Buckets:
| <meta charset="utf-8" /><meta name="hf:doc:metadata" content="{"title":"고정 길이 모델의 펄플렉서티(Perplexity)","local":"perplexity-of-fixedlength-models","sections":[{"title":"고정 길이 모델의 펄플렉서티(PPL) 계산하기","local":"calculating-ppl-with-fixedlength-models","sections":[],"depth":2},{"title":"예제: 🤗 Transformers에서 GPT-2로 펄플렉서티(perplexity) 계산하기","local":"example-calculating-perplexity-with-gpt2-in-transformers","sections":[],"depth":2}],"depth":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/45.9db944af.js"> | |
| <link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/CodeBlock.54a9f38d.js"> | |
| <link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/DocNotebookDropdown.41f65cb5.js"> | |
| <link rel="modulepreload" href="/docs/transformers/main/ko/_app/immutable/chunks/globals.7f7f1b26.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="{"title":"고정 길이 모델의 펄플렉서티(Perplexity)","local":"perplexity-of-fixedlength-models","sections":[{"title":"고정 길이 모델의 펄플렉서티(PPL) 계산하기","local":"calculating-ppl-with-fixedlength-models","sections":[],"depth":2},{"title":"예제: 🤗 Transformers에서 GPT-2로 펄플렉서티(perplexity) 계산하기","local":"example-calculating-perplexity-with-gpt2-in-transformers","sections":[],"depth":2}],"depth":1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <h1 class="relative group"><a id="perplexity-of-fixedlength-models" 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="#perplexity-of-fixedlength-models"><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>고정 길이 모델의 펄플렉서티(Perplexity)</span></h1> <div class="flex space-x-1 absolute z-10 right-0 top-0"> <div class="relative colab-dropdown "> <button class=" " type="button"> <img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg"> </button> </div> <div class="relative colab-dropdown "> <button class=" " type="button"> <img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg"> </button> </div></div> <p data-svelte-h="svelte-11w5q0v">펄플렉서티(Perplexity, PPL)는 가장 일반적인 언어 모델 평가지표 중 하나입니다. | |
| 자세히 알아보기 전에 이 평가지표는 고전적인 언어 모델(자기회귀 또는 인과적 언어 모델이라고도 함)에만 적용되며 BERT와 같은 마스킹된 언어 모델에는 잘 적용하지 않습니다 (BERT는 <a href="../en/model_summary">summary of the models</a> 문서를 참고하세요).</p> <p>펄플렉서티는 시퀀스의 음의 로그 우도(negative log-likelihood, NLL) 값의 평균에 지수(exponentiate)를 취한 값으로 정의됩니다. | |
| 토큰화된 시퀀스<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>X</mi><mo>=</mo><mo stretchy="false">(</mo><msub><mi>x</mi><mn>0</mn></msub><mo separator="true">,</mo><msub><mi>x</mi><mn>1</mn></msub><mo separator="true">,</mo><mo>…</mo><mo separator="true">,</mo><msub><mi>x</mi><mi>t</mi></msub><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">X = (x_0, x_1, \dots, x_t)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6833em;"></span><span class="mord mathnormal" style="margin-right:0.07847em;">X</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:1em;vertical-align:-0.25em;"></span><span class="mopen">(</span><span class="mord"><span class="mord mathnormal">x</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">0</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="mpunct">,</span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mord"><span class="mord mathnormal">x</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">1</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="mpunct">,</span><span class="mspace" style="margin-right:0.1667em;"></span><span class="minner">…</span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mpunct">,</span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.2806em;"><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 mathnormal mtight">t</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="mclose">)</span></span></span></span><!-- HTML_TAG_END --> 가 있을 때,<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>X</mi></mrow><annotation encoding="application/x-tex">X</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6833em;"></span><span class="mord mathnormal" style="margin-right:0.07847em;">X</span></span></span></span><!-- HTML_TAG_END --> 의 펄플렉서티는 아래 수식과 같이 구할 수 있습니다. | |
