Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / course /pr_1054 /id /chapter3 /3.html
rtrm's picture
rtrm
download
raw
70.2 kB
<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Fine-tuning Model dengan Trainer API&quot;,&quot;local&quot;:&quot;fine-tuning-a-model-with-the-trainer-api&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Pelatihan&quot;,&quot;local&quot;:&quot;training&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Evaluasi&quot;,&quot;local&quot;:&quot;evaluation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Fitur Pelatihan Lanjutan&quot;,&quot;local&quot;:&quot;advanced-training-features&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Kuis Bagian&quot;,&quot;local&quot;:&quot;section-quiz&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;1. Apa tujuan dari parameter <code> processing_class </code> dalam Trainer?&quot;,&quot;local&quot;:&quot;1-apa-tujuan-dari-parameter-code-processingclass-code-dalam-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;2. Parameter TrainingArguments mana yang mengontrol seberapa sering evaluasi dilakukan saat pelatihan?&quot;,&quot;local&quot;:&quot;2-parameter-trainingarguments-mana-yang-mengontrol-seberapa-sering-evaluasi-dilakukan-saat-pelatihan&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;3. Apa yang diaktifkan oleh <code> fp16=True </code> dalam TrainingArguments?&quot;,&quot;local&quot;:&quot;3-apa-yang-diaktifkan-oleh-code-fp16true-code-dalam-trainingarguments&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;4. Apa peran fungsi <code> compute_metrics </code> dalam Trainer?&quot;,&quot;local&quot;:&quot;4-apa-peran-fungsi-code-computemetrics-code-dalam-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;5. Apa yang terjadi jika Anda tidak menyediakan <code> eval_dataset </code> ke Trainer?&quot;,&quot;local&quot;:&quot;5-apa-yang-terjadi-jika-anda-tidak-menyediakan-code-evaldataset-code-ke-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;6. Apa itu gradient accumulation dan bagaimana cara mengaktifkannya?&quot;,&quot;local&quot;:&quot;6-apa-itu-gradient-accumulation-dan-bagaimana-cara-mengaktifkannya&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}">
 <link href="/docs/course/pr_1054/id/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/entry/start.4f92af03.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/scheduler.36a0863c.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/singletons.7dc7b9a4.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/index.733708bb.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/paths.cf097d06.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/entry/app.19cef1b6.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/index.156fee99.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/nodes/0.1203e4a0.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/each.e59479a4.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/nodes/25.d7d00efc.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/Tip.8a648467.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/Youtube.a5d6d567.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/CodeBlock.4cf998e6.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/CourseFloatingBanner.16bb8bff.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/Question.aeeaa180.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/stores.8f4efe8a.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/FrameworkSwitchCourse.d8719b67.js">
 <link rel="modulepreload" href="/docs/course/pr_1054/id/_app/immutable/chunks/getInferenceSnippets.472bc46d.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;Fine-tuning Model dengan Trainer API&quot;,&quot;local&quot;:&quot;fine-tuning-a-model-with-the-trainer-api&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;Pelatihan&quot;,&quot;local&quot;:&quot;training&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Evaluasi&quot;,&quot;local&quot;:&quot;evaluation&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Fitur Pelatihan Lanjutan&quot;,&quot;local&quot;:&quot;advanced-training-features&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;Kuis Bagian&quot;,&quot;local&quot;:&quot;section-quiz&quot;,&quot;sections&quot;:[{&quot;title&quot;:&quot;1. Apa tujuan dari parameter <code> processing_class </code> dalam Trainer?&quot;,&quot;local&quot;:&quot;1-apa-tujuan-dari-parameter-code-processingclass-code-dalam-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;2. Parameter TrainingArguments mana yang mengontrol seberapa sering evaluasi dilakukan saat pelatihan?