{"id":3026,"date":"2024-01-19T11:05:24","date_gmt":"2024-01-19T03:05:24","guid":{"rendered":"https:\/\/www.1ai.net\/?p=3026"},"modified":"2024-01-19T11:05:24","modified_gmt":"2024-01-19T03:05:24","slug":"%e8%8b%b1%e4%bc%9f%e8%be%be%e5%8f%91%e5%b8%83chatqa%e6%a8%a1%e5%9e%8b-%e6%80%a7%e8%83%bd%e8%be%be%e5%88%b0gpt-4%e7%ba%a7%e5%88%ab","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/3026.html","title":{"rendered":"Nvidia releases ChatQA model with GPT-4 performance"},"content":{"rendered":"<p><a href=\"https:\/\/www.1ai.net\/en\/tag\/%e8%8b%b1%e4%bc%9f%e8%be%be\" title=\"\u3010\u67e5\u770b\u542b\u6709[\u82f1\u4f1f\u8fbe]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">\u82f1\u4f1f\u8fbe<\/a>\u63a8\u51fa\u4e86<a href=\"https:\/\/www.1ai.net\/en\/tag\/chatqa\" title=\"\u3010\u67e5\u770b\u542b\u6709[ChatQA]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">ChatQA<\/a>\u6a21\u578b\uff0c\u636e\u6089\u5176\u6027\u80fd\u53ef\u4ee5\u5bf9\u98da<a href=\"https:\/\/www.1ai.net\/en\/tag\/gpt-4\" title=\"\u3010\u67e5\u770b\u542b\u6709[GPT-4]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">GPT-4<\/a>\uff0c\u4f7f\u7528\u4e24\u9636\u6bb5\u6307\u4ee4\u8c03\u6574\u548c\u6539\u8fdb\u7684\u4e0a\u4e0b\u6587\u68c0\u7d22\u7b49\u9ad8\u6548\u7684\u8bad\u7ec3\u65b9\u6cd5\u3002<\/p>\n<p class=\"article-content__img\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-3027\" title=\"201811151633429961_46\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/01\/201811151633429961_46.jpg\" alt=\"201811151633429961_46\" width=\"600\" height=\"399\" \/><\/p>\n<p>ChatQA\u662f\u4e00\u7ec4\u5bf9\u8bdd\u5f0f\u95ee\u7b54 \uff08QA\uff09 \u6a21\u578b\uff0c\u53ef\u4ee5\u83b7\u5f97GPT-4\u7ea7\u522b\u7684\u7cbe\u5ea6\u3002\u5177\u4f53\u6765\u8bf4\uff0c\u5f00\u53d1\u56e2\u961f\u63d0\u51fa\u4e86\u4e00\u79cd\u4e24\u9636\u6bb5\u6307\u4ee4\u8c03\u4f18\u65b9\u6cd5\uff0c\u53ef\u4ee5\u663e\u7740\u6539\u5584\u5927\u578b\u8bed\u8a00\u6a21\u578b(LLM)\u7684\u96f6\u6837\u672c\u5bf9\u8bddQA\u7ed3\u679c\u3002<\/p>\n<p>\u4e3a\u4e86\u5904\u7406\u5bf9\u8bdd\u5f0f QA \u4e2d\u7684\u68c0\u7d22\uff0c\u5728\u591a\u8f6e QA \u6570\u636e\u96c6\u4e0a\u5fae\u8c03\u4e86\u5bc6\u96c6\u68c0\u7d22\u5668\uff0c\u8fd9\u63d0\u4f9b\u4e86\u4e0e\u4f7f\u7528<span class=\"spamTxt\">\u6700\u5148<\/span>\u8fdb\u7684\u67e5\u8be2\u91cd\u5199\u6a21\u578b\u76f8\u5f53\u7684\u7ed3\u679c\uff0c\u540c\u65f6\u5927\u5927\u964d\u4f4e\u4e86\u90e8\u7f72\u6210\u672c\u3002\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0cChatQA-70B \u572810\u4e2a\u5bf9\u8bdd\u5f0f QA \u6570\u636e\u96c6\uff0854.14\u5bf953.90\uff09\u4e0a\u7684\u5e73\u5747\u5f97\u5206\u65b9\u9762\u4f18\u4e8e GPT-4\uff0c\u800c\u65e0\u9700\u4f9d\u8d56\u6765\u81ea OpenAI GPT \u6a21\u578b\u7684\u4efb\u4f55\u5408\u6210\u6570\u636e\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>NVIDIA has introduced the ChatQA model, which is said to perform against Biao GPT-4, using efficient training methods such as two-stage instruction tuning and improved context retrieval. ChatQA is a set of conversational question-and-answer (QA) models that can achieve GPT-4 level accuracy. Specifically, the development team proposes a two-stage instruction tuning approach that significantly improves zero-sample conversational QA results for large language models (LLMs). To handle retrieval in conversational QA, dense retrievers are fine-tuned on a multi-round QA dataset, which provides results comparable to using state-of-the-art query rewriting models while significantly reducing deployment costs. It is worth noting that ChatQA-70B provides a good result on 10 conversational QA datasets (54.14 vs. 5<\/p>","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[146],"tags":[966,510,239],"collection":[],"class_list":["post-3026","post","type-post","status-publish","format-standard","hentry","category-news","tag-chatqa","tag-gpt-4","tag-239"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/3026","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/comments?post=3026"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/3026\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=3026"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=3026"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=3026"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=3026"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}