.$$.fragment,t),B=!0)},o(t){h(l.$$.fragment,t),h(U.$$.fragment,t),h(C.$$.fragment,t),h(Z.$$.fragment,t),B=!1},d(t){t&&(a(m),a(e),a(g),a(v),a(k),a(G),a(b),a(x),a(W),a(_),a(A)),d(l,t),d(U,t),d(C,t),d(Z,t)}}}function ie(I){let l,m;return l=new Gs({props:{$$slots:{default:[re]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function oe(I){let l,m='Keras を使用したモデルの微調整に慣れていない場合は、<a href="../training#train-a-tensorflow-model-with-keras">こちら</a> の基本的なチュートリアルをご覧ください。';return{c(){l=y("p"),l.innerHTML=m},l(e){l=J(e,"P",{"data-svelte-h":!0}),T(l)!=="svelte-1jwo7q8"&&(l.innerHTML=m)},m(e,M){n(e,l,M)},p:ts,d(e){e&&a(l)}}}function me(I){let l,m,e,M=`</ヒント>
	TensorFlow でモデルを微調整するには、オプティマイザー関数、学習率スケジュール、およびいくつかのトレーニングハイパーパラメーターをセットアップすることから始めます。`,g,U,v,k,X='次に、<a href="/docs/transformers/pr_30690/ja/model_doc/auto#transformers.TFAutoModelForQuestionAnswering">TFAutoModelForQuestionAnswering</a> を使用して DistilBERT をロードできます。',G,b,E,x,C='<a href="/docs/transformers/pr_30690/ja/main_classes/model#transformers.TFPreTrainedModel.prepare_tf_dataset">prepare_tf_dataset()</a> を使用して、データセットを <code>tf.data.Dataset</code> 形式に変換します。',W,_,V,A,Z='<a href="https://keras.io/api/models/model_training_apis/#compile-method" rel="nofollow"><code>compile</code></a> を使用してトレーニング用のモデルを設定します。',B,t,c,z,es='トレーニングを開始する前に最後にセットアップすることは、モデルをハブにプッシュする方法を提供することです。これは、モデルとトークナイザーを <a href="/docs/transformers/pr_30690/ja/main_classes/keras_callbacks#transformers.PushToHubCallback">PushToHubCallback</a> でプッシュする場所を指定することで実行できます。',H,Y,L,F,ls='ついに、モデルのトレーニングを開始する準備が整いました。トレーニングおよび検証データセット、エポック数、コールバックを指定して <a href="https://keras.io/api/models/model_training_apis/#fit-method" rel="nofollow"><code>fit</code></a> を呼び出し、モデルを微調整します。',Q,S,D,N,as="トレーニングが完了すると、モデルは自動的にハブにアップロードされ、誰でも使用できるようになります。",q;return l=new Tt({props:{$$slots:{default:[oe]},$$scope:{ctx:I}}}),U=new R({props:{code:"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",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> create_optimizer

	<span class="hljs-meta">>>> </span>batch_size = <span class="hljs-number">16</span>
	<span class="hljs-meta">>>> </span>num_epochs = <span class="hljs-number">2</span>
	<span class="hljs-meta">>>> </span>total_train_steps = (<span class="hljs-built_in">len</span>(tokenized_squad[<span class="hljs-string">"train"</span>]) // batch_size) * num_epochs
	<span class="hljs-meta">>>> </span>optimizer, schedule = create_optimizer(
	<span class="hljs-meta">... </span> init_lr=<span class="hljs-number">2e-5</span>,
	<span class="hljs-meta">... </span> num_warmup_steps=<span class="hljs-number">0</span>,
	<span class="hljs-meta">... </span> num_train_steps=total_train_steps,
	<span class="hljs-meta">... </span>)`,wrap:!1}}),b=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMFRGQXV0b01vZGVsRm9yUXVlc3Rpb25BbnN3ZXJpbmclMEElMEFtb2RlbCUyMCUzRCUyMFRGQXV0b01vZGVsRm9yUXVlc3Rpb25BbnN3ZXJpbmcoJTIyZGlzdGlsYmVydCUyRmRpc3RpbGJlcnQtYmFzZS11bmNhc2VkJTIyKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> TFAutoModelForQuestionAnswering

	<span class="hljs-meta">>>> </span>model = TFAutoModelForQuestionAnswering(<span class="hljs-string">"distilbert/distilbert-base-uncased"</span>)`,wrap:!1}}),_=new R({props:{code:"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",highlighted:`<span class="hljs-meta">>>> </span>tf_train_set = model.prepare_tf_dataset(
	<span class="hljs-meta">... </span> tokenized_squad[<span class="hljs-string">"train"</span>],
	<span class="hljs-meta">... </span> shuffle=<span class="hljs-literal">True</span>,
	<span class="hljs-meta">... </span> batch_size=<span class="hljs-number">16</span>,
	<span class="hljs-meta">... </span> collate_fn=data_collator,
	<span class="hljs-meta">... </span>)

	<span class="hljs-meta">>>> </span>tf_validation_set = model.prepare_tf_dataset(
	<span class="hljs-meta">... </span> tokenized_squad[<span class="hljs-string">"test"</span>],
	<span class="hljs-meta">... </span> shuffle=<span class="hljs-literal">False</span>,
	<span class="hljs-meta">... </span> batch_size=<span class="hljs-number">16</span>,
	<span class="hljs-meta">... </span> collate_fn=data_collator,
	<span class="hljs-meta">... </span>)`,wrap:!1}}),t=new R({props:{code:"aW1wb3J0JTIwdGVuc29yZmxvdyUyMGFzJTIwdGYlMEElMEFtb2RlbC5jb21waWxlKG9wdGltaXplciUzRG9wdGltaXplcik=",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">import</span> tensorflow <span class="hljs-keyword">as</span> tf

	<span class="hljs-meta">>>> </span>model.<span class="hljs-built_in">compile</span>(optimizer=optimizer)`,wrap:!1}}),Y=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycy5rZXJhc19jYWxsYmFja3MlMjBpbXBvcnQlMjBQdXNoVG9IdWJDYWxsYmFjayUwQSUwQWNhbGxiYWNrJTIwJTNEJTIwUHVzaFRvSHViQ2FsbGJhY2soJTBBJTIwJTIwJTIwJTIwb3V0cHV0X2RpciUzRCUyMm15X2F3ZXNvbWVfcWFfbW9kZWwlMjIlMkMlMEElMjAlMjAlMjAlMjB0b2tlbml6ZXIlM0R0b2tlbml6ZXIlMkMlMEEp",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers.keras_callbacks <span class="hljs-keyword">import</span> PushToHubCallback

