transformers/pr_28250/zh/bertology.html

<meta charset="utf-8" /><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;基于BERT进行的相关研究（BERTology）&quot;,&quot;local&quot;:&quot;基于bert进行的相关研究bertology&quot;,&quot;sections&quot;:[],&quot;depth&quot;:1}">
		<link href="/docs/transformers/pr_28250/zh/_app/immutable/assets/0.e3b0c442.css" rel="modulepreload">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/entry/start.359aefb2.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/scheduler.9991993c.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/singletons.0d57c151.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/index.02cfeb18.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/paths.03bfa24f.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/entry/app.6c335e9f.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/index.ed60ef0f.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/nodes/0.2e1efaaf.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/each.e59479a4.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/nodes/6.521b236d.js">
		<link rel="modulepreload" href="/docs/transformers/pr_28250/zh/_app/immutable/chunks/EditOnGithub.ba269039.js"><!-- HEAD_svelte-u9bgzb_START --><meta name="hf:doc:metadata" content="{&quot;title&quot;:&quot;基于BERT进行的相关研究（BERTology）&quot;,&quot;local&quot;:&quot;基于bert进行的相关研究bertology&quot;,&quot;sections&quot;:[],&quot;depth&quot;:1}"><!-- HEAD_svelte-u9bgzb_END -->      <p></p>   <h1 class="relative group"><a id="基于bert进行的相关研究bertology" 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="#基于bert进行的相关研究bertology"><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>基于BERT进行的相关研究（BERTology）</span></h1> <p data-svelte-h="svelte-122li14">当前，一个新兴的研究领域正致力于探索大规模 transformer 模型（如BERT）的内部工作机制，一些人称之为“BERTology”。以下是这个领域的一些典型示例：</p> <ul data-svelte-h="svelte-1e34lw"><li>BERT Rediscovers the Classical NLP Pipeline by Ian Tenney, Dipanjan Das, Ellie Pavlick:
<a href="https://arxiv.org/abs/1905.05950" rel="nofollow">https://arxiv.org/abs/1905.05950</a></li> <li>Are Sixteen Heads Really Better than One? by Paul Michel, Omer Levy, Graham Neubig: <a href="https://arxiv.org/abs/1905.10650" rel="nofollow">https://arxiv.org/abs/1905.10650</a></li> <li>What Does BERT Look At? An Analysis of BERT’s Attention by Kevin Clark, Urvashi Khandelwal, Omer Levy, Christopher D.
Manning: <a href="https://arxiv.org/abs/1906.04341" rel="nofollow">https://arxiv.org/abs/1906.04341</a></li> <li>CAT-probing: A Metric-based Approach to Interpret How Pre-trained Models for Programming Language Attend Code Structure: <a href="https://arxiv.org/abs/2210.04633" rel="nofollow">https://arxiv.org/abs/2210.04633</a></li></ul> <p data-svelte-h="svelte-kha6wh">为了助力这一新兴领域的发展，我们在BERT/GPT/GPT-2模型中增加了一些附加功能，方便人们访问其内部表示，这些功能主要借鉴了Paul Michel的杰出工作(<a href="https://arxiv.org/abs/1905.10650)%EF%BC%9A" rel="nofollow">https://arxiv.org/abs/1905.10650)：</a></p> <ul data-svelte-h="svelte-1dvlo65"><li>访问BERT/GPT/GPT-2的所有隐藏状态，</li> <li>访问BERT/GPT/GPT-2每个注意力头的所有注意力权重，</li> <li>检索注意力头的输出值和梯度，以便计算头的重要性得分并对头进行剪枝，详情可见论文：<a href="https://arxiv.org/abs/1905.10650%E3%80%82" rel="nofollow">https://arxiv.org/abs/1905.10650。</a></li></ul> <p data-svelte-h="svelte-j2vorw">为了帮助您理解和使用这些功能，我们添加了一个具体的示例脚本：<a href="https://github.com/huggingface/transformers/tree/main/examples/research_projects/bertology/run_bertology.py" rel="nofollow">bertology.py</a>，该脚本可以对一个在 GLUE 数据集上预训练的模型进行信息提取与剪枝。</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/zh/bertology.md" target="_blank"><span data-svelte-h="svelte-1kd6by1">&lt;</span> <span data-svelte-h="svelte-x0xyl0">&gt;</span> <span data-svelte-h="svelte-1dajgef"><span class="underline ml-1.5">Update</span> on GitHub</span></a>  <p></p> 
			
			<script>
				{
					__sveltekit_qheazb = {
						assets: "/docs/transformers/pr_28250/zh",
						base: "/docs/transformers/pr_28250/zh",
						env: {}
					};

					const element = document.currentScript.parentElement;

					const data = [null,null];

					Promise.all([
						import("/docs/transformers/pr_28250/zh/_app/immutable/entry/start.359aefb2.js"),
						import("/docs/transformers/pr_28250/zh/_app/immutable/entry/app.6c335e9f.js")
					]).then(([kit, app]) => {
						kit.start(app, element, {
							node_ids: [0, 6],
							data,
							form: null,
							error: null
						});
					});
				}
			</script>