&quot;,&quot;local&quot;:&quot;2-parameter-trainingarguments-mana-yang-mengontrol-seberapa-sering-evaluasi-dilakukan-saat-pelatihan&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;3. Apa yang diaktifkan oleh <code> fp16=True </code> dalam TrainingArguments?&quot;,&quot;local&quot;:&quot;3-apa-yang-diaktifkan-oleh-code-fp16true-code-dalam-trainingarguments&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;4. Apa peran fungsi <code> compute_metrics </code> dalam Trainer?&quot;,&quot;local&quot;:&quot;4-apa-peran-fungsi-code-computemetrics-code-dalam-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;5. Apa yang terjadi jika Anda tidak menyediakan <code> eval_dataset </code> ke Trainer?&quot;,&quot;local&quot;:&quot;5-apa-yang-terjadi-jika-anda-tidak-menyediakan-code-evaldataset-code-ke-trainer&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3},{&quot;title&quot;:&quot;6. Apa itu gradient accumulation dan bagaimana cara mengaktifkannya?&quot;,&quot;local&quot;:&quot;6-apa-itu-gradient-accumulation-dan-bagaimana-cara-mengaktifkannya&quot;,&quot;sections&quot;:[],&quot;depth&quot;:3}],&quot;depth&quot;:2}],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END --> <p></p> <div class="bg-white leading-none border border-gray-100 rounded-lg flex p-0.5 w-56 text-sm mb-4"><a class="flex justify-center flex-1 py-1.5 px-2.5 focus:outline-none !no-underline rounded-l bg-red-50 dark:bg-transparent text-red-600" href="?fw=pt"><svg class="mr-1.5" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><defs><clipPath id="a"><rect x="3.05" y="0.5" width="25.73" height="31" fill="none"></rect></clipPath></defs><g clip-path="url(#a)"><path d="M24.94,9.51a12.81,12.81,0,0,1,0,18.16,12.68,12.68,0,0,1-18,0,12.81,12.81,0,0,1,0-18.16l9-9V5l-.84.83-6,6a9.58,9.58,0,1,0,13.55,0ZM20.44,9a1.68,1.68,0,1,1,1.67-1.67A1.68,1.68,0,0,1,20.44,9Z" fill="#ee4c2c"></path></g></svg> Pytorch </a><a class="flex justify-center flex-1 py-1.5 px-2.5 focus:outline-none !no-underline rounded-r text-gray-500 filter grayscale" href="?fw=tf"><svg class="mr-1.5" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" aria-hidden="true" focusable="false" role="img" width="0.94em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 256 274"><path d="M145.726 42.065v42.07l72.861 42.07v-42.07l-72.86-42.07zM0 84.135v42.07l36.43 21.03V105.17L0 84.135zm109.291 21.035l-36.43 21.034v126.2l36.43 21.035v-84.135l36.435 21.035v-42.07l-36.435-21.034V105.17z" fill="#E55B2D"></path><path d="M145.726 42.065L36.43 105.17v42.065l72.861-42.065v42.065l36.435-21.03v-84.14zM255.022 63.1l-36.435 21.035v42.07l36.435-21.035V63.1zm-72.865 84.135l-36.43 21.035v42.07l36.43-21.036v-42.07zm-36.43 63.104l-36.436-21.035v84.135l36.435-21.035V210.34z" fill="#ED8E24"></path><path d="M145.726 0L0 84.135l36.43 21.035l109.296-63.105l72.861 42.07L255.022 63.1L145.726 0zm0 126.204l-36.435 21.03l36.435 21.036l36.43-21.035l-36.43-21.03z" fill="#F8BF3C"></path></svg> TensorFlow </a></div> <h1 class="relative group"><a id="fine-tuning-a-model-with-the-trainer-api" 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="#fine-tuning-a-model-with-the-trainer-api"><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>Fine-tuning Model dengan Trainer API</span></h1> <div class="flex space-x-1 absolute z-10 right-0 top-0"><a href="https://discuss.huggingface.co/t/chapter-3-questions" target="_blank"><img alt="Ask a Question" class="!m-0" src="https://img.shields.io/badge/Ask%20a%20question-ffcb4c.svg?logo=data:image/svg+xml;base64,PHN2ZyB4bWxucz0iaHR0cDovL3d3dy53My5vcmcvMjAwMC9zdmciIHZpZXdCb3g9IjAgLTEgMTA0IDEwNiI+PGRlZnM+PHN0eWxlPi5jbHMtMXtmaWxsOiMyMzFmMjA7fS5jbHMtMntmaWxsOiNmZmY5YWU7fS5jbHMtM3tmaWxsOiMwMGFlZWY7fS5jbHMtNHtmaWxsOiMwMGE5NGY7fS5jbHMtNXtmaWxsOiNmMTVkMjI7fS5jbHMtNntmaWxsOiNlMzFiMjM7fTwvc3R5bGU+PC9kZWZzPjx0aXRsZT5EaXNjb3Vyc2VfbG9nbzwvdGl0bGU+PGcgaWQ9IkxheWVyXzIiPjxnIGlkPSJMYXllcl8zIj48cGF0aCBjbGFzcz0iY2xzLTEiIGQ9Ik01MS44NywwQzIzLjcxLDAsMCwyMi44MywwLDUxYzAsLjkxLDAsNTIuODEsMCw1Mi44MWw1MS44Ni0uMDVjMjguMTYsMCw1MS0yMy43MSw1MS01MS44N1M4MCwwLDUxLjg3LDBaIi8+PHBhdGggY2xhc3M9ImNscy0yIiBkPSJNNTIuMzcsMTkuNzRBMzEuNjIsMzEuNjIsMCwwLDAsMjQuNTgsNjYuNDFsLTUuNzIsMTguNEwzOS40LDgwLjE3YTMxLjYxLDMxLjYxLDAsMSwwLDEzLTYwLjQzWiIvPjxwYXRoIGNsYXNzPSJjbHMtMyIgZD0iTTc3LjQ1LDMyLjEyYTMxLjYsMzEuNiwwLDAsMS0zOC4wNSw0OEwxOC44Niw4NC44MmwyMC45MS0yLjQ3QTMxLjYsMzEuNiwwLDAsMCw3Ny40NSwzMi4xMloiLz48cGF0aCBjbGFzcz0iY2xzLTQiIGQ9Ik03MS42MywyNi4yOUEzMS42LDMxLjYsMCwwLDEsMzguOCw3OEwxOC44Niw4NC44MiwzOS40LDgwLjE3QTMxLjYsMzEuNiwwLDAsMCw3MS42MywyNi4yOVoiLz48cGF0aCBjbGFzcz0iY2xzLTUiIGQ9Ik0yNi40Nyw2Ny4xMWEzMS42MSwzMS42MSwwLDAsMSw1MS0zNUEzMS42MSwzMS42MSwwLDAsMCwyNC41OCw2Ni40MWwtNS43MiwxOC40WiIvPjxwYXRoIGNsYXNzPSJjbHMtNiIgZD0iTTI0LjU4LDY2LjQxQTMxLjYxLDMxLjYxLDAsMCwxLDcxLjYzLDI2LjI5YTMxLjYxLDMxLjYxLDAsMCwwLTQ5LDM5LjYzbC0zLjc2LDE4LjlaIi8+PC9nPjwvZz48L3N2Zz4="></a> <a