	<span class="hljs-meta">>>> </span>callback = PushToHubCallback(
	<span class="hljs-meta">... </span> output_dir=<span class="hljs-string">"my_awesome_qa_model"</span>,
	<span class="hljs-meta">... </span> tokenizer=tokenizer,
	<span class="hljs-meta">... </span>)`,wrap:!1}}),S=new R({props:{code:"bW9kZWwuZml0KHglM0R0Zl90cmFpbl9zZXQlMkMlMjB2YWxpZGF0aW9uX2RhdGElM0R0Zl92YWxpZGF0aW9uX3NldCUyQyUyMGVwb2NocyUzRDMlMkMlMjBjYWxsYmFja3MlM0QlNUJjYWxsYmFjayU1RCk=",highlighted:'<span class="hljs-meta">>>> </span>model.fit(x=tf_train_set, validation_data=tf_validation_set, epochs=<span class="hljs-number">3</span>, callbacks=[callback])',wrap:!1}}),{c(){f(l.$$.fragment),m=i(),e=y("p"),e.textContent=M,g=i(),f(U.$$.fragment),v=i(),k=y("p"),k.innerHTML=X,G=i(),f(b.$$.fragment),E=i(),x=y("p"),x.innerHTML=C,W=i(),f(_.$$.fragment),V=i(),A=y("p"),A.innerHTML=Z,B=i(),f(t.$$.fragment),c=i(),z=y("p"),z.innerHTML=es,H=i(),f(Y.$$.fragment),L=i(),F=y("p"),F.innerHTML=ls,Q=i(),f(S.$$.fragment),D=i(),N=y("p"),N.textContent=as},l(r){u(l.$$.fragment,r),m=o(r),e=J(r,"P",{"data-svelte-h":!0}),T(e)!=="svelte-1v9esfa"&&(e.textContent=M),g=o(r),u(U.$$.fragment,r),v=o(r),k=J(r,"P",{"data-svelte-h":!0}),T(k)!=="svelte-efobh8"&&(k.innerHTML=X),G=o(r),u(b.$$.fragment,r),E=o(r),x=J(r,"P",{"data-svelte-h":!0}),T(x)!=="svelte-eom40g"&&(x.innerHTML=C),W=o(r),u(_.$$.fragment,r),V=o(r),A=J(r,"P",{"data-svelte-h":!0}),T(A)!=="svelte-kzu0b7"&&(A.innerHTML=Z),B=o(r),u(t.$$.fragment,r),c=o(r),z=J(r,"P",{"data-svelte-h":!0}),T(z)!=="svelte-1rpvjj8"&&(z.innerHTML=es),H=o(r),u(Y.$$.fragment,r),L=o(r),F=J(r,"P",{"data-svelte-h":!0}),T(F)!=="svelte-ffgub5"&&(F.innerHTML=ls),Q=o(r),u(S.$$.fragment,r),D=o(r),N=J(r,"P",{"data-svelte-h":!0}),T(N)!=="svelte-vh7z0v"&&(N.textContent=as)},m(r,$){j(l,r,$),n(r,m,$),n(r,e,$),n(r,g,$),j(U,r,$),n(r,v,$),n(r,k,$),n(r,G,$),j(b,r,$),n(r,E,$),n(r,x,$),n(r,W,$),j(_,r,$),n(r,V,$),n(r,A,$),n(r,B,$),j(t,r,$),n(r,c,$),n(r,z,$),n(r,H,$),j(Y,r,$),n(r,L,$),n(r,F,$),n(r,Q,$),j(S,r,$),n(r,D,$),n(r,N,$),q=!0},p(r,$){const vs={};$&2&&(vs.$$scope={dirty:$,ctx:r}),l.$set(vs)},i(r){q\|\|(w(l.$$.fragment,r),w(U.$$.fragment,r),w(b.$$.fragment,r),w(_.$$.fragment,r),w(t.$$.fragment,r),w(Y.$$.fragment,r),w(S.$$.fragment,r),q=!0)},o(r){h(l.$$.fragment,r),h(U.$$.fragment,r),h(b.$$.fragment,r),h(_.$$.fragment,r),h(t.$$.fragment,r),h(Y.$$.fragment,r),h(S.$$.fragment,r),q=!1},d(r){r&&(a(m),a(e),a(g),a(v),a(k),a(G),a(E),a(x),a(W),a(V),a(A),a(B),a(c),a(z),a(H),a(L),a(F),a(Q),a(D),a(N)),d(l,r),d(U,r),d(b,r),d(_,r),d(t,r),d(Y,r),d(S,r)}}}function Me(I){let l,m;return l=new Gs({props:{$$slots:{default:[me]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function ce(I){let l,m=`質問応答用のモデルを微調整する方法の詳細な例については、対応するドキュメントを参照してください。
	<a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/question_answering.ipynb" rel="nofollow">PyTorch ノートブック</a>
	または <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/main/examples/question_answering-tf.ipynb" rel="nofollow">TensorFlow ノートブック</a>。`;return{c(){l=y("p"),l.innerHTML=m},l(e){l=J(e,"P",{"data-svelte-h":!0}),T(l)!=="svelte-1n4kqih"&&(l.innerHTML=m)},m(e,M){n(e,l,M)},p:ts,d(e){e&&a(l)}}}function fe(I){let l,m="テキストをトークン化して PyTorch テンソルを返します。",e,M,g,U,v="入力をモデルに渡し、<code>logits</code>を返します。",k,X,G,b,E="モデル出力から開始位置と終了位置の最も高い確率を取得します。",x,C,W,_,V="予測されたトークンをデコードして答えを取得します。",A,Z,B;return M=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMEF1dG9Ub2tlbml6ZXIlMEElMEF0b2tlbml6ZXIlMjAlM0QlMjBBdXRvVG9rZW5pemVyLmZyb21fcHJldHJhaW5lZCglMjJteV9hd2Vzb21lX3FhX21vZGVsJTIyKSUwQWlucHV0cyUyMCUzRCUyMHRva2VuaXplcihxdWVzdGlvbiUyQyUyMGNvbnRleHQlMkMlMjByZXR1cm5fdGVuc29ycyUzRCUyMnB0JTIyKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer

	<span class="hljs-meta">>>> </span>tokenizer = AutoTokenizer.from_pretrained(<span class="hljs-string">"my_awesome_qa_model"</span>)
	<span class="hljs-meta">>>> </span>inputs = tokenizer(question, context, return_tensors=<span class="hljs-string">"pt"</span>)`,wrap:!1}}),X=new R({props:{code:"aW1wb3J0JTIwdG9yY2glMEFmcm9tJTIwdHJhbnNmb3JtZXJzJTIwaW1wb3J0JTIwQXV0b01vZGVsRm9yUXVlc3Rpb25BbnN3ZXJpbmclMEElMEFtb2RlbCUyMCUzRCUyMEF1dG9Nb2RlbEZvclF1ZXN0aW9uQW5zd2VyaW5nLmZyb21fcHJldHJhaW5lZCglMjJteV9hd2Vzb21lX3FhX21vZGVsJTIyKSUwQXdpdGglMjB0b3JjaC5ub19ncmFkKCklM0ElMEElMjAlMjAlMjAlMjBvdXRwdXRzJTIwJTNEJTIwbW9kZWwoKippbnB1dHMp",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">import</span> torch
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoModelForQuestionAnswering

	<span class="hljs-meta">>>> </span>model = AutoModelForQuestionAnswering.from_pretrained(<span class="hljs-string">"my_awesome_qa_model"</span>)
	<span class="hljs-meta">>>> </span><span class="hljs-keyword">with</span> torch.no_grad():
	<span class="hljs-meta">... </span> outputs = model(**inputs)`,wrap:!1}}),C=new R({props:{code:"YW5zd2VyX3N0YXJ0X2luZGV4JTIwJTNEJTIwb3V0cHV0cy5zdGFydF9sb2dpdHMuYXJnbWF4KCklMEFhbnN3ZXJfZW5kX2luZGV4JTIwJTNEJTIwb3V0cHV0cy5lbmRfbG9naXRzLmFyZ21heCgp",highlighted:`<span class="hljs-meta">>>> </span>answer_start_index = outputs.start_logits.argmax()
	<span class="hljs-meta">>>> </span>answer_end_index = outputs.end_logits.argmax()`,wrap:!1}}),Z=new R({props:{code:"cHJlZGljdF9hbnN3ZXJfdG9rZW5zJTIwJTNEJTIwaW5wdXRzLmlucHV0X2lkcyU1QjAlMkMlMjBhbnN3ZXJfc3RhcnRfaW5kZXglMjAlM0ElMjBhbnN3ZXJfZW5kX2luZGV4JTIwJTJCJTIwMSU1RCUwQXRva2VuaXplci5kZWNvZGUocHJlZGljdF9hbnN3ZXJfdG9rZW5zKQ==",highlighted:`<span class="hljs-meta">>>> </span>predict_answer_tokens = inputs.input_ids[<span class="hljs-number">0</span>, answer_start_index : answer_end_index + <span class="hljs-number">1</span>]
	<span class="hljs-meta">>>> </span>tokenizer.decode(predict_answer_tokens)
	<span class="hljs-string">'176 billion parameters and can generate text in 46 languages natural languages and 13'</span>`,wrap:!1}}),{c(){l=y("p"),l.textContent=m,e=i(),f(M.$$.fragment),g=i(),U=y("p"),U.innerHTML=v,k=i(),f(X.$$.fragment),G=i(),b=y("p"),b.textContent=E,x=i(),f(C.$$.fragment),W=i(),_=y("p"),_.textContent=V,A=i(),f(Z.$$.fragment)},l(t){l=J(t,"P",{"data-svelte-h":!0}),T(l)!=="svelte-1n0k6or"&&(l.textContent=m),e=o(t),u(M.$$.fragment,t),g=o(t),U=J(t,"P",{"data-svelte-h":!0}),T(U)!=="svelte-1rvunpz"&&(U.innerHTML=v),k=o(t),u(X.$$.fragment,t),G=o(t),b=J(t,"P",{"data-svelte-h":!0}),T(b)!=="svelte-bytl0h"&&(b.textContent=E),x=o(t),u(C.$$.fragment,t),W=o(t),_=J(t,"P",{"data-svelte-h":!0}),T(_)!=="svelte-1yq6861"&&(_.textContent=V),A=o(t),u(Z.$$.fragment,t)},m(t,c){n(t,l,c),n(t,e,c),j(M,t,c),n(t,g,c),n(t,U,c),n(t,k,c),j(X,t,c),n(t,G,c),n(t,b,c),n(t,x,c),j(C,t,c),n(t,W,c),n(t,_,c),n(t,A,c),j(Z,t,c),B=!0},p:ts,i(t){B\|\|(w(M.$$.fragment,t),w(X.$$.fragment,t),w(C.$$.fragment,t),w(Z.$$.fragment,t),B=!0)},o(t){h(M.$$.fragment,t),h(X.$$.fragment,t),h(C.$$.fragment,t),h(Z.$$.fragment,t),B=!1},d(t){t&&(a(l),a(e),a(g),a(U),a(k),a(G),a(b),a(x),a(W),a(_),a(A)),d(M,t),d(X,t),d(C,t),d(Z,t)}}}function ue(I){let l,m;return l=new Gs({props:{$$slots:{default:[fe]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function je(I){let l,m="テキストをトークン化し、TensorFlow テンソルを返します。",e,M,g,U,v="入力をモデルに渡し、<code>logits</code>を返します。",k,X,G,b,E="モデル出力から開始位置と終了位置の最も高い確率を取得します。",x,C,W,_,V="予測されたトークンをデコードして答えを取得します。",A,Z,B;return M=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMEF1dG9Ub2tlbml6ZXIlMEElMEF0b2tlbml6ZXIlMjAlM0QlMjBBdXRvVG9rZW5pemVyLmZyb21fcHJldHJhaW5lZCglMjJteV9hd2Vzb21lX3FhX21vZGVsJTIyKSUwQWlucHV0cyUyMCUzRCUyMHRva2VuaXplcihxdWVzdGlvbiUyQyUyMHRleHQlMkMlMjByZXR1cm5fdGVuc29ycyUzRCUyMnRmJTIyKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer

	<span class="hljs-meta">>>> </span>tokenizer = AutoTokenizer.from_pretrained(<span class="hljs-string">"my_awesome_qa_model"</span>)
	<span class="hljs-meta">>>> </span>inputs = tokenizer(question, text, return_tensors=<span class="hljs-string">"tf"</span>)`,wrap:!1}}),X=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMFRGQXV0b01vZGVsRm9yUXVlc3Rpb25BbnN3ZXJpbmclMEElMEFtb2RlbCUyMCUzRCUyMFRGQXV0b01vZGVsRm9yUXVlc3Rpb25BbnN3ZXJpbmcuZnJvbV9wcmV0cmFpbmVkKCUyMm15X2F3ZXNvbWVfcWFfbW9kZWwlMjIpJTBBb3V0cHV0cyUyMCUzRCUyMG1vZGVsKCoqaW5wdXRzKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> TFAutoModelForQuestionAnswering