| <!-- HTML_TAG_START --><span class="katex-display"><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML" display="block"><semantics><mrow><mtext>PPL</mtext><mo stretchy="false">(</mo><mi>X</mi><mo stretchy="false">)</mo><mo>=</mo><mi>exp</mi><mo></mo><mrow><mo fence="true">{</mo><mrow><mo>−</mo><mfrac><mn>1</mn><mi>t</mi></mfrac><munderover><mo>∑</mo><mi>i</mi><mi>t</mi></munderover><mi>log</mi><mo></mo><msub><mi>p</mi><mi>θ</mi></msub><mo stretchy="false">(</mo><msub><mi>x</mi><mi>i</mi></msub><mi mathvariant="normal">∣</mi><msub><mi>x</mi><mrow><mo><</mo><mi>i</mi></mrow></msub><mo stretchy="false">)</mo></mrow><mo fence="true">}</mo></mrow></mrow><annotation encoding="application/x-tex">\text{PPL}(X) = \exp \left\{ {-\frac{1}{t}\sum_i^t \log p_\theta (x_i|x_{<i}) } \right\}</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 text"><span class="mord">PPL</span></span><span class="mopen">(</span><span class="mord mathnormal" style="margin-right:0.07847em;">X</span><span class="mclose">)</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:3.0582em;vertical-align:-1.2777em;"></span><span class="mop">exp</span><span class="mspace" style="margin-right:0.1667em;"></span><span class="minner"><span class="mopen delimcenter" style="top:0em;"><span class="delimsizing size4">{</span></span><span class="mord"><span class="mord">−</span><span class="mord"><span class="mopen nulldelimiter"></span><span class="mfrac"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:1.3214em;"><span style="top:-2.314em;"><span class="pstrut" style="height:3em;"></span><span class="mord"><span class="mord mathnormal">t</span></span></span><span style="top:-3.23em;"><span class="pstrut" style="height:3em;"></span><span class="frac-line" style="border-bottom-width:0.04em;"></span></span><span style="top:-3.677em;"><span class="pstrut" style="height:3em;"></span><span class="mord"><span class="mord">1</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.686em;"><span></span></span></span></span></span><span class="mclose nulldelimiter"></span></span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mop op-limits"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:1.7806em;"><span style="top:-1.8723em;margin-left:0em;"><span class="pstrut" style="height:3.05em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mathnormal mtight">i</span></span></span><span style="top:-3.05em;"><span class="pstrut" style="height:3.05em;"></span><span><span class="mop op-symbol large-op">∑</span></span></span><span style="top:-4.3em;margin-left:0em;"><span class="pstrut" style="height:3.05em;"></span><span class="sizing reset-size6 size3 mtight"><span class="mord mathnormal mtight">t</span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:1.2777em;"><span></span></span></span></span></span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mop">lo<span style="margin-right:0.01389em;">g</span></span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mord"><span class="mord mathnormal">p</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><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 mathnormal mtight" style="margin-right:0.02778em;">θ</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="mopen">(</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3117em;"><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 mathnormal mtight">i</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="mord">∣</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3117em;"><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="mrel mtight"><</span><span class="mord mathnormal mtight">i</span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.1774em;"><span></span></span></span></span></span></span><span class="mclose">)</span></span><span class="mclose delimcenter" style="top:0em;"><span class="delimsizing size4">}</span></span></span></span></span></span></span><!