href="https://colab.research.google.com/github/huggingface/notebooks/blob/master/course/en/chapter3/section3.ipynb" target="_blank"><img alt="Open In Colab" class="!m-0" src="https://colab.research.google.com/assets/colab-badge.svg"></a> <a href="https://studiolab.sagemaker.aws/import/github/huggingface/notebooks/blob/master/course/en/chapter3/section3.ipynb" target="_blank"><img alt="Open In Studio Lab" class="!m-0" src="https://studiolab.sagemaker.aws/studiolab.svg"></a></div> <iframe class="w-full xl:w-4/6 h-80" src="https://www.youtube-nocookie.com/embed/nvBXf7s7vTI" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> <p data-svelte-h="svelte-1d6rr4m">🤗 Transformers menyediakan kelas <code>Trainer</code> untuk membantu Anda melakukan fine-tuning model pralatih apa pun pada dataset Anda menggunakan praktik terbaik modern. Setelah semua praproses data selesai di bagian sebelumnya, hanya ada beberapa langkah lagi untuk mendefinisikan <code>Trainer</code>. Bagian yang paling menantang kemungkinan besar adalah menyiapkan lingkungan untuk menjalankan <code>Trainer.train()</code>, karena akan berjalan sangat lambat di CPU. Jika Anda tidak memiliki GPU, Anda bisa menggunakan GPU atau TPU gratis di <a href="https://colab.research.google.com/" rel="nofollow">Google Colab</a>.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-h22p9w">📚 <strong>Sumber Latihan</strong>: Sebelum mulai melatih, kenali terlebih dahulu <a href="https://huggingface.co/docs/transformers/main/en/training" rel="nofollow">panduan pelatihan 🤗 Transformers</a> dan contoh praktiknya di <a href="https://huggingface.co/learn/cookbook/en/fine_tuning_code_llm_on_single_gpu" rel="nofollow">cookbook fine-tuning</a>.</p></div> <p data-svelte-h="svelte-a23cjl">Contoh kode di bawah mengasumsikan bahwa Anda telah menjalankan contoh di bagian sebelumnya. Berikut ringkasan singkat apa yang Anda perlukan:</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
<span class="hljs-keyword">from</span> transformers <span class="hljs-keyword">import</span> AutoTokenizer, DataCollatorWithPadding
raw_datasets = load_dataset(<span class="hljs-string">&quot;glue&quot;</span>, <span class="hljs-string">&quot;mrpc&quot;</span>)
checkpoint = <span class="hljs-string">&quot;bert-base-uncased&quot;</span>
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
<span class="hljs-keyword">def</span> <span class="hljs-title function_">tokenize_function</span>(<span class="hljs-params">example</span>):
 <span class="hljs-keyword">return</span> tokenizer(example[<span class="hljs-string">&quot;sentence1&quot;</span>], example[<span class="hljs-string">&quot;sentence2&quot;</span>], truncation=<span class="hljs-literal">True</span>)
tokenized_datasets = raw_datasets.<span class="hljs-built_in">map</span>(tokenize_function, batched=<span class="hljs-literal">True</span>)
data_collator = DataCollatorWithPadding(tokenizer=tokenizer)<!-- HTML_TAG_END --></pre></div> <h3 class="relative group"><a id="training" 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="#training"><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>Pelatihan</span></h3> <p data-svelte-h="svelte-194qie3">Langkah pertama sebelum kita dapat mendefinisikan <code>Trainer</code> adalah membuat kelas <code>TrainingArguments</code> yang akan menyimpan semua hyperparameter yang digunakan <code>Trainer</code> untuk pelatihan dan evaluasi. Satu-satunya argumen wajib adalah direktori tempat model hasil pelatihan akan disimpan, termasuk checkpoint-checkpoint selama pelatihan. Sisanya bisa dibiarkan default, yang sudah cukup baik untuk fine-tuning dasar.</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> TrainingArguments
training_args = TrainingArguments(<span class="hljs-string">&quot;test-trainer&quot;</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1gxzrir">Jika Anda ingin model Anda secara otomatis diunggah ke Hub selama pelatihan, tambahkan <code>push_to_hub=True</code> dalam <code>TrainingArguments</code>. Kita akan mempelajari lebih lanjut tentang ini di <a href="/course/chapter4/3">Bab 4</a></p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1tgflm4">🚀 <strong>Konfigurasi Lanjutan</strong>: Untuk detail lengkap semua argumen pelatihan dan strategi optimasi, lihat <a href="https://huggingface.co/docs/transformers/main/en/main_classes/trainer#transformers.TrainingArguments" rel="nofollow">dokumentasi TrainingArguments</a> dan <a href="https://huggingface.co/learn/cookbook/en/fine_tuning_code_llm_on_single_gpu" rel="nofollow">cookbook konfigurasi pelatihan</a>.</p></div> <p data-svelte-h="svelte-1xyo4pf">Langkah kedua adalah mendefinisikan model. Seperti di <a href="/course/chapter2">bab sebelumnya</a>, kita akan menggunakan <code>AutoModelForSequenceClassification</code> dengan dua label:</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> AutoModelForSequenceClassification
model = AutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=<span class="hljs-number">2</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1hv7i3p">Anda akan melihat bahwa, tidak seperti di <a href="/course/chapter2">Bab 2</a>, Anda akan mendapatkan peringatan setelah menginstansiasi model pretrained ini. Hal ini terjadi karena BERT tidak dilatih sebelumnya untuk mengklasifikasikan pasangan kalimat, sehingga kepala (head) dari model pretrained telah dibuang dan diganti dengan kepala baru yang sesuai untuk tugas klasifikasi sekuens. Peringatan tersebut menunjukkan bahwa beberapa bobot tidak digunakan (yaitu yang terkait dengan kepala pretraining yang dihapus), dan beberapa lainnya diinisialisasi secara acak (yaitu yang digunakan untuk kepala baru). Peringatan ini diakhiri dengan anjuran untuk melatih model tersebut — dan itulah yang akan kita lakukan sekarang.</p> <p data-svelte-h="svelte-4d0wt7">Setelah kita memiliki model, kita bisa mendefinisikan <code>Trainer</code> dengan memberikan semua objek yang sudah kita buat sejauh ini — <code>model</code>, <code>training_args</code>, dataset pelatihan dan validasi, <code>data_collator</code>, serta <code>processing_class</code>. Parameter <code>processing_class</code> adalah fitur baru yang memberi tahu Trainer tokenizer mana yang digunakan:</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> Trainer
trainer = Trainer(
 model,
 training_args,
 train_dataset=tokenized_datasets[<span class="hljs-string">&quot;train&quot;</span>],
 eval_dataset=tokenized_datasets[<span class="hljs-string">&quot;validation&quot;</span>],
 data_collator=data_collator,
 processing_class=tokenizer,
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1a9iw5v">Jika Anda memberikan tokenizer sebagai <code>processing_class</code>, <code>Trainer</code> akan secara otomatis menggunakan <code>DataCollatorWithPadding</code>. Baris <code>data_collator=...</code> bisa diabaikan, tapi kami sertakan di sini untuk menunjukkan bagian penting dari pipeline ini.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-12xc9rd">📖 <strong>Pelajari Lebih Lanjut</strong>: Untuk detail lengkap tentang kelas Trainer dan parameternya, kunjungi <a href="https://huggingface.co/docs/transformers/main/en/main_classes/trainer" rel="nofollow">dokumentasi Trainer API</a> dan contoh lanjutan di <a href="https://huggingface.co/learn/cookbook/en/fine_tuning_code_llm_on_single_gpu" rel="nofollow">cookbook pelatihan</a>.</p></div> <p data-svelte-h="svelte-1acpzfq">Untuk melakukan fine-tuning, kita cukup memanggil metode <code>train()</code>:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START -->trainer.train()<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1pp0n0z">Ini akan memulai pelatihan dan melaporkan loss setiap 500 langkah. Namun, ini tidak akan menunjukkan seberapa baik model Anda karena:</p> <ol data-svelte-h="svelte-1xy13n"><li>Kita belum memberi tahu <code>Trainer</code> untuk melakukan evaluasi selama pelatihan dengan mengatur <code>eval_strategy</code> di <code>TrainingArguments</code> ke salah satu dari <code>&quot;steps&quot;</code> (melakukan evaluasi setiap <code>eval_steps</code>) atau <code>&quot;epoch&quot;</code> (melakukan evaluasi di akhir setiap epoch).</li> <li>Kita juga belum memberikan fungsi <code>compute_metrics()</code> ke <code>Trainer</code> untuk menghitung metrik selama evaluasi tersebut (jika tidak, evaluasi hanya akan mencetak nilai loss, yang bukan angka yang mudah diinterpretasikan).</li></ol> <h3 class="relative group"><a id="evaluation" 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="#evaluation"><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>Evaluasi</span></h3> <p data-svelte-h="svelte-1p1t1o2">Mari kita lihat bagaimana cara membangun fungsi <code>compute_metrics()</code> yang berguna dan menggunakannya saat kita melatih model berikutnya. Fungsi ini harus menerima objek <code>EvalPrediction</code> (yang merupakan tuple bernama dengan field <code>predictions</code> dan <code>label_ids</code>) dan mengembalikan dictionary yang memetakan string ke angka desimal (string adalah nama metrik yang dikembalikan, dan angka desimal adalah nilai metrik tersebut). Untuk mendapatkan prediksi dari model kita, kita dapat menggunakan perintah <code>Trainer.predict()</code>:</p> <div class="code-block relative "><div class="absolute top-2.5 right-4"><button class="inline-flex items-center relative text-sm focus:text-green-500 cursor-pointer focus:outline-none transition duration-200 ease-in-out opacity-0 mx-0.5 text-gray-600 " title="code excerpt" type="button"><svg class="" xmlns="http://www.w3.org/2000/svg" aria-hidden="true" fill="currentColor" focusable="false" role="img" width="1em" height="1em" preserveAspectRatio="xMidYMid meet" viewBox="0 0 32 32"><path d="M28,10V28H10V10H28m0-2H10a2,2,0,0,0-2,2V28a2,2,0,0,0,2,2H28a2,2,0,0,0,2-2V10a2,2,0,0,0-2-2Z" transform="translate(0)"></path><path d="M4,18H2V4A2,2,0,0,1,4,2H18V4H4Z" transform="translate(0)"></path><rect fill="none" width="32" height="32"></rect></svg> <div class="absolute pointer-events-none transition-opacity bg-black text-white py-1 px-2 leading-tight rounded font-normal shadow left-1/2 top-full transform -translate-x-1/2 translate-y-2 opacity-0"><div class="absolute bottom-full left-1/2 transform -translate-x-1/2 w-0 h-0 border-black border-4 border-t-0" style="border-left-color: transparent; border-right-color: transparent; "></div> Copied</div></button></div> <pre class=""><!-- HTML_TAG_START -->predictions = trainer.predict(tokenized_datasets[<span class="hljs-string">&quot;validation&quot;</span>])
<span class="hljs-built_in">print</span>(predictions.predictions.shape, predictions.label_ids.shape)<!-- HTML_TAG_END --></pre></div> <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-number">408</span>, <span class="hljs-number">2</span>) (<span class="hljs-number">408</span>,)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1elf410">Output dari metode <code>predict()</code> adalah tuple bernama lain yang memiliki tiga field: <code>predictions</code>, <code>label_ids</code>, dan <code>metrics</code>. Field <code>metrics</code> hanya akan berisi loss pada dataset yang diberikan, serta metrik waktu (berapa lama waktu yang dibutuhkan untuk melakukan prediksi, secara total dan rata-rata). Setelah kita melengkapi fungsi <code>compute_metrics()</code> dan memberikannya ke <code>Trainer</code>, field tersebut juga akan berisi metrik yang dikembalikan oleh <code>compute_metrics()</code>.</p> <p data-svelte-h="svelte-13batns">Seperti yang bisa kamu lihat, <code>predictions</code> adalah array dua dimensi dengan bentuk 408 x 2 (408 adalah jumlah elemen dalam dataset yang kita gunakan). Ini adalah nilai logit untuk setiap elemen dari dataset yang kita berikan ke <code>predict()</code> (seperti yang telah kamu lihat di <a href="/course/chapter2">bab sebelumnya</a>, semua model Transformer mengembalikan logit). Untuk mengubahnya menjadi prediksi yang dapat kita bandingkan dengan label kita, kita perlu mengambil indeks dengan nilai maksimum pada sumbu kedua:</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> numpy <span class="hljs-keyword">as</span> np
preds = np.argmax(predictions.predictions, axis=-<span class="hljs-number">1</span>)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1qyl80t">Sekarang kita bisa membandingkan <code>preds</code> tersebut dengan label yang sebenarnya. Untuk membangun fungsi <code>compute_metrics()</code>, kita akan menggunakan metrik dari pustaka 🤗 <a href="https://github.com/huggingface/evaluate/" rel="nofollow">Evaluate</a>. Kita bisa memuat metrik yang digunakan untuk dataset MRPC semudah kita memuat dataset-nya, kali ini dengan fungsi <code>evaluate.load()</code>. Objek yang dikembalikan memiliki metode <code>compute()</code> yang bisa kita gunakan untuk menghitung metrik evaluasi:</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> evaluate
metric = evaluate.load(<span class="hljs-string">&quot;glue&quot;</span>, <span class="hljs-string">&quot;mrpc&quot;</span>)
metric.compute(predictions=preds, references=predictions.label_ids)<!-- HTML_TAG_END --></pre></div> <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-string">&#x27;accuracy&#x27;</span>: <span class="hljs-number">0.8578431372549019</span>, <span class="hljs-string">&#x27;f1&#x27;</span>: <span class="hljs-number">0.8996539792387542</span>}<!-- HTML_TAG_END --></pre></div> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-dwktpq">Pelajari metrik evaluasi lebih lanjut di <a href="https://huggingface.co/docs/evaluate/" rel="nofollow">dokumentasi 🤗 Evaluate</a>.</p></div> <p data-svelte-h="svelte-10xjo82">Hasil yang Anda peroleh bisa berbeda-beda, karena inisialisasi acak pada kepala model dapat memengaruhi metrik yang dicapai. Di sini, kita dapat melihat bahwa model kita memiliki akurasi sebesar 85,78% pada set validasi dan skor F1 sebesar 89,97. Kedua metrik tersebut digunakan untuk mengevaluasi hasil pada dataset MRPC dalam benchmark GLUE. Tabel pada <a href="https://arxiv.org/pdf/1810.04805.pdf" rel="nofollow">makalah BERT</a> melaporkan skor F1 sebesar 88,9 untuk model dasar. Model tersebut adalah model <code>uncased</code>, sementara kita saat ini menggunakan model <code>cased</code>, yang menjelaskan hasil yang lebih baik.</p> <p data-svelte-h="svelte-4z06rr">Menggabungkan semuanya, kita mendapatkan fungsi <code>compute_metrics()</code> kita:</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">def</span> <span class="hljs-title function_">compute_metrics</span>(<span class="hljs-params">eval_preds</span>):