	<span class="hljs-meta">>>> </span>model = TFAutoModelForQuestionAnswering.from_pretrained(<span class="hljs-string">"my_awesome_qa_model"</span>)
	<span class="hljs-meta">>>> </span>outputs = model(**inputs)`,wrap:!1}}),C=new R({props:{code:"YW5zd2VyX3N0YXJ0X2luZGV4JTIwJTNEJTIwaW50KHRmLm1hdGguYXJnbWF4KG91dHB1dHMuc3RhcnRfbG9naXRzJTJDJTIwYXhpcyUzRC0xKSU1QjAlNUQpJTBBYW5zd2VyX2VuZF9pbmRleCUyMCUzRCUyMGludCh0Zi5tYXRoLmFyZ21heChvdXRwdXRzLmVuZF9sb2dpdHMlMkMlMjBheGlzJTNELTEpJTVCMCU1RCk=",highlighted:`<span class="hljs-meta">>>> </span>answer_start_index = <span class="hljs-built_in">int</span>(tf.math.argmax(outputs.start_logits, axis=-<span class="hljs-number">1</span>)[<span class="hljs-number">0</span>])
	<span class="hljs-meta">>>> </span>answer_end_index = <span class="hljs-built_in">int</span>(tf.math.argmax(outputs.end_logits, axis=-<span class="hljs-number">1</span>)[<span class="hljs-number">0</span>])`,wrap:!1}}),Z=new R({props:{code:"cHJlZGljdF9hbnN3ZXJfdG9rZW5zJTIwJTNEJTIwaW5wdXRzLmlucHV0X2lkcyU1QjAlMkMlMjBhbnN3ZXJfc3RhcnRfaW5kZXglMjAlM0ElMjBhbnN3ZXJfZW5kX2luZGV4JTIwJTJCJTIwMSU1RCUwQXRva2VuaXplci5kZWNvZGUocHJlZGljdF9hbnN3ZXJfdG9rZW5zKQ==",highlighted:`<span class="hljs-meta">>>> </span>predict_answer_tokens = inputs.input_ids[<span class="hljs-number">0</span>, answer_start_index : answer_end_index + <span class="hljs-number">1</span>]
	<span class="hljs-meta">>>> </span>tokenizer.decode(predict_answer_tokens)
	<span class="hljs-string">'176 billion parameters and can generate text in 46 languages natural languages and 13'</span>`,wrap:!1}}),{c(){l=y("p"),l.textContent=m,e=i(),f(M.$$.fragment),g=i(),U=y("p"),U.innerHTML=v,k=i(),f(X.$$.fragment),G=i(),b=y("p"),b.textContent=E,x=i(),f(C.$$.fragment),W=i(),_=y("p"),_.textContent=V,A=i(),f(Z.$$.fragment)},l(t){l=J(t,"P",{"data-svelte-h":!0}),T(l)!=="svelte-1bnmfqq"&&(l.textContent=m),e=o(t),u(M.$$.fragment,t),g=o(t),U=J(t,"P",{"data-svelte-h":!0}),T(U)!=="svelte-1rvunpz"&&(U.innerHTML=v),k=o(t),u(X.$$.fragment,t),G=o(t),b=J(t,"P",{"data-svelte-h":!0}),T(b)!=="svelte-bytl0h"&&(b.textContent=E),x=o(t),u(C.$$.fragment,t),W=o(t),_=J(t,"P",{"data-svelte-h":!0}),T(_)!=="svelte-1yq6861"&&(_.textContent=V),A=o(t),u(Z.$$.fragment,t)},m(t,c){n(t,l,c),n(t,e,c),j(M,t,c),n(t,g,c),n(t,U,c),n(t,k,c),j(X,t,c),n(t,G,c),n(t,b,c),n(t,x,c),j(C,t,c),n(t,W,c),n(t,_,c),n(t,A,c),j(Z,t,c),B=!0},p:ts,i(t){B\|\|(w(M.$$.fragment,t),w(X.$$.fragment,t),w(C.$$.fragment,t),w(Z.$$.fragment,t),B=!0)},o(t){h(M.$$.fragment,t),h(X.$$.fragment,t),h(C.$$.fragment,t),h(Z.$$.fragment,t),B=!1},d(t){t&&(a(l),a(e),a(g),a(U),a(k),a(G),a(b),a(x),a(W),a(_),a(A)),d(M,t),d(X,t),d(C,t),d(Z,t)}}}function we(I){let l,m;return l=new Gs({props:{$$slots:{default:[je]},$$scope:{ctx:I}}}),{c(){f(l.$$.fragment)},l(e){u(l.$$.fragment,e)},m(e,M){j(l,e,M),m=!0},p(e,M){const g={};M&2&&(g.$$scope={dirty:M,ctx:e}),l.$set(g)},i(e){m\|\|(w(l.$$.fragment,e),m=!0)},o(e){h(l.$$.fragment,e),m=!1},d(e){d(l,e)}}}function he(I){let l,m,e,M,g,U,v,k,X,G,b,E="質問応答タスクは、質問に対して回答を返します。 Alexa、Siri、Google などの仮想アシスタントに天気を尋ねたことがあるなら、質問応答モデルを使用したことがあるはずです。質問応答タスクには一般的に 2 つのタイプがあります。",x,C,W="<li>抽出: 与えられたコンテキストから回答を抽出します。</li> <li>抽象的: 質問に正しく答えるコンテキストから回答を生成します。</li>",_,V,A="このガイドでは、次の方法を説明します。",Z,B,t='<li>抽出的質問応答用に <a href="https://huggingface.co/datasets/squad" rel="nofollow">SQuAD</a> データセット上の <a href="https://huggingface.co/distilbert/distilbert-base-uncased" rel="nofollow">DistilBERT</a> を微調整します。</li> <li>微調整したモデルを推論に使用します。</li>',c,z,es,H,Y="始める前に、必要なライブラリがすべてインストールされていることを確認してください。",L,F,ls,Q,S="モデルをアップロードしてコミュニティと共有できるように、Hugging Face アカウントにログインすることをお勧めします。プロンプトが表示されたら、トークンを入力してログインします。",D,N,as,q,r,$,vs="まず、🤗 データセットライブラリから SQuAD データセットの小さいサブセットを読み込みます。これにより、完全なデータセットのトレーニングにさらに時間を費やす前に、実験してすべてが機能することを確認する機会が得られます。",Vs,ns,zs,ps,bt="<code>train_test_split</code> メソッドを使用して、データセットの <code>train</code> 分割をトレインセットとテストセットに分割します。",Es,rs,Fs,is,$t="次に、例を見てみましょう。",Ns,os,Hs,ms,Ut="ここにはいくつかの重要なフィールドがあります。",Qs,Ms,_t="<li><code>answers</code>: 回答トークンと回答テキストの開始位置。</li> <li><code>context</code>: モデルが答えを抽出するために必要な背景情報。</li> <li><code>question</code>: モデルが答える必要がある質問。</li>",qs,cs,Ys,fs,Ss,us,Ct="次のステップでは、DistilBERT トークナイザーをロードして<code>question</code>フィールドと<code>context</code>フィールドを処理します。",Ls,js,Ds,ws,It="質問応答タスクに特有の、注意すべき前処理手順がいくつかあります。",Ps,hs,Zt=`<li>データセット内の一部の例には、モデルの最大入力長を超える非常に長い「コンテキスト」が含まれる場合があります。より長いシーケンスを処理するには、<code>truncation="only_second"</code> を設定して <code>context</code> のみを切り捨てます。</li> <li>次に、設定によって、回答の開始位置と終了位置を元の <code>context</code>にマッピングします。
	「<code>return_offset_mapping=True</code>」。</li> <li>マッピングが手元にあるので、答えの開始トークンと終了トークンを見つけることができます。 <code>sequence_ids</code> メソッドを使用して、
	オフセットのどの部分が<code>question</code>に対応し、どの部分が<code>context</code>に対応するかを見つけます。</li>`,Ks,ds,kt="以下に、<code>answer</code>の開始トークンと終了トークンを切り詰めて<code>context</code>にマッピングする関数を作成する方法を示します。",Os,ys,st,Js,Xt="データセット全体に前処理関数を適用するには、🤗 Datasets <code>map</code> 関数を使用します。 <code>batched=True</code> を設定してデータセットの複数の要素を一度に処理することで、<code>map</code> 関数を高速化できます。不要な列を削除します。",tt,Ts,et,gs,Rt="次に、<code>DefaultDataCollator</code> を使用してサンプルのバッチを作成します。 🤗 Transformers の他のデータ照合器とは異なり、<code>DefaultDataCollator</code> はパディングなどの追加の前処理を適用しません。",lt,P,at,bs,nt,K,pt,O,rt,$s,it,Us,xt='質問応答の評価には、大量の後処理が必要です。時間がかかりすぎないように、このガイドでは評価ステップを省略しています。 <a href="/docs/transformers/pr_30690/ja/main_classes/trainer#transformers.Trainer">Trainer</a> はトレーニング中に評価損失を計算するため、モデルのパフォーマンスについて完全に分からないわけではありません。',ot,_s,Bt='もっと時間があり、質問応答用のモデルを評価する方法に興味がある場合は、<a href="https://huggingface.co/course/chapter7/7?fw=pt#postprocessing" rel="nofollow">質問応答</a> の章を参照してください。 🤗ハグフェイスコースから！',mt,Cs,Mt,Is,At="モデルを微調整したので、それを推論に使用できるようになりました。",ct,Zs,Gt="質問と、モデルに予測させたいコンテキストを考え出します。",ft,ks,ut,Xs,vt='推論用に微調整されたモデルを試す最も簡単な方法は、それを <a href="/docs/transformers/pr_30690/ja/main_classes/pipelines#transformers.pipeline">pipeline()</a> で使用することです。モデルを使用して質問応答用の<code>pipeline</code>をインスタンス化し、それにテキストを渡します。',jt,Rs,wt,xs,Wt="必要に応じて、<code>pipeline</code>の結果を手動で複製することもできます。",ht,ss,dt,Bs,yt,Ws,Jt;return g=new As({props:{title:"Question answering",local:"question-answering",headingTag:"h1"}}),v=new Ot({props:{classNames:"absolute z-10 right-0 top-0",options:[{label:"Mixed",value:"https://colab.research.google.com/github/huggingface/notebooks/blob/main/transformers_doc/ja/question_answering.ipynb"},{label:"PyTorch",value:"https://colab.research.google.com/github/huggingface/notebooks/blob/main/transformers_doc/ja/pytorch/question_answering.ipynb"},{label:"TensorFlow",value:"https://colab.research.google.com/github/huggingface/notebooks/blob/main/transformers_doc/ja/tensorflow/question_answering.ipynb"},{label:"Mixed",value:"https://studiolab.sagemaker.aws/import/github/huggingface/notebooks/blob/main/transformers_doc/ja/question_answering.ipynb"},{label:"PyTorch",value:"https://studiolab.sagemaker.aws/import/github/huggingface/notebooks/blob/main/transformers_doc/ja/pytorch/question_answering.ipynb"},{label:"TensorFlow",value:"https://studiolab.sagemaker.aws/import/github/huggingface/notebooks/blob/main/transformers_doc/ja/tensorflow/question_answering.ipynb"}]}}),X=new qt({props:{id:"ajPx5LwJD-I"}}),z=new Tt({props:{$$slots:{default:[te]},$$scope:{ctx:I}}}),F=new R({props:{code:"cGlwJTIwaW5zdGFsbCUyMHRyYW5zZm9ybWVycyUyMGRhdGFzZXRzJTIwZXZhbHVhdGU=",highlighted:"pip install transformers datasets evaluate",wrap:!1}}),N=new R({props:{code:"ZnJvbSUyMGh1Z2dpbmdmYWNlX2h1YiUyMGltcG9ydCUyMG5vdGVib29rX2xvZ2luJTBBJTBBbm90ZWJvb2tfbG9naW4oKQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> huggingface_hub <span class="hljs-keyword">import</span> notebook_login

