{"id":477506,"date":"2023-08-09T09:15:57","date_gmt":"2023-08-09T09:15:57","guid":{"rendered":""},"modified":"2023-09-05T11:14:50","modified_gmt":"2023-09-05T11:14:50","slug":"hugging-face","status":"publish","type":"wiki","link":"https:\/\/oneproxy.pro\/cn\/wiki\/hugging-face\/","title":{"rendered":"\u62b1\u8138"},"content":{"rendered":"<p>Hugging Face \u662f\u4e00\u5bb6\u4e13\u6ce8\u4e8e\u81ea\u7136\u8bed\u8a00\u5904\u7406 (NLP) \u548c\u4eba\u5de5\u667a\u80fd (AI) \u7684\u5148\u950b\u516c\u53f8\u548c\u5f00\u6e90\u793e\u533a\u3002Hugging Face \u4ee5\u5176 Transformer \u6a21\u578b\u4ee5\u53ca\u76f8\u5173\u7684 PyTorch \u548c TensorFlow \u5e93\u800c\u95fb\u540d\uff0c\u5df2\u6210\u4e3a NLP \u7814\u7a76\u548c\u5f00\u53d1\u9886\u57df\u7684\u9886\u519b\u529b\u91cf\u3002<\/p>\n<h2>\u62e5\u62b1\u8138\u7684\u8d77\u6e90<\/h2>\n<p>Hugging Face, Inc. \u7531 Clement Delangue \u548c Julien Chaumond \u4e8e 2016 \u5e74\u5728\u7ebd\u7ea6\u5171\u540c\u521b\u7acb\u3002\u6700\u521d\uff0c\u8be5\u516c\u53f8\u4e13\u6ce8\u4e8e\u5f00\u53d1\u5177\u6709\u9c9c\u660e\u4e2a\u6027\u7684\u804a\u5929\u673a\u5668\u4eba\uff0c\u7c7b\u4f3c\u4e8e Siri \u548c Alexa\u3002\u7136\u800c\uff0c\u4ed6\u4eec\u7684\u91cd\u70b9\u5728 2018 \u5e74\u53d1\u751f\u4e86\u8f6c\u53d8\uff0c\u63a8\u51fa\u4e86\u4e00\u4e2a\u540d\u4e3a Transformers \u7684\u5f00\u6e90\u5e93\uff0c\u4ee5\u5e94\u5bf9\u65b0\u5174\u7684\u57fa\u4e8e Transformer \u7684\u6a21\u578b\u9886\u57df\uff0c\u8fd9\u4e9b\u6a21\u578b\u6b63\u5728\u5f7b\u5e95\u6539\u53d8 NLP \u9886\u57df\u3002<\/p>\n<h2>\u89e3\u5f00\u62e5\u62b1\u7684\u8138<\/h2>\n<p>Hugging Face \u7684\u6838\u5fc3\u76ee\u6807\u662f\u8ba9\u4eba\u5de5\u667a\u80fd\u666e\u53ca\uff0c\u5e76\u4e3a\u793e\u533a\u63d0\u4f9b\u5de5\u5177\uff0c\u8ba9\u6240\u6709\u4eba\u90fd\u80fd\u4f7f\u7528\u6700\u5148\u8fdb\u7684 NLP\u3002Hugging Face \u56e2\u961f\u7ef4\u62a4\u7740\u4e00\u4e2a\u540d\u4e3a Transformers \u7684\u5e93\uff0c\u8be5\u5e93\u63d0\u4f9b\u4e86\u6570\u5343\u4e2a\u9884\u5148\u8bad\u7ec3\u7684\u6a21\u578b\u6765\u6267\u884c\u6587\u672c\u4efb\u52a1\uff0c\u4f8b\u5982\u6587\u672c\u5206\u7c7b\u3001\u4fe1\u606f\u63d0\u53d6\u3001\u81ea\u52a8\u6458\u8981\u3001\u7ffb\u8bd1\u548c\u6587\u672c\u751f\u6210\u3002<\/p>\n<p>Hugging Face \u5e73\u53f0\u8fd8\u5305\u62ec\u534f\u4f5c\u8bad\u7ec3\u73af\u5883\u3001\u63a8\u7406 API \u548c\u6a21\u578b\u4e2d\u5fc3\u3002\u6a21\u578b\u4e2d\u5fc3\u5141\u8bb8\u7814\u7a76\u4eba\u5458\u548c\u5f00\u53d1\u4eba\u5458\u5171\u4eab\u6a21\u578b\u5e76\u8fdb\u884c\u534f\u4f5c\uff0c\u4ece\u800c\u6709\u52a9\u4e8e\u5e73\u53f0\u7684\u5f00\u653e\u6027\u3002<\/p>\n<h2>\u62e5\u62b1\u8138\u90e8\u7684\u5185\u90e8\u8fd0\u4f5c\u539f\u7406<\/h2>\n<p>Hugging