 metric = evaluate.load(<span class="hljs-string">&quot;glue&quot;</span>, <span class="hljs-string">&quot;mrpc&quot;</span>)
 logits, labels = eval_preds
 predictions = np.argmax(logits, axis=-<span class="hljs-number">1</span>)
 <span class="hljs-keyword">return</span> metric.compute(predictions=predictions, references=labels)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-m088q">Dan untuk melihatnya digunakan secara langsung dalam melaporkan metrik di akhir setiap epoch, berikut adalah cara kita mendefinisikan <code>Trainer</code> baru dengan fungsi <code>compute_metrics()</code> ini:</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 -->training_args = TrainingArguments(<span class="hljs-string">&quot;test-trainer&quot;</span>, eval_strategy=<span class="hljs-string">&quot;epoch&quot;</span>)
model = AutoModelForSequenceClassification.from_pretrained(checkpoint, num_labels=<span class="hljs-number">2</span>)
trainer = Trainer(
 model,
 training_args,
 train_dataset=tokenized_datasets[<span class="hljs-string">&quot;train&quot;</span>],
 eval_dataset=tokenized_datasets[<span class="hljs-string">&quot;validation&quot;</span>],
 data_collator=data_collator,
 processing_class=tokenizer,
 compute_metrics=compute_metrics,
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1ju7nmh">Perlu dicatat bahwa kita membuat objek <code>TrainingArguments</code> baru dengan <code>eval_strategy</code> diatur ke <code>&quot;epoch&quot;</code> dan juga model baru — jika tidak, kita hanya akan melanjutkan pelatihan dari model yang sudah kita latih sebelumnya. Untuk memulai proses pelatihan baru, kita jalankan:</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 -->trainer.train()<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-1dbi3nz">Kali ini, proses pelatihan akan melaporkan <em>validation loss</em> dan metrik evaluasi di akhir setiap epoch, selain <em>training loss</em>. Sekali lagi, akurasi atau skor F1 yang Anda dapatkan mungkin sedikit berbeda karena inisialisasi acak pada kepala model, namun tetap akan berada di kisaran yang serupa.</p> <h3 class="relative group"><a id="advanced-training-features" 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="#advanced-training-features"><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>Fitur Pelatihan Lanjutan</span></h3> <p data-svelte-h="svelte-jebpr5"><code>Trainer</code> memiliki banyak fitur bawaan yang membuat praktik terbaik deep learning modern lebih mudah diakses:</p> <p data-svelte-h="svelte-46h5mv"><strong>Pelatihan dengan Presisi Campuran (Mixed Precision)</strong>: Gunakan <code>fp16=True</code> dalam training arguments untuk pelatihan yang lebih cepat dan penggunaan memori yang lebih rendah:</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 -->training_args = TrainingArguments(
 <span class="hljs-string">&quot;test-trainer&quot;</span>,
 eval_strategy=<span class="hljs-string">&quot;epoch&quot;</span>,
 fp16=<span class="hljs-literal">True</span>, <span class="hljs-comment"># Aktifkan pelatihan dengan presisi campuran</span>
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-15wws5i"><strong>Akumulasi Gradien (Gradient Accumulation)</strong>: Berguna untuk mendapatkan ukuran batch efektif yang lebih besar saat memori GPU terbatas:</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 -->training_args = TrainingArguments(
 <span class="hljs-string">&quot;test-trainer&quot;</span>,
 eval_strategy=<span class="hljs-string">&quot;epoch&quot;</span>,
 per_device_train_batch_size=<span class="hljs-number">4</span>,
 gradient_accumulation_steps=<span class="hljs-number">4</span>, <span class="hljs-comment"># Ukuran batch efektif = 4 * 4 = 16</span>
)<!-- HTML_TAG_END --></pre></div> <p data-svelte-h="svelte-kxznaz"><strong>Pengaturan Learning Rate</strong>: <code>Trainer</code> secara default menggunakan peluruhan linear (linear decay), tetapi Anda bisa menyesuaikannya:</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 -->training_args = TrainingArguments(