	<span class="hljs-meta">>>> </span>notebook_login()`,wrap:!1}}),q=new As({props:{title:"Load SQuAD dataset",local:"load-squad-dataset",headingTag:"h2"}}),ns=new R({props:{code:"ZnJvbSUyMGRhdGFzZXRzJTIwaW1wb3J0JTIwbG9hZF9kYXRhc2V0JTBBJTBBc3F1YWQlMjAlM0QlMjBsb2FkX2RhdGFzZXQoJTIyc3F1YWQlMjIlMkMlMjBzcGxpdCUzRCUyMnRyYWluJTVCJTNBNTAwMCU1RCUyMik=",highlighted:`<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>squad = load_dataset(<span class="hljs-string">"squad"</span>, split=<span class="hljs-string">"train[:5000]"</span>)`,wrap:!1}}),rs=new R({props:{code:"c3F1YWQlMjAlM0QlMjBzcXVhZC50cmFpbl90ZXN0X3NwbGl0KHRlc3Rfc2l6ZSUzRDAuMik=",highlighted:'<span class="hljs-meta">>>> </span>squad = squad.train_test_split(test_size=<span class="hljs-number">0.2</span>)',wrap:!1}}),os=new R({props:{code:"c3F1YWQlNUIlMjJ0cmFpbiUyMiU1RCU1QjAlNUQ=",highlighted:`<span class="hljs-meta">>>> </span>squad[<span class="hljs-string">"train"</span>][<span class="hljs-number">0</span>]
	{<span class="hljs-string">'answers'</span>: {<span class="hljs-string">'answer_start'</span>: [<span class="hljs-number">515</span>], <span class="hljs-string">'text'</span>: [<span class="hljs-string">'Saint Bernadette Soubirous'</span>]},
	<span class="hljs-string">'context'</span>: <span class="hljs-string">'Architecturally, the school has a Catholic character. Atop the Main Building\\'s gold dome is a golden statue of the Virgin Mary. Immediately in front of the Main Building and facing it, is a copper statue of Christ with arms upraised with the legend "Venite Ad Me Omnes". Next to the Main Building is the Basilica of the Sacred Heart. Immediately behind the basilica is the Grotto, a Marian place of prayer and reflection. It is a replica of the grotto at Lourdes, France where the Virgin Mary reputedly appeared to Saint Bernadette Soubirous in 1858. At the end of the main drive (and in a direct line that connects through 3 statues and the Gold Dome), is a simple, modern stone statue of Mary.'</span>,
	<span class="hljs-string">'id'</span>: <span class="hljs-string">'5733be284776f41900661182'</span>,
	<span class="hljs-string">'question'</span>: <span class="hljs-string">'To whom did the Virgin Mary allegedly appear in 1858 in Lourdes France?'</span>,
	<span class="hljs-string">'title'</span>: <span class="hljs-string">'University_of_Notre_Dame'</span>
	}`,wrap:!1}}),cs=new As({props:{title:"Preprocess",local:"preprocess",headingTag:"h2"}}),fs=new qt({props:{id:"qgaM0weJHpA"}}),js=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMEF1dG9Ub2tlbml6ZXIlMEElMEF0b2tlbml6ZXIlMjAlM0QlMjBBdXRvVG9rZW5pemVyLmZyb21fcHJldHJhaW5lZCglMjJkaXN0aWxiZXJ0JTJGZGlzdGlsYmVydC1iYXNlLXVuY2FzZWQlMjIp",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer

	<span class="hljs-meta">>>> </span>tokenizer = AutoTokenizer.from_pretrained(<span class="hljs-string">"distilbert/distilbert-base-uncased"</span>)`,wrap:!1}}),ys=new R({props:{code:"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",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">def</span> <span class="hljs-title function_">preprocess_function</span>(<span class="hljs-params">examples</span>):
	<span class="hljs-meta">... </span> questions = [q.strip() <span class="hljs-keyword">for</span> q <span class="hljs-keyword">in</span> examples[<span class="hljs-string">"question"</span>]]
	<span class="hljs-meta">... </span> inputs = tokenizer(
	<span class="hljs-meta">... </span> questions,
	<span class="hljs-meta">... </span> examples[<span class="hljs-string">"context"</span>],
	<span class="hljs-meta">... </span> max_length=<span class="hljs-number">384</span>,
	<span class="hljs-meta">... </span> truncation=<span class="hljs-string">"only_second"</span>,
	<span class="hljs-meta">... </span> return_offsets_mapping=<span class="hljs-literal">True</span>,
	<span class="hljs-meta">... </span> padding=<span class="hljs-string">"max_length"</span>,
	<span class="hljs-meta">... </span> )

	<span class="hljs-meta">... </span> offset_mapping = inputs.pop(<span class="hljs-string">"offset_mapping"</span>)
	<span class="hljs-meta">... </span> answers = examples[<span class="hljs-string">"answers"</span>]
	<span class="hljs-meta">... </span> start_positions = []
	<span class="hljs-meta">... </span> end_positions = []

	<span class="hljs-meta">... </span> <span class="hljs-keyword">for</span> i, offset <span class="hljs-keyword">in</span> <span class="hljs-built_in">enumerate</span>(offset_mapping):
	<span class="hljs-meta">... </span> answer = answers[i]
	<span class="hljs-meta">... </span> start_char = answer[<span class="hljs-string">"answer_start"</span>][<span class="hljs-number">0</span>]
	<span class="hljs-meta">... </span> end_char = answer[<span class="hljs-string">"answer_start"</span>][<span class="hljs-number">0</span>] + <span class="hljs-built_in">len</span>(answer[<span class="hljs-string">"text"</span>][<span class="hljs-number">0</span>])
	<span class="hljs-meta">... </span> sequence_ids = inputs.sequence_ids(i)

