transformers/main/ko/attention.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;어텐션 메커니즘&quot;,&quot;local&quot;:&quot;attention_mechanisms&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;LSH 어텐션&quot;,&quot;local&quot;:&quot;lsh_attention&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;지역 어텐션&quot;,&quot;local&quot;:&quot;local_attention&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;다른 방법들&quot;,&quot;local&quot;:&quot;other_tricks&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;축별 위치 인코딩&quot;,&quot;local&quot;:&quot;axial_positional_encodings&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
		<link href="/docs/transformers/main/ko/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/entry/start.9aa88961.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/scheduler.9bc65507.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/singletons.9eec45c3.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/index.3b203c72.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/paths.566078f7.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/entry/app.84fb67c3.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/index.707bf1b6.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/nodes/0.1c99376b.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/each.e59479a4.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/nodes/5.6269901b.js">
		<link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/EditOnGithub.922df6ba.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;어텐션 메커니즘&quot;,&quot;local&quot;:&quot;attention_mechanisms&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;LSH 어텐션&quot;,&quot;local&quot;:&quot;lsh_attention&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;지역 어텐션&quot;,&quot;local&quot;:&quot;local_attention&quot;,&quot;sections&quot;:[],&quot;depth&quot;:2},{&quot;title&quot;:&quot;다른 방법들&quot;,&quot;local&quot;:&quot;other_tricks&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;축별 위치 인코딩&quot;,&quot;local&quot;:&quot;axial_positional_encodings&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>   <h1 class="relative group"><a id="attention_mechanisms" 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="#attention_mechanisms"><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>어텐션 메커니즘</span></h1> <p data-svelte-h="svelte-jt294z">대부분의 트랜스포머 모델은 정방행렬인 전체 어텐션을 사용합니다.
하지만 이는 긴 텍스트를 다룰 때는 큰 계산 병목 현상을 유발할 수 있습니다.
<code>Longformer</code>와 <code>Reformer</code>는 훈련 속도를 높이기 위해 어텐션 행렬의 희소 버전을 사용하여 효율을 높이려는 모델입니다.</p>  <h2 class="relative group"><a id="lsh_attention" 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="#lsh_attention"><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>LSH 어텐션</span></h2> <p data-svelte-h="svelte-6wbt6"><a href="model_doc/reformer">Reformer</a>는 LSH(Locality Sensitive Hashing) 어텐션을 사용합니다. softmax(QK^t)에서는 행렬 QK^t의 (softmax 차원에서) 가장 큰 요소들만 유용한 기여를 할 것입니다.
따라서 각각의 쿼리 q에 대해, q와 가까운 키 k만 고려할 수 있습니다. 해시 함수는 q와 k가 가까운지 여부를 결정하는 데 사용됩니다.
어텐션 마스크는 현재 토큰을 마스킹하여 변경됩니다. 이 때 첫 번째 위치의 토큰은 제외합니다. 왜냐하면 쿼리와 키가 동일한 값을 갖게 되기 때문입니다(서로 매우 유사함).
해시는 약간의 무작위성을 가질 수 있으므로, 실제로는 여러 개의 해시 함수가 사용되고 (<code>n_rounds</code> 매개변수에 의해 결정됨) 그 후에 평균값을 취하게 됩니다.</p>  <h2 class="relative group"><a id="local_attention" 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="#local_attention"><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>지역 어텐션</span></h2> <p data-svelte-h="svelte-1195vp3"><a href="model_doc/longformer">Longformer</a>는 지역 어텐션을 사용합니다. 종종 특정 토큰에 대해 지역 컨텍스트(예: 왼쪽과 오른쪽에 있는 두 개의 토큰은 무엇인가요?)만으로도 작업을 수행하는데 충분합니다.
또한 작은 창(window)을 가진 어텐션 레이어를 쌓음으로써 마지막 레이어는 창 내의 토큰뿐만 아니라 더 많은 수의 토큰에 대한 수용 영역(receptive field)을 갖게 되어 전체 문장의 표현을 구축할 수 있습니다.</p> <p data-svelte-h="svelte-rbnmsa">사전에 선택된 일부 입력 토큰들은 전역 어텐션을 받습니다. 이 몇 개의 토큰에 대해서는 어텐션 행렬이 모든 토큰에 접근할 수 있으며, 이 과정은 대칭적으로 이루어집니다.