 <span class="hljs-string">&quot;test-trainer&quot;</span>,
 eval_strategy=<span class="hljs-string">&quot;epoch&quot;</span>,
 learning_rate=<span class="hljs-number">2e-5</span>,
 lr_scheduler_type=<span class="hljs-string">&quot;cosine&quot;</span>, <span class="hljs-comment"># Coba scheduler lain</span>
)<!-- HTML_TAG_END --></pre></div> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-pnyltu">🎯 <strong>Optimasi Performa</strong>: Untuk teknik pelatihan lanjutan seperti pelatihan terdistribusi, optimasi memori, dan optimasi spesifik perangkat keras, jelajahi <a href="https://huggingface.co/docs/transformers/main/en/performance" rel="nofollow">panduan performa 🤗 Transformers</a>.</p></div> <p data-svelte-h="svelte-tk7j0"><code>Trainer</code> langsung dapat digunakan di beberapa GPU atau TPU, dan menyediakan banyak opsi untuk pelatihan terdistribusi. Kita akan membahas semua fitur tersebut di Bab 10.</p> <p data-svelte-h="svelte-5r2pgf">Ini menutup pengenalan tentang fine-tuning menggunakan API <code>Trainer</code>. Contoh penerapan fine-tuning pada berbagai tugas NLP umum akan dijelaskan di <a href="/course/chapter7">Bab 7</a>, tetapi untuk sekarang mari kita lihat bagaimana melakukan hal yang sama menggunakan <em>training loop</em> PyTorch murni.</p> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1v6h1bi">📝 <strong>Lebih Banyak Contoh</strong>: Lihat koleksi lengkap <a href="https://huggingface.co/docs/transformers/main/en/notebooks" rel="nofollow">notebook 🤗 Transformers</a>.</p></div> <h2 class="relative group"><a id="section-quiz" 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="#section-quiz"><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>Kuis Bagian</span></h2> <p data-svelte-h="svelte-1v50w22">Uji pemahaman Anda tentang API <code>Trainer</code> dan konsep fine-tuning:</p> <h3 class="relative group"><a id="1-apa-tujuan-dari-parameter-code-processingclass-code-dalam-trainer" 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="#1-apa-tujuan-dari-parameter-code-processingclass-code-dalam-trainer"><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>1. Apa tujuan dari parameter &lt;code> processing_class &lt;/code> dalam Trainer?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->Ini menentukan arsitektur model yang digunakan.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->Ini memberi tahu Trainer tokenizer mana yang digunakan untuk memproses data.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->Ini menentukan ukuran batch untuk pelatihan.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->Ini mengontrol frekuensi evaluasi.<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <h3 class="relative group"><a id="2-parameter-trainingarguments-mana-yang-mengontrol-seberapa-sering-evaluasi-dilakukan-saat-pelatihan" 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="#2-parameter-trainingarguments-mana-yang-mengontrol-seberapa-sering-evaluasi-dilakukan-saat-pelatihan"><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>2. Parameter TrainingArguments mana yang mengontrol seberapa sering evaluasi dilakukan saat pelatihan?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->eval_frequency<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->eval_strategy<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->evaluation_steps<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->do_eval<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <h3 class="relative group"><a id="3-apa-yang-diaktifkan-oleh-code-fp16true-code-dalam-trainingarguments" 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="#3-apa-yang-diaktifkan-oleh-code-fp16true-code-dalam-trainingarguments"><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>3. Apa yang diaktifkan oleh &lt;code> fp16=True &lt;/code> dalam TrainingArguments?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->Presisi integer 16-bit untuk pelatihan yang lebih cepat.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->Pelatihan dengan presisi campuran menggunakan angka floating-point 16-bit untuk pelatihan lebih cepat dan penggunaan memori lebih rendah.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->Pelatihan selama tepat 16 epoch.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->Menggunakan 16 GPU untuk pelatihan terdistribusi.<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <h3 class="relative group"><a id="4-apa-peran-fungsi-code-computemetrics-code-dalam-trainer" 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="#4-apa-peran-fungsi-code-computemetrics-code-dalam-trainer"><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>4. Apa peran fungsi &lt;code> compute_metrics &lt;/code> dalam Trainer?