	<span class="hljs-meta">... </span> <span class="hljs-comment"># Find the start and end of the context</span>
	<span class="hljs-meta">... </span> idx = <span class="hljs-number">0</span>
	<span class="hljs-meta">... </span> <span class="hljs-keyword">while</span> sequence_ids[idx] != <span class="hljs-number">1</span>:
	<span class="hljs-meta">... </span> idx += <span class="hljs-number">1</span>
	<span class="hljs-meta">... </span> context_start = idx
	<span class="hljs-meta">... </span> <span class="hljs-keyword">while</span> sequence_ids[idx] == <span class="hljs-number">1</span>:
	<span class="hljs-meta">... </span> idx += <span class="hljs-number">1</span>
	<span class="hljs-meta">... </span> context_end = idx - <span class="hljs-number">1</span>

	<span class="hljs-meta">... </span> <span class="hljs-comment"># If the answer is not fully inside the context, label it (0, 0)</span>
	<span class="hljs-meta">... </span> <span class="hljs-keyword">if</span> offset[context_start][<span class="hljs-number">0</span>] > end_char <span class="hljs-keyword">or</span> offset[context_end][<span class="hljs-number">1</span>] < start_char:
	<span class="hljs-meta">... </span> start_positions.append(<span class="hljs-number">0</span>)
	<span class="hljs-meta">... </span> end_positions.append(<span class="hljs-number">0</span>)
	<span class="hljs-meta">... </span> <span class="hljs-keyword">else</span>:
	<span class="hljs-meta">... </span> <span class="hljs-comment"># Otherwise it's the start and end token positions</span>
	<span class="hljs-meta">... </span> idx = context_start
	<span class="hljs-meta">... </span> <span class="hljs-keyword">while</span> idx <= context_end <span class="hljs-keyword">and</span> offset[idx][<span class="hljs-number">0</span>] <= start_char:
	<span class="hljs-meta">... </span> idx += <span class="hljs-number">1</span>
	<span class="hljs-meta">... </span> start_positions.append(idx - <span class="hljs-number">1</span>)

	<span class="hljs-meta">... </span> idx = context_end
	<span class="hljs-meta">... </span> <span class="hljs-keyword">while</span> idx >= context_start <span class="hljs-keyword">and</span> offset[idx][<span class="hljs-number">1</span>] >= end_char:
	<span class="hljs-meta">... </span> idx -= <span class="hljs-number">1</span>
	<span class="hljs-meta">... </span> end_positions.append(idx + <span class="hljs-number">1</span>)

	<span class="hljs-meta">... </span> inputs[<span class="hljs-string">"start_positions"</span>] = start_positions
	<span class="hljs-meta">... </span> inputs[<span class="hljs-string">"end_positions"</span>] = end_positions
	<span class="hljs-meta">... </span> <span class="hljs-keyword">return</span> inputs`,wrap:!1}}),Ts=new R({props:{code:"dG9rZW5pemVkX3NxdWFkJTIwJTNEJTIwc3F1YWQubWFwKHByZXByb2Nlc3NfZnVuY3Rpb24lMkMlMjBiYXRjaGVkJTNEVHJ1ZSUyQyUyMHJlbW92ZV9jb2x1bW5zJTNEc3F1YWQlNUIlMjJ0cmFpbiUyMiU1RC5jb2x1bW5fbmFtZXMp",highlighted:'<span class="hljs-meta">>>> </span>tokenized_squad = squad.<span class="hljs-built_in">map</span>(preprocess_function, batched=<span class="hljs-literal">True</span>, remove_columns=squad[<span class="hljs-string">"train"</span>].column_names)',wrap:!1}}),P=new gt({props:{pytorch:!0,tensorflow:!0,jax:!1,$$slots:{tensorflow:[ne],pytorch:[le]},$$scope:{ctx:I}}}),bs=new As({props:{title:"Train",local:"train",headingTag:"h2"}}),K=new gt({props:{pytorch:!0,tensorflow:!0,jax:!1,$$slots:{tensorflow:[Me],pytorch:[ie]},$$scope:{ctx:I}}}),O=new Tt({props:{$$slots:{default:[ce]},$$scope:{ctx:I}}}),$s=new As({props:{title:"Evaluate",local:"evaluate",headingTag:"h2"}}),Cs=new As({props:{title:"Inference",local:"inference",headingTag:"h2"}}),ks=new R({props:{code:"cXVlc3Rpb24lMjAlM0QlMjAlMjJIb3clMjBtYW55JTIwcHJvZ3JhbW1pbmclMjBsYW5ndWFnZXMlMjBkb2VzJTIwQkxPT00lMjBzdXBwb3J0JTNGJTIyJTBBY29udGV4dCUyMCUzRCUyMCUyMkJMT09NJTIwaGFzJTIwMTc2JTIwYmlsbGlvbiUyMHBhcmFtZXRlcnMlMjBhbmQlMjBjYW4lMjBnZW5lcmF0ZSUyMHRleHQlMjBpbiUyMDQ2JTIwbGFuZ3VhZ2VzJTIwbmF0dXJhbCUyMGxhbmd1YWdlcyUyMGFuZCUyMDEzJTIwcHJvZ3JhbW1pbmclMjBsYW5ndWFnZXMuJTIy",highlighted:`<span class="hljs-meta">>>> </span>question = <span class="hljs-string">"How many programming languages does BLOOM support?"</span>
	<span class="hljs-meta">>>> </span>context = <span class="hljs-string">"BLOOM has 176 billion parameters and can generate text in 46 languages natural languages and 13 programming languages."</span>`,wrap:!1}}),Rs=new R({props:{code:"ZnJvbSUyMHRyYW5zZm9ybWVycyUyMGltcG9ydCUyMHBpcGVsaW5lJTBBJTBBcXVlc3Rpb25fYW5zd2VyZXIlMjAlM0QlMjBwaXBlbGluZSglMjJxdWVzdGlvbi1hbnN3ZXJpbmclMjIlMkMlMjBtb2RlbCUzRCUyMm15X2F3ZXNvbWVfcWFfbW9kZWwlMjIpJTBBcXVlc3Rpb25fYW5zd2VyZXIocXVlc3Rpb24lM0RxdWVzdGlvbiUyQyUyMGNvbnRleHQlM0Rjb250ZXh0KQ==",highlighted:`<span class="hljs-meta">>>> </span><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> pipeline