-- HTML_TAG_END --> <!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>log</mi><mo></mo><msub><mi>p</mi><mi>θ</mi></msub><mo stretchy="false">(</mo><msub><mi>x</mi><mi>i</mi></msub><mi mathvariant="normal">∣</mi><msub><mi>x</mi><mrow><mo><</mo><mi>i</mi></mrow></msub><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">\log p_\theta (x_i|x_{<i})</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="mop">lo<span style="margin-right:0.01389em;">g</span></span><span class="mspace" style="margin-right:0.1667em;"></span><span class="mord"><span class="mord mathnormal">p</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><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 mathnormal mtight" style="margin-right:0.02778em;">θ</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="mopen">(</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3117em;"><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 mathnormal mtight">i</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="mord">∣</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3117em;"><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="mrel mtight"><</span><span class="mord mathnormal mtight">i</span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.1774em;"><span></span></span></span></span></span></span><span class="mclose">)</span></span></span></span><!-- HTML_TAG_END --> 는 모델에 i번째 이전까지 토큰이 주어졌을 때 i번째 토큰의 로그 우도값입니다.</p> <p data-svelte-h="svelte-1b4v3xq">직관적으로 말뭉치에서 지정된 토큰 집합을 균일하게 예측하는 모델의 능력에 대한 평가로 생각할 수 있습니다. | |
| 중요한 점은 토큰화 과정이 모델의 펄플렉서티에 직접적인 영향을 미치므로 서로 다른 모델을 비교할 때 항상 이를 고려해야 합니다.</p> <p data-svelte-h="svelte-33m9h2">이는 데이터와 모델 예측 간의 cross-entropy 값에 지수를 취한 것과 동일합니다. | |
| 펄플렉서티와 문자당 비트 수(BPC) 및 데이터 압축과의 관계에 대해 더 직관적인 이해를 원하신다면 다음 글 | |
| <a href="https://thegradient.pub/understanding-evaluation-metrics-for-language-models/" rel="nofollow">fantastic blog post on The Gradient</a>을 확인하세요.</p> <h2 class="relative group"><a id="calculating-ppl-with-fixedlength-models" 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="#calculating-ppl-with-fixedlength-models"><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>고정 길이 모델의 펄플렉서티(PPL) 계산하기</span></h2> <p data-svelte-h="svelte-1sg414w">모델의 컨텍스트 크기가 정해져있지 않다면, | |
| 아래와 같이 시퀀스를 자동 회귀적으로 분해하고 각 단계에서 선행 하는 전체 시퀀스를 조건부 확률에 넣어 모델의 펄플렉서티를 계산할 것입니다.</p> <img width="600" alt="Full decomposition of a sequence with unlimited context length" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_full.gif"> <p>그러나 모델의 근사치를 구할 때는 일반적으로 모델이 처리할 수 있는 토큰 수에 제한이 있습니다. | |
| 예를 들어, 가장 큰 버전의 <a href="model_doc/gpt2" data-svelte-h="svelte-1kdeo4m">GPT-2</a>는 토큰의 길이가 1024로 고정되어 있습니다. | |
| 따라서<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>t</mi></mrow><annotation encoding="application/x-tex">t</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.6151em;"></span><span class="mord mathnormal">t</span></span></span></span><!-- HTML_TAG_END --> 가 1024보다 큰 경우에<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>p</mi><mi>θ</mi></msub><mo stretchy="false">(</mo><msub><mi>x</mi><mi>t</mi></msub><mi mathvariant="normal">∣</mi><msub><mi>x</mi><mrow><mo><</mo><mi>t</mi></mrow></msub><mo stretchy="false">)</mo></mrow><annotation encoding="application/x-tex">p_\theta(x_t|x_{<t})</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"><span class="mord mathnormal">p</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.3361em;"><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 mathnormal mtight" style="margin-right:0.02778em;">θ</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="mopen">(</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.2806em;"><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 mathnormal mtight">t</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="mord">∣</span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.2806em;"><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="mrel mtight"><</span><span class="mord mathnormal mtight">t</span></span></span></span></span><span class="vlist-s"></span></span><span class="vlist-r"><span class="vlist" style="height:0.1774em;"><span></span></span></span></span></span></span><span class="mclose">)</span></span></span></span><!