Face \u5728 Transformer \u67b6\u6784\u7684\u57fa\u7840\u4e0a\u8fd0\u884c\uff0c\u5b83\u5229\u7528\u81ea\u6ce8\u610f\u529b\u673a\u5236\u6765\u7406\u89e3\u53e5\u5b50\u4e2d\u5355\u8bcd\u7684\u4e0a\u4e0b\u6587\u76f8\u5173\u6027\u3002 Transformer \u6a21\u578b\u662f\u5728\u5927\u578b\u6587\u672c\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u9884\u8bad\u7ec3\u7684\uff0c\u5e76\u4e14\u53ef\u4ee5\u9488\u5bf9\u7279\u5b9a\u4efb\u52a1\u8fdb\u884c\u5fae\u8c03\u3002<\/p>\n<p>\u5728\u540e\u7aef\uff0cTransformers \u5e93\u652f\u6301 PyTorch \u548c TensorFlow \u8fd9\u4e24\u79cd\u6700\u5e7f\u6cdb\u4f7f\u7528\u7684\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6\u3002\u8fd9\u4f7f\u5f97\u5b83\u975e\u5e38\u901a\u7528\uff0c\u5e76\u5141\u8bb8\u7528\u6237\u5728\u8fd9\u4e24\u4e2a\u6846\u67b6\u4e4b\u95f4\u65e0\u7f1d\u5207\u6362\u3002<\/p>\n<h2>\u62e5\u62b1\u8138\u7684\u4e3b\u8981\u7279\u70b9<\/h2>\n<ul>\n<li><strong>\u591a\u6837\u5316\u7684\u9884\u8bad\u7ec3\u6a21\u578b<\/strong>\uff1aHugging Face \u7684 Transformers \u5e93\u63d0\u4f9b\u4e86\u5927\u91cf\u9884\u8bad\u7ec3\u6a21\u578b\uff0c\u4f8b\u5982 BERT\u3001GPT-2\u3001T5 \u548c RoBERTa \u7b49\u3002<\/li>\n<li><strong>\u5e7f\u6cdb\u7684\u8bed\u8a00\u652f\u6301<\/strong>\uff1a\u6a21\u578b\u53ef\u4ee5\u5904\u7406\u591a\u79cd\u8bed\u8a00\uff0c\u7279\u5b9a\u6a21\u578b\u662f\u5728\u975e\u82f1\u8bed\u6570\u636e\u96c6\u4e0a\u8fdb\u884c\u8bad\u7ec3\u7684\u3002<\/li>\n<li><strong>\u5fae\u8c03\u80fd\u529b<\/strong>\uff1a\u6a21\u578b\u53ef\u4ee5\u8f7b\u677e\u5730\u9488\u5bf9\u7279\u5b9a\u4efb\u52a1\u8fdb\u884c\u5fae\u8c03\uff0c\u4ece\u800c\u5728\u5404\u79cd\u7528\u4f8b\u4e2d\u63d0\u4f9b\u591a\u529f\u80fd\u6027\u3002<\/li>\n<li><strong>\u793e\u533a\u9a71\u52a8<\/strong>\uff1aHugging Face \u5728\u5176\u793e\u533a\u4e2d\u84ec\u52c3\u53d1\u5c55\u3002\u5b83\u9f13\u52b1\u7528\u6237\u4e3a\u6a21\u578b\u505a\u51fa\u8d21\u732e\uff0c\u4ece\u800c\u63d0\u9ad8\u53ef\u7528\u6a21\u578b\u7684\u6574\u4f53\u8d28\u91cf\u548c\u591a\u6837\u6027\u3002<\/li>\n<\/ul>\n<h2>\u62e5\u62b1\u8138\u90e8\u6a21\u578b\u7684\u7c7b\u578b<\/h2>\n<p>\u4ee5\u4e0b\u662f Hugging Face \u53d8\u5f62\u91d1\u521a\u5e93\u4e2d\u4e00\u4e9b\u6700\u6d41\u884c\u7684\u53d8\u5f62\u91d1\u521a\u6a21\u578b\u7684\u5217\u8868\uff1a<\/p>\n<table>\n<thead>\n<tr>\n<th>\u578b\u53f7\u540d\u79f0<\/th>\n<th>\u63cf\u8ff0<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>\u4f2f\u7279<\/td>\n<td>Transformers \u7684\u53cc\u5411\u7f16\u7801\u5668\u8868\u793a\uff0c\u7528\u4e8e\u6839\u636e\u672a\u6807\u8bb0\u6587\u672c\u9884\u8bad\u7ec3\u6df1\u5ea6\u53cc\u5411\u8868\u793a<\/td>\n<\/tr>\n<tr>\n<td>GPT-2<\/td>\n<td>\u7528\u4e8e\u8bed\u8a00\u751f\u6210\u4efb\u52a1\u7684\u751f\u6210\u5f0f\u9884\u8bad\u7ec3 Transformer 2<\/td>\n<\/tr>\n<tr>\n<td>T5<\/td>\n<td>\u9002\u7528\u4e8e\u5404\u79cd NLP \u4efb\u52a1\u7684\u6587\u672c\u5230\u6587\u672c\u4f20\u8f93\u8f6c\u6362\u5668<\/td>\n<\/tr>\n<tr>\n<td>\u7f57\u4f2f\u5854<\/td>\n<td>\u7ecf\u8fc7\u7a33\u5065\u4f18\u5316\u7684 BERT \u7248\u672c\uff0c\u53ef\u83b7\u5f97\u66f4\u51c6\u786e\u7684\u7ed3\u679c<\/td>\n<\/tr>\n<tr>\n<td>\u84b8\u998f\u4f2f\u7279<\/td>\n<td>BERT \u7684\u7cbe\u70bc\u7248\u672c\uff0c\u66f4\u8f7b\u3001\u66f4\u5feb<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<h2>\u5229\u7528\u62e5\u62b1\u548c\u5e94\u5bf9\u6311\u6218<\/h2>\n<p>Hugging Face \u6a21\u578b\u53ef\u7528\u4e8e\u5e7f\u6cdb\u7684\u4efb\u52a1\uff0c\u4ece\u60c5\u611f\u5206\u6790\u548c\u6587\u672c\u5206\u7c7b\u5230\u673a\u5668\u7ffb\u8bd1\u548c\u6587\u672c\u6458\u8981\u3002\u7136\u800c\uff0c\u4e0e\u6240\u6709\u4eba\u5de5\u667a\u80fd\u6a21\u578b\u4e00\u6837\uff0c\u5b83\u4eec\u4e5f\u53ef\u80fd\u5e26\u6765\u6311\u6218\uff0c\u4f8b\u5982\u9700\u8981\u5927\u91cf\u6570\u636e\u8fdb\u884c\u8bad\u7ec3\u4ee5\u53ca\u6a21\u578b\u5b58\u5728\u504f\u5dee\u98ce\u9669\u3002 Hugging Face \u901a\u8fc7\u63d0\u4f9b\u8be6\u7ec6\u7684\u6a21\u578b\u5fae\u8c03\u6307\u5357\u548c\u591a\u79cd\u9884\u8bad\u7ec3\u6a21\u578b\u6765\u89e3\u51b3\u8fd9\u4e9b\u6311\u6218\u3002<\/p>\n<h2>\u4e0e\u7c7b\u4f3c\u5de5\u5177\u7684\u6bd4\u8f83<\/h2>\n<p>\u867d\u7136 Hugging Face \u662f\u5e7f\u6cdb\u6d41\u884c\u7684 NLP \u4efb\u52a1\u5e73\u53f0\uff0c\u4f46\u8fd8\u6709\u5176\u4ed6\u53ef\u7528\u5de5\u5177\uff0c\u4f8b\u5982 spaCy\u3001NLTK \u548c StanleyNLP\u3002\u7136\u800c\uff0cHugging Face \u7684\u4e0e\u4f17\u4e0d\u540c\u4e4b\u5904\u5728\u4e8e\u5176\u5e7f\u6cdb\u7684\u9884\u8bad\u7ec3\u6a21\u578b\u4ee5\u53ca\u4e0e PyTorch \u548c TensorFlow \u7684\u65e0\u7f1d\u96c6\u6210\u3002<\/p>\n<h2>\u62e5\u62b1\u8138\u7684\u672a\u6765<\/h2>\n<p>Hugging Face \u975e\u5e38\u91cd\u89c6\u793e\u533a\uff0c\u4e0d\u65ad\u7a81\u7834 NLP \u548c\u4eba\u5de5\u667a\u80fd\u7814\u7a76\u7684\u754c\u9650\u3002\u4ed6\u4eec\u6700\u8fd1\u7684\u91cd\u70b9\u662f GPT-4 \u7b49\u5927\u578b\u8bed\u8a00\u6a21\u578b\u9886\u57df\u4ee5\u53ca\u8fd9\u4e9b\u6a21\u578b\u5728\u901a\u7528\u4efb\u52a1\u4e2d\u53d1\u6325\u7684\u4f5c\u7528\u3002\u4ed6\u4eec\u8fd8\u6df1\u5165\u7814\u7a76\u8bbe\u5907\u4e0a\u548c\u9690\u79c1\u4fdd\u62a4\u673a\u5668\u5b66\u4e60\u7b49\u9886\u57df\u3002<\/p>\n<h2>\u4ee3\u7406\u670d\u52a1\u5668\u548c\u62e5\u62b1\u8138<\/h2>\n<p>\u4ee3\u7406\u670d\u52a1\u5668\u53ef\u4ee5\u4e0e Hugging Face \u7ed3\u5408\u4f7f\u7528\u6765\u6267\u884c\u7f51\u9875\u6293\u53d6\u7b49\u4efb\u52a1\uff0c\u5176\u4e2d IP \u8f6e\u6362\u5bf9\u4e8e\u533f\u540d\u81f3\u5173\u91cd\u8981\u3002\u4ee3\u7406\u670d\u52a1\u5668\u7684\u4f7f\u7528\u5141\u8bb8\u5f00\u53d1\u4eba\u5458\u4ece\u7f51\u7edc\u8bbf\u95ee\u548c\u68c0\u7d22\u6570\u636e\uff0c\u8fd9\u4e9b\u6570\u636e\u53ef\u4ee5\u8f93\u5165\u5230 Hugging Face \u6a21\u578b\u4e2d\u4ee5\u6267\u884c\u5404\u79cd NLP \u4efb\u52a1\u3002<\/p>\n<h2>\u76f8\u5173\u94fe\u63a5<\/h2>\n<ul>\n<li>\u62b1\u8138\u7f51\u7ad9\uff1a <a href=\"https:\/\/huggingface.co\/\" target=\"_new\" rel=\"noopener nofollow\">https:\/\/huggingface.co\/<\/a><\/li>\n<li>GitHub \u4e0a\u7684 Transformers \u5e93\uff1a <a href=\"https:\/\/github.com\/huggingface\/transformers\" target=\"_new\" rel=\"noopener nofollow\">https:\/\/github.com\/huggingface\/transformers<\/a><\/li>\n<li>\u62e5\u62b1\u8138\u90e8\u6a21\u578b\u4e2d\u5fc3\uff1a <a href=\"https:\/\/huggingface.co\/models\" target=\"_new\" rel=\"noopener nofollow\">https:\/\/huggingface.co\/models<\/a><\/li>\n<li>\u5b98\u65b9\u62b1\u8138\u8bfe\u7a0b\uff1a <a href=\"https:\/\/huggingface.co\/course\/chapter1\" target=\"_new\" rel=\"noopener nofollow\">https:\/\/huggingface.co\/course\/chapter1<\/a><\/li>\n<\/ul>","protected":false},"featured_media":0,"menu_order":0,"template":"","meta":{"_acf_changed":false,"content-type":"","inline_featured_image":false,"footnotes":""},"class_list":["post-477506","wiki","type-wiki","status-publish","hentry"],"acf":{"faq_title":"Frequently Asked Questions about <mark>Hugging Face: An In-Depth Guide to the Transformer Revolution<\/mark>","faq_items":[{"question":"What is Hugging Face?","answer":"<p>Hugging Face is a company and open-source community specializing in natural language processing (NLP) and artificial intelligence (AI). They are known for their Transformers library, which offers a vast array of pre-trained models for various NLP tasks.<\/p>"},{"question":"Who founded Hugging Face and when?","answer":"<p>Hugging Face was co-founded by Clement Delangue and Julien Chaumond in 2016 in New York City. Initially, the company focused on developing a chatbot, but their focus shifted towards transformer-based models for NLP in 2018.