	<span class="hljs-meta">>>> </span>question_answerer = pipeline(<span class="hljs-string">"question-answering"</span>, model=<span class="hljs-string">"my_awesome_qa_model"</span>)
	<span class="hljs-meta">>>> </span>question_answerer(question=question, context=context)
	{<span class="hljs-string">'score'</span>: <span class="hljs-number">0.2058267742395401</span>,
	<span class="hljs-string">'start'</span>: <span class="hljs-number">10</span>,
	<span class="hljs-string">'end'</span>: <span class="hljs-number">95</span>,
	<span class="hljs-string">'answer'</span>: <span class="hljs-string">'176 billion parameters and can generate text in 46 languages natural languages and 13'</span>}`,wrap:!1}}),ss=new gt({props:{pytorch:!0,tensorflow:!0,jax:!1,$$slots:{tensorflow:[we],pytorch:[ue]},$$scope:{ctx:I}}}),Bs=new se({props:{source:"https://github.com/huggingface/transformers/blob/main/docs/source/ja/tasks/question_answering.md"}}),{c(){l=y("meta"),m=i(),e=y("p"),M=i(),f(g.$$.fragment),U=i(),f(v.$$.fragment),k=i(),f(X.$$.fragment),G=i(),b=y("p"),b.textContent=E,x=i(),C=y("ul"),C.innerHTML=W,_=i(),V=y("p"),V.textContent=A,Z=i(),B=y("ol"),B.innerHTML=t,c=i(),f(z.$$.fragment),es=i(),H=y("p"),H.textContent=Y,L=i(),f(F.$$.fragment),ls=i(),Q=y("p"),Q.textContent=S,D=i(),f(N.$$.fragment),as=i(),f(q.$$.fragment),r=i(),$=y("p"),$.textContent=vs,Vs=i(),f(ns.$$.fragment),zs=i(),ps=y("p"),ps.innerHTML=bt,Es=i(),f(rs.$$.fragment),Fs=i(),is=y("p"),is.textContent=$t,Ns=i(),f(os.$$.fragment),Hs=i(),ms=y("p"),ms.textContent=Ut,Qs=i(),Ms=y("ul"),Ms.innerHTML=_t,qs=i(),f(cs.$$.fragment),Ys=i(),f(fs.$$.fragment),Ss=i(),us=y("p"),us.innerHTML=Ct,Ls=i(),f(js.$$.fragment),Ds=i(),ws=y("p"),ws.textContent=It,Ps=i(),hs=y("ol"),hs.innerHTML=Zt,Ks=i(),ds=y("p"),ds.innerHTML=kt,Os=i(),f(ys.$$.fragment),st=i(),Js=y("p"),Js.innerHTML=Xt,tt=i(),f(Ts.$$.fragment),et=i(),gs=y("p"),gs.innerHTML=Rt,lt=i(),f(P.$$.fragment),at=i(),f(bs.$$.fragment),nt=i(),f(K.$$.fragment),pt=i(),f(O.$$.fragment),rt=i(),f($s.$$.fragment),it=i(),Us=y("p"),Us.innerHTML=xt,ot=i(),_s=y("p"),_s.innerHTML=Bt,mt=i(),f(Cs.$$.fragment),Mt=i(),Is=y("p"),Is.textContent=At,ct=i(),Zs=y("p"),Zs.textContent=Gt,ft=i(),f(ks.$$.fragment),ut=i(),Xs=y("p"),Xs.innerHTML=vt,jt=i(),f(Rs.$$.fragment),wt=i(),xs=y("p"),xs.innerHTML=Wt,ht=i(),f(ss.$$.fragment),dt=i(),f(Bs.$$.fragment),yt=i(),Ws=y("p"),this.h()},l(s){const p=Pt("svelte-u9bgzb",document.head);l=J(p,"META",{name:!0,content:!0}),p.forEach(a),m=o(s),e=J(s,"P",{}),Ht(e).forEach(a),M=o(s),u(g.$$.fragment,s),U=o(s),u(v.$$.fragment,s),k=o(s),u(X.$$.fragment,s),G=o(s),b=J(s,"P",{"data-svelte-h":!0}),T(b)!=="svelte-1tgwu4t"&&(b.textContent=E),x=o(s),C=J(s,"UL",{"data-svelte-h":!0}),T(C)!=="svelte-ku9lav"&&(C.innerHTML=W),_=o(s),V=J(s,"P",{"data-svelte-h":!0}),T(V)!=="svelte-w5jzhi"&&(V.textContent=A),Z=o(s),B=J(s,"OL",{"data-svelte-h":!0}),T(B)!=="svelte-o857bo"&&(B.innerHTML=t),c=o(s),u(z.$$.fragment,s),es=o(s),H=J(s,"P",{"data-svelte-h":!0}),T(H)!=="svelte-1lya3k8"&&(H.textContent=Y),L=o(s),u(F.$$.fragment,s),ls=o(s),Q=J(s,"P",{"data-svelte-h":!0}),T(Q)!=="svelte-193zy02"&&(Q.textContent=S),D=o(s),u(N.$$.fragment,s),as=o(s),u(q.$$.fragment,s),r=o(s),$=J(s,"P",{"data-svelte-h":!0}),T($)!=="svelte-12zmynt"&&($.textContent=vs),Vs=o(s),u(ns.$$.fragment,s),zs=o(s),ps=J(s,"P",{"data-svelte-h":!0}),T(ps)!=="svelte-m4w3yq"&&(ps.innerHTML=bt),Es=o(s),u(rs.$$.fragment,s),Fs=o(s),is=J(s,"P",{"data-svelte-h":!0}),T(is)!=="svelte-1r6oj5w"&&(is.textContent=$t),Ns=o(s),u(os.$$.fragment,s),Hs=o(s),ms=J(s,"P",{"data-svelte-h":!0}),T(ms)!=="svelte-y9dab"&&(ms.textContent=Ut),Qs=o(s),Ms=J(s,"UL",{"data-svelte-h":!0}),T(Ms)!=="svelte-1kxnjgl"&&(Ms.innerHTML=_t),qs=o(s),u(cs.$$.fragment,s),Ys=o(s),u(fs.$$.fragment,s),Ss=o(s),us=J(s,"P",{"data-svelte-h":!0}),T(us)!=="svelte-1bamyze"&&(us.innerHTML=Ct),Ls=o(s),u(js.$$.fragment,s),Ds=o(s),ws=J(s,"P",{"data-svelte-h":!0}),T(ws)!=="svelte-lhw9mt"&&(ws.textContent=It),Ps=o(s),hs=J(s,"OL",{"data-svelte-h":!0}),T(hs)!=="svelte-1syxwtd"&&(hs.innerHTML=Zt),Ks=o(s),ds=J(s,"P",{"data-svelte-h":!0}),T(ds)!=="svelte-1ryfrne"&&(ds.innerHTML=kt),Os=o(s),u(ys.$$.fragment,s),st=o(s),Js=J(s,"P",{"data-svelte-h":!0}),T(Js)!=="svelte-t1eqy4"&&(Js.innerHTML=Xt),tt=o(s),u(Ts.$$.fragment,s),et=o(s),gs=J(s,"P",{"data-svelte-h":!0}),T(gs)!=="svelte-me6jz3"&&(gs.innerHTML=Rt),lt=o(s),u(P.$$.fragment,s),at=o(s),u(bs.$$.fragment,s),nt=o(s),u(K.$$.fragment,s),pt=o(s),u(O.$$.fragment,s),rt=o(s),u($s.