-- HTML_TAG_END --> 을 계산할 수 없습니다.</p> <p>대신 시퀀스는 일반적으로 모델의 최대 입력 크기와 동일한 길이는 가지는 부분 시퀀스로 쪼갭니다. | |
| 만약 모델의 최대 입력 길이가<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>k</mi></mrow><annotation encoding="application/x-tex">k</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:0.03148em;">k</span></span></span></span><!-- HTML_TAG_END --> 라면, | |
| 토큰<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><msub><mi>x</mi><mi>t</mi></msub></mrow><annotation encoding="application/x-tex">x_t</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="base"><span class="strut" style="height:0.5806em;vertical-align:-0.15em;"></span><span class="mord"><span class="mord mathnormal">x</span><span class="msupsub"><span class="vlist-t vlist-t2"><span class="vlist-r"><span class="vlist" style="height:0.2806em;"><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 mathnormal mtight">t</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 --> 의 우도 값을 계산할 때 이전 토큰을 모두 사용하지 않고,<!-- HTML_TAG_START --><span class="katex"><span class="katex-mathml"><math xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mi>k</mi><mo>−</mo><mn>1</mn></mrow><annotation encoding="application/x-tex">k-1</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:0.03148em;">k</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:0.6444em;"></span><span class="mord">1</span></span></span></span><!-- HTML_TAG_END --> 토큰까지 사용해 대략적인 우도 값을 추정합니다.</p> <p data-svelte-h="svelte-p30g18">모델의 시퀀스에 대한 펄플렉서티를 계산할 때, | |
| 수월하지만 차선책은 시퀀스를 청크로 쪼개고 분해된 각 부분의 로그 우도 값을 독립적으로 합산하는 것입니다.</p> <img width="600" alt="Suboptimal PPL not taking advantage of full available context" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_chunked.gif"> <p data-svelte-h="svelte-1b7o98p">이 방법은 각 부분의 펄플렉서티를 한 번의 포워드 패스로 계산할 수 있어 빠르지만 일반적으로 더 높은(더 나쁜) PPL을 산출합니다. | |
| 왜냐하면 대부분의 예측 단계에서 모델의 컨텍스트가 적기 때문입니다.</p> <p data-svelte-h="svelte-10v5nla">대신, 고정 길이 모델의 PPL은 슬라이딩 윈도우 전략으로 평가해야 합니다. | |
| 이 전략에는 컨텍스트 윈도우을 반복적으로 슬라이딩해 모델이 각 예측을 수행할 때 더 많은 컨텍스트를 갖도록 하는 작업이 포함됩니다.</p> <img width="600" alt="Sliding window PPL taking advantage of all available context" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/ppl_sliding.gif"> <p data-svelte-h="svelte-1bx3afe">이는 시퀀스 확률의 실제 분해에 더 가까운 근사치이며 일반적으로 더 유리한 점수를 산출합니다. | |
| 단점은 말뭉치의 각 토큰에 대해 별도의 포워드 패스가 필요하다는 것입니다. | |
| 현실적으로 좋은 절충안은 한 번에 한 토큰씩 슬라이딩하는 것이 아니라 더 큰 간격으로 컨텍스트를 이동하는 스트라이드가 적용된 슬라이딩 윈도우을 사용하는 것입니다. | |
| 이렇게 하면 계산을 훨씬 더 빠르게 진행하면서도 모델에 각 단계에서 예측을 수행할 수 있는 긴 컨텍스트를 제공할 수 있습니다.</p> <h2 class="relative group"><a id="example-calculating-perplexity-with-gpt2-in-transformers" 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="#example-calculating-perplexity-with-gpt2-in-transformers"><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>예제: 🤗 Transformers에서 GPT-2로 펄플렉서티(perplexity) 계산하기</span></h2> <p data-svelte-h="svelte-1scr742">이제 GPT-2로 위의 과정을 시연해 보겠습니다.</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 --><span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> GPT2LMHeadModel, GPT2TokenizerFast | |
| device = <span class="hljs-string">"cuda"</span> | |
| model_id = <span class="hljs-string">"openai-community/gpt2-large"</span> | |
| model = GPT2LMHeadModel.from_pretrained(model_id).to(device) | |