<\/p>"},{"question":"What are some key features of Hugging Face?","answer":"<p>Hugging Face offers diverse pre-trained models, broad language support, fine-tuning capabilities for specific tasks, and a thriving community-driven approach. These features make Hugging Face a leading platform for NLP tasks.<\/p>"},{"question":"What types of Hugging Face models exist?","answer":"<p>Hugging Face's Transformers library provides many transformer models, such as BERT, GPT-2, T5, RoBERTa, and DistilBERT, which can be used for a range of NLP tasks like text classification, information extraction, automatic summarization, translation, and text generation.<\/p>"},{"question":"What challenges can occur when using Hugging Face?","answer":"<p>Some challenges when using Hugging Face models may include the requirement of large amounts of data for training and the risk of bias in the models. Hugging Face addresses these challenges by providing detailed guides for fine-tuning models and a diverse range of pre-trained models.<\/p>"},{"question":"How does Hugging Face compare to similar tools?","answer":"<p>While other NLP tools like spaCy, NLTK, and StanfordNLP exist, Hugging Face stands out due to its extensive range of pre-trained models and its seamless integration with popular deep learning frameworks like PyTorch and TensorFlow.<\/p>"},{"question":"What is the future perspective of Hugging Face?","answer":"<p>Hugging Face continues to push the boundaries of NLP and AI research. They are focusing on the development and use of large language models like GPT-4 and exploring fields such as on-device and privacy-preserving machine learning.<\/p>"},{"question":"How can proxy servers be used with Hugging Face?","answer":"<p>Proxy servers can be used with Hugging Face for tasks like web scraping. The use of proxy servers allows for IP rotation for anonymity and facilitates the retrieval of web data, which can be processed using Hugging Face models for various NLP tasks.<\/p>"}]},"_links":{"self":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki\/477506","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki"}],"about":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/types\/wiki"}],"version-history":[{"count":0,"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/wiki\/477506\/revisions"}],"wp:attachment":[{"href":"https:\/\/oneproxy.pro\/cn\/wp-json\/wp\/v2\/media?parent=477506"}],"curies":[{"name":"\u53ef\u6e7f\u6027\u7c89\u5242","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}