$$.fragment,s),it=o(s),Us=J(s,"P",{"data-svelte-h":!0}),T(Us)!=="svelte-f1f1ff"&&(Us.innerHTML=xt),ot=o(s),_s=J(s,"P",{"data-svelte-h":!0}),T(_s)!=="svelte-ihy4ok"&&(_s.innerHTML=Bt),mt=o(s),u(Cs.$$.fragment,s),Mt=o(s),Is=J(s,"P",{"data-svelte-h":!0}),T(Is)!=="svelte-cyrfc8"&&(Is.textContent=At),ct=o(s),Zs=J(s,"P",{"data-svelte-h":!0}),T(Zs)!=="svelte-151wogl"&&(Zs.textContent=Gt),ft=o(s),u(ks.$$.fragment,s),ut=o(s),Xs=J(s,"P",{"data-svelte-h":!0}),T(Xs)!=="svelte-1igkpvv"&&(Xs.innerHTML=vt),jt=o(s),u(Rs.$$.fragment,s),wt=o(s),xs=J(s,"P",{"data-svelte-h":!0}),T(xs)!=="svelte-p649vi"&&(xs.innerHTML=Wt),ht=o(s),u(ss.$$.fragment,s),dt=o(s),u(Bs.$$.fragment,s),yt=o(s),Ws=J(s,"P",{}),Ht(Ws).forEach(a),this.h()},h(){Qt(l,"name","hf:doc:metadata"),Qt(l,"content",de)},m(s,p){Kt(document.head,l),n(s,m,p),n(s,e,p),n(s,M,p),j(g,s,p),n(s,U,p),j(v,s,p),n(s,k,p),j(X,s,p),n(s,G,p),n(s,b,p),n(s,x,p),n(s,C,p),n(s,_,p),n(s,V,p),n(s,Z,p),n(s,B,p),n(s,c,p),j(z,s,p),n(s,es,p),n(s,H,p),n(s,L,p),j(F,s,p),n(s,ls,p),n(s,Q,p),n(s,D,p),j(N,s,p),n(s,as,p),j(q,s,p),n(s,r,p),n(s,$,p),n(s,Vs,p),j(ns,s,p),n(s,zs,p),n(s,ps,p),n(s,Es,p),j(rs,s,p),n(s,Fs,p),n(s,is,p),n(s,Ns,p),j(os,s,p),n(s,Hs,p),n(s,ms,p),n(s,Qs,p),n(s,Ms,p),n(s,qs,p),j(cs,s,p),n(s,Ys,p),j(fs,s,p),n(s,Ss,p),n(s,us,p),n(s,Ls,p),j(js,s,p),n(s,Ds,p),n(s,ws,p),n(s,Ps,p),n(s,hs,p),n(s,Ks,p),n(s,ds,p),n(s,Os,p),j(ys,s,p),n(s,st,p),n(s,Js,p),n(s,tt,p),j(Ts,s,p),n(s,et,p),n(s,gs,p),n(s,lt,p),j(P,s,p),n(s,at,p),j(bs,s,p),n(s,nt,p),j(K,s,p),n(s,pt,p),j(O,s,p),n(s,rt,p),j($s,s,p),n(s,it,p),n(s,Us,p),n(s,ot,p),n(s,_s,p),n(s,mt,p),j(Cs,s,p),n(s,Mt,p),n(s,Is,p),n(s,ct,p),n(s,Zs,p),n(s,ft,p),j(ks,s,p),n(s,ut,p),n(s,Xs,p),n(s,jt,p),j(Rs,s,p),n(s,wt,p),n(s,xs,p),n(s,ht,p),j(ss,s,p),n(s,dt,p),j(Bs,s,p),n(s,yt,p),n(s,Ws,p),Jt=!0},p(s,[p]){const Vt={};p&2&&(Vt.$$scope={dirty:p,ctx:s}),z.$set(Vt);const zt={};p&2&&(zt.$$scope={dirty:p,ctx:s}),P.$set(zt);const Et={};p&2&&(Et.$$scope={dirty:p,ctx:s}),K.$set(Et);const Ft={};p&2&&(Ft.$$scope={dirty:p,ctx:s}),O.$set(Ft);const Nt={};p&2&&(Nt.$$scope={dirty:p,ctx:s}),ss.$set(Nt)},i(s){Jt\|\|(w(g.$$.fragment,s),w(v.$$.fragment,s),w(X.$$.fragment,s),w(z.$$.fragment,s),w(F.$$.fragment,s),w(N.$$.fragment,s),w(q.$$.fragment,s),w(ns.$$.fragment,s),w(rs.$$.fragment,s),w(os.$$.fragment,s),w(cs.$$.fragment,s),w(fs.$$.fragment,s),w(js.$$.fragment,s),w(ys.$$.fragment,s),w(Ts.$$.fragment,s),w(P.$$.fragment,s),w(bs.$$.fragment,s),w(K.$$.fragment,s),w(O.$$.fragment,s),w($s.$$.fragment,s),w(Cs.$$.fragment,s),w(ks.$$.fragment,s),w(Rs.$$.fragment,s),w(ss.$$.fragment,s),w(Bs.$$.fragment,s),Jt=!0)},o(s){h(g.$$.fragment,s),h(v.$$.fragment,s),h(X.$$.fragment,s),h(z.$$.fragment,s),h(F.$$.fragment,s),h(N.$$.fragment,s),h(q.$$.fragment,s),h(ns.$$.fragment,s),h(rs.$$.fragment,s),h(os.$$.fragment,s),h(cs.$$.fragment,s),h(fs.$$.fragment,s),h(js.$$.fragment,s),h(ys.$$.fragment,s),h(Ts.$$.fragment,s),h(P.$$.fragment,s),h(bs.$$.fragment,s),h(K.$$.fragment,s),h(O.$$.fragment,s),h($s.$$.fragment,s),h(Cs.$$.fragment,s),h(ks.$$.fragment,s),h(Rs.$$.fragment,s),h(ss.$$.fragment,s),h(Bs.$$.fragment,s),Jt=!1},d(s){s&&(a(m),a(e),a(M),a(U),a(k),a(G),a(b),a(x),a(C),a(_),a(V),a(Z),a(B),a(c),a(es),a(H),a(L),a(ls),a(Q),a(D),a(as),a(r),a($),a(Vs),a(zs),a(ps),a(Es),a(Fs),a(is),a(Ns),a(Hs),a(ms),a(Qs),a(Ms),a(qs),a(Ys),a(Ss),a(us),a(Ls),a(Ds),a(ws),a(Ps),a(hs),a(Ks),a(ds),a(Os),a(st),a(Js),a(tt),a(et),a(gs),a(lt),a(at),a(nt),a(pt),a(rt),a(it),a(Us),a(ot),a(_s),a(mt),a(Mt),a(Is),a(ct),a(Zs),a(ft),a(ut),a(Xs),a(jt),a(wt),a(xs),a(ht),a(dt),a(yt),a(Ws)),a(l),d(g,s),d(v,s),d(X,s),d(z,s),d(F,s),d(N,s),d(q,s),d(ns,s),d(rs,s),d(os,s),d(cs,s),d(fs,s),d(js,s),d(ys,s),d(Ts,s),d(P,s),d(bs,s),d(K,s),d(O,s),d($s,s),d(Cs,s),d(ks,s),d(Rs,s),d(ss,s),d(Bs,s)}}}const de='{"title":"Question answering","local":"question-answering","sections":[{"title":"Load SQuAD dataset","local":"load-squad-dataset","sections":[],"depth":2},{"title":"Preprocess","local":"preprocess","sections":[],"depth":2},{"title":"Train","local":"train","sections":[],"depth":2},{"title":"Evaluate","local":"evaluate","sections":[],"depth":2},{"title":"Inference","local":"inference","sections":[],"depth":2}],"depth":1}';function ye(I){return St(()=>{new URLSearchParams(window.location.search).get("fw")}),[]}class Ie extends Lt{constructor(l){super(),Dt(this,l,ye,he,Yt,{})}}export{Ie as component};

Xet Storage Details

Size:: 60.5 kB
Xet hash:: 0b50cde2125f14566ab1f16595ac5551ef6d754693145ae91ff3469363d887e7

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