| tokenizer = GPT2TokenizerFast.from_pretrained(model_id)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-g326s0">WikiText-2 데이터 세트를 가져오고 몇 가지 슬라이딩 윈도우 전략을 사용해 펄플렉서티를 계산해보겠습니다. | |
| 이 데이터 세트는 크기가 작고 포워드 패스 한 번만 수행하기 때문에 전체 데이터 세트를 메모리에 가져오고 인코딩할 수 있습니다.</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 --><span class="hljs-keyword">from</span> datasets <span class="hljs-keyword">import</span> load_dataset | |
| test = load_dataset(<span class="hljs-string">"wikitext"</span>, <span class="hljs-string">"wikitext-2-raw-v1"</span>, split=<span class="hljs-string">"test"</span>) | |
| encodings = tokenizer(<span class="hljs-string">"\n\n"</span>.join(test[<span class="hljs-string">"text"</span>]), return_tensors=<span class="hljs-string">"pt"</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-aa9p9w">🤗 Transformers를 사용하면 모델의 <code>labels</code>로 <code>input_ids</code>를 전달해 각 토큰에 대한 평균 음의 우도 값을 손실로 반환할 수 있습니다. | |
| 하지만 슬라이딩 윈도우 방식을 사용하면 각 반복마다 모델에 전달하는 토큰이 겹칩니다. | |
| 컨텍스트로 처리하는 토큰에 대한 로그 우도 값이 손실에 포함되는 것을 원하지 않기 때문에 이러한 토큰의 <code>input_ids</code>를 <code>-100</code>으로 설정하여 무시할 수 있습니다.</p> <p data-svelte-h="svelte-bou0a7">다음은 스트라이드(stride)를 <code>512</code>로 사용한 예시입니다. | |
| 즉, 모델이 한 토큰의 조건부 우도 값을 계산할 때 컨텍스트에 최소한 512개의 토큰이 포함되어있다는 의미입니다 (해당 토큰 앞에 512개의 토큰이 있는 경우).</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 --><span class="hljs-keyword">import</span> torch | |
| <span class="hljs-keyword">from</span> tqdm <span class="hljs-keyword">import</span> tqdm | |
| max_length = model.config.n_positions | |
| stride = <span class="hljs-number">512</span> | |
| seq_len = encodings.input_ids.size(<span class="hljs-number">1</span>) | |
| nlls = [] | |
| prev_end_loc = <span class="hljs-number">0</span> | |
| <span class="hljs-keyword">for</span> begin_loc <span class="hljs-keyword">in</span> tqdm(<span class="hljs-built_in">range</span>(<span class="hljs-number">0</span>, seq_len, stride)): | |
| end_loc = <span class="hljs-built_in">min</span>(begin_loc + max_length, seq_len) | |
| trg_len = end_loc - prev_end_loc <span class="hljs-comment"># 마지막 루프의 스트라이드 값과 다를 수 있음</span> | |
| input_ids = encodings.input_ids[:, begin_loc:end_loc].to(device) | |
| target_ids = input_ids.clone() | |
| target_ids[:, :-trg_len] = -<span class="hljs-number">100</span> | |
| <span class="hljs-keyword">with</span> torch.no_grad(): | |
| outputs = model(input_ids, labels=target_ids) | |
| <span class="hljs-comment"># 손실은 모든 유효한 레이블에 대한 평균값을 구하는 교차 엔트로피(cross entropy)로 계산됩니다.</span> | |
| <span class="hljs-comment"># 나이브 베이지안 모델은 내부적으로 레이블을 왼쪽으로 1개씩 밀기 때문에, (타켓 - 1)개 만큼의 레이블에 대해 손실을 계산합니다.</span> | |
| neg_log_likelihood = outputs.loss | |
| nlls.append(neg_log_likelihood) | |
| prev_end_loc = end_loc | |
| <span class="hljs-keyword">if</span> end_loc == seq_len: | |
| <span class="hljs-keyword">break</span> | |
| ppl = torch.exp(torch.stack(nlls).mean())<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1ro3gyl">스트라이드를 최대 입력 길이와 동일하게 설정하면 위에서 설명한 차선책인 비슬라이딩 윈도우 전략과 동일합니다. | |
| 일반적으로 스트라이드가 작을수록 모델이 각 예측을 할 때 더 많은 컨텍스트를 볼 수 있게 되어 펄플렉서티 값이 좋아집니다.</p> <p data-svelte-h="svelte-1cfqojp">위의 계산을 토큰이 겹치지 않도록 <code>stride = 1024</code>로 설정하면 PPL은 <code>19.44</code>로 GPT-2 논문에서 보고된 <code>19.93</code>과 거의 동일합니다. | |
| <code>stride = 512</code>로 슬라이딩 윈도우 전략을 사용하면 PPL은 <code>16.45</code>로 떨어집니다. | |
| 이는 더 좋은 점수일 뿐만 아니라 시퀀스 확률의 실제 자동 회귀 분해에 더 가까운 방식으로 계산됩니다.</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/perplexity.md" target="_blank"><span data-svelte-h="svelte-1kd6by1"><</span> <span data-svelte-h="svelte-x0xyl0">></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, 45], | |
| data, | |
| form: null, | |
| error: null | |
| }); | |
| }); | |
| } | |
| </script> | |
Xet Storage Details
- Size:
- 38.1 kB
- Xet hash:
- cff8208593af0801b8c85ef8a11d71cefdabb76a94bdb55787d4fdc0067ea7bc
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.