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->Ia menghitung loss selama pelatihan.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->Ia mengubah logit menjadi prediksi dan menghitung metrik evaluasi seperti akurasi dan F1.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->Ia menentukan optimizer mana yang akan digunakan.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->Ia memproses data pelatihan.<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <h3 class="relative group"><a id="5-apa-yang-terjadi-jika-anda-tidak-menyediakan-code-evaldataset-code-ke-trainer" 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="#5-apa-yang-terjadi-jika-anda-tidak-menyediakan-code-evaldataset-code-ke-trainer"><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>5. Apa yang terjadi jika Anda tidak menyediakan &lt;code> eval_dataset &lt;/code> ke Trainer?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->Pelatihan akan gagal dengan error.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->Trainer akan secara otomatis membagi data pelatihan untuk evaluasi.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->Anda tidak akan mendapatkan metrik evaluasi selama pelatihan, tapi pelatihan tetap akan berjalan.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->Model akan menggunakan data pelatihan untuk evaluasi.<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <h3 class="relative group"><a id="6-apa-itu-gradient-accumulation-dan-bagaimana-cara-mengaktifkannya" 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="#6-apa-itu-gradient-accumulation-dan-bagaimana-cara-mengaktifkannya"><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>6. Apa itu gradient accumulation dan bagaimana cara mengaktifkannya?</span></h3> <div><form><label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="0"> <!-- HTML_TAG_START -->Ia menyimpan gradien ke disk, diaktifkan dengan save_gradients=True.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="1"> <!-- HTML_TAG_START -->Ia mengakumulasi gradien dari beberapa batch sebelum melakukan update, diaktifkan dengan gradient_accumulation_steps.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="2"> <!-- HTML_TAG_START -->Ia mempercepat perhitungan gradien, diaktifkan otomatis dengan fp16.<!-- HTML_TAG_END --></label> <label class="block"><input autocomplete="off" class="form-input -mt-1.5 mr-2" name="choice" type="checkbox" value="3"> <!-- HTML_TAG_START -->Ia mencegah overflow gradien, diaktifkan dengan gradient_clipping=True.<!-- HTML_TAG_END --></label> <div class="flex flex-row items-center mt-3"><button class="btn px-4 mr-4" type="submit" disabled>Submit</button> </div></form></div> <div class="course-tip bg-gradient-to-br dark:bg-gradient-to-r before:border-green-500 dark:before:border-green-800 from-green-50 dark:from-gray-900 to-white dark:to-gray-950 border border-green-50 text-green-700 dark:text-gray-400"><p data-svelte-h="svelte-1gdfk5g">💡 <strong>Inti Penting:</strong></p> <ul data-svelte-h="svelte-o7lpvg"><li>API <code>Trainer</code> menyediakan antarmuka tingkat tinggi yang menangani sebagian besar kompleksitas pelatihan</li> <li>Gunakan <code>processing_class</code> untuk menentukan tokenizer yang tepat dalam pemrosesan data</li> <li><code>TrainingArguments</code> mengontrol semua aspek pelatihan: learning rate, batch size, strategi evaluasi, dan optimasi</li> <li><code>compute_metrics</code> memungkinkan evaluasi dengan metrik khusus, bukan hanya loss pelatihan</li> <li>Fitur modern seperti presisi campuran (<code>fp16=True</code>) dan gradient accumulation bisa secara signifikan meningkatkan efisiensi pelatihan</li></ul></div> <a class="!text-gray-400 !no-underline text-sm flex items-center not-prose mt-4" href="https://github.com/huggingface/course/blob/main/chapters/id/chapter3/3.mdx" 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_ojy514 = {
 assets: "/docs/course/pr_1054/id",
 base: "/docs/course/pr_1054/id",
 env: {}
 };
 const element = document.currentScript.parentElement;
 const data = [null,null];
 Promise.all([
 import("/docs/course/pr_1054/id/_app/immutable/entry/start.4f92af03.js"),
 import("/docs/course/pr_1054/id/_app/immutable/entry/app.19cef1b6.js")
 ]).then(([kit, app]) => {
 kit.start(app, element, {
 node_ids: [0, 25],
 data,
 form: null,
 error: null
 });
 });
 }
 </script>

Xet Storage Details

Size:
70.2 kB
·
Xet hash:
4c9c5dc883a41a4ce861e2d70cb9057c42375fe1ea1d1155169e9f7e17c39d9e

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