다른 모든 토큰들은 로컬 창 내의 토큰들에 더해 해당 특정 토큰들에도 접근할 수 있습니다. 이는 논문의 Figure 2d에서 나타나며, 아래에 샘플 어텐션 마스크가 제시되어 있습니다:</p> <div class="flex justify-center" data-svelte-h="svelte-d4kpls"><img scale="50 %" align="center" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/local_attention_mask.png"></div> <p data-svelte-h="svelte-1jaot4s">적은 파라미터의 어텐션 행렬을 사용하면 모델이 더 큰 시퀀스 입력 길이를 가질 수 있습니다.</p>  <h2 class="relative group"><a id="other_tricks" 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="#other_tricks"><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>다른 방법들</span></h2>  <h3 class="relative group"><a id="axial_positional_encodings" 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="#axial_positional_encodings"><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>축별 위치 인코딩</span></h3> <p><a href="model_doc/reformer" data-svelte-h="svelte-zf8r8v">Reformer</a>는 축별 위치 인코딩(axial positional encodings)을 사용합니다. 기존의 트랜스포머 모델에서는 위치 인코딩 행렬 E는 크기가<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>l</mi><mo>×</mo><mi>d</mi></mrow><annotation encoding="application/x-tex">l \times d</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.7778em;vertical-align:-0.0833em;"></span><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord mathnormal">d</span></span></span></span><!-- HTML_TAG_END -->인 행렬이며,
여기서<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>l</mi></mrow><annotation encoding="application/x-tex">l</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord mathnormal" style="margin-right:0.01968em;">l</span></span></span></span><!-- HTML_TAG_END -->은 시퀀스 길이(sequence length)이고<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>d</mi></mrow><annotation encoding="application/x-tex">d</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord mathnormal">d</span></span></span></span><!-- HTML_TAG_END -->는 숨겨진 상태(hidden state)의 차원입니다. 매우 긴 텍스트의 경우, 이 행렬은 매우 크며 GPU 상에서 공간을 많이 차지할 수 있습니다.
이를 완화하기 위해, 축별 위치 인코딩은 큰 행렬 E를 두 개의 작은 행렬 E1과 E2로 분해합니다. 이때 E1의 크기는<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>l</mi><mn>1</mn></msub><mo>×</mo><msub><mi>d</mi><mn>1</mn></msub></mrow><annotation encoding="application/x-tex">l_{1} \times d_{1}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:-0.0197em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">1</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">d</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">1</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span><!-- HTML_TAG_END -->이고, E2의 크기는<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>l</mi><mn>2</mn></msub><mo>×</mo><msub><mi>d</mi><mn>2</mn></msub></mrow><annotation encoding="application/x-tex">l_{2} \times d_{2}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:-0.0197em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">d</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span></span></span></span><!-- HTML_TAG_END -->입니다.
이때<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>l</mi><mn>1</mn></msub><mo>×</mo><msub><mi>l</mi><mn>2</mn></msub><mo>=</mo><mi>l</mi></mrow><annotation encoding="application/x-tex">l_{1} \times l_{2} = l</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:-0.0197em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">1</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">×</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:-0.0197em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2778em;"></span></span><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord mathnormal" style="margin-right:0.01968em;">l</span></span></span></span><!-- HTML_TAG_END -->이고<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>d</mi><mn>1</mn></msub><mo>+</mo><msub><mi>d</mi><mn>2</mn></msub><mo>=</mo><mi>d</mi></mrow><annotation encoding="application/x-tex">d_{1} + d_{2} = d</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">d</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">1</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2222em;"></span><span class="mbin">+</span><span class="mspace" style="margin-right:0.2222em;"></span></span><span class="base"><span class="strut" style="height:0.8444em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">d</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3011em;"><span style="top:-2.55em;margin-left:0em;margin-right:0.05em;"><span class="pstrut" style="height:2.7em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mtight"><span class="mord mtight">2</span></span></span></span></span><span class="vlist-s">​</span></span><span class="vlist-r"><span class="vlist" style="height:0.15em;"><span></span></span></span></span></span></span><span class="mspace" style="margin-right:0.2778em;"></span><span class="mrel">=</span><span class="mspace" style="margin-right:0.2778em;"></span></span><span class="base"><span class="strut" style="height:0.6944em;"></span><span class="mord mathnormal">d</span></span></span></span><!-- HTML_TAG_END -->(길이에 대한 곱셈 연산을 사용하면 훨씬 작아집니다). E의 시간 단계 j에 대한 임베딩은 E1에서 시간 단계<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>j</mi><mi mathvariant="normal">%</mi><mi>l</mi><mn>1</mn></mrow><annotation encoding="application/x-tex">j \% l1</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.9444em;vertical-align:-0.1944em;"></span><span class="mord mathnormal" style="margin-right:0.05724em;">j</span><span class="mord">%</span><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="mord">1</span></span></span></span><!-- HTML_TAG_END -->의 임베딩과 E2에서 시간 단계 <!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>j</mi><mi mathvariant="normal">/</mi><mi mathvariant="normal">/</mi><mi>l</mi><mn>1</mn></mrow><annotation encoding="application/x-tex">j // l1</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:1em;vertical-align:-0.25em;"></span><span class="mord mathnormal" style="margin-right:0.05724em;">j</span><span class="mord">//</span><span class="mord mathnormal" style="margin-right:0.01968em;">l</span><span class="mord">1</span></span></span></span><!-- HTML_TAG_END -->의 임베딩을 연결하여 얻습니다.</p> <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/ko/attention.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_1hrx8 = {
						assets: "/docs/transformers/main/ko",
						base: "/docs/transformers/main/ko",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/transformers/main/ko/_app/immutable/entry/start.9aa88961.js"),
						import("/docs/transformers/main/ko/_app/immutable/entry/app.84fb67c3.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 5],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>