{"id":91,"date":"2024-07-21T18:18:12","date_gmt":"2024-07-21T10:18:12","guid":{"rendered":"https:\/\/www.onetts.com\/ai\/?p=91"},"modified":"2024-07-22T12:55:40","modified_gmt":"2024-07-22T04:55:40","slug":"qwen2-72b","status":"publish","type":"post","link":"https:\/\/www.onetts.com\/ai\/models\/qwen2-72b\/","title":{"rendered":"Qwen2-72B"},"content":{"rendered":"<p>Qwen2-72B \u662f Qwen2 \u7cfb\u5217\u4e2d\u7684\u65d7\u8230\u6a21\u578b\uff0c\u4e00\u4e2a\u5177\u6709 72 \u4ebf\u53c2\u6570\u7684\u5927\u578b\u8bed\u8a00\u6a21\u578b\u3002\u5b83\u57fa\u4e8e Transformer \u67b6\u6784\uff0c\u91c7\u7528 SwiGLU \u6fc0\u6d3b\u51fd\u6570\u3001\u6ce8\u610f\u529b QKV \u504f\u7f6e\u548c\u5206\u7ec4\u67e5\u8be2\u6ce8\u610f\u529b\u7b49\u5148\u8fdb\u6280\u672f\u3002Qwen2-72B \u65e8\u5728\u63d0\u4f9b\u5353\u8d8a\u7684\u81ea\u7136\u8bed\u8a00\u5904\u7406\u80fd\u529b\uff0c\u5305\u62ec\u8bed\u8a00\u7406\u89e3\u3001\u751f\u6210\u3001\u591a\u8bed\u8a00\u5904\u7406\u3001\u7f16\u7a0b\u3001\u6570\u5b66\u63a8\u7406\u7b49\u3002<\/p>\n<h2>\u6a21\u578b\u8bc4\u6d4b<\/h2>\n<p>Qwen2-72B \u5728\u591a\u4e2a\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8868\u73b0\u51fa\u8272\uff0c\u8d85\u8d8a\u4e86\u4f17\u591a\u5f00\u6e90\u6a21\u578b\uff0c\u5e76\u5728\u4e0e\u4e13\u6709\u6a21\u578b\u7684\u7ade\u4e89\u4e2d\u5c55\u73b0\u4e86\u5176\u7ade\u4e89\u529b\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u5173\u952e\u7684\u8bc4\u6d4b\u7ed3\u679c\uff1a<\/p>\n<ul>\n<li><strong>\u82f1\u8bed\u4efb\u52a1<\/strong>\uff1a\u5728 MMLU\u3001MMLU-Pro\u3001GPQA \u7b49\u6570\u636e\u96c6\u4e0a\u53d6\u5f97\u4e86\u9886\u5148\u7684\u6210\u7ee9\u3002<\/li>\n<li><strong>\u7f16\u7a0b\u4efb\u52a1<\/strong>\uff1a\u5728 HumanEval\u3001MBPP \u7b49\u7f16\u7a0b\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u663e\u793a\u51fa\u5f3a\u5927\u7684\u6027\u80fd\u3002<\/li>\n<li><strong>\u6570\u5b66\u4efb\u52a1<\/strong>\uff1a\u5728 GSM8K\u3001MATH \u6570\u636e\u96c6\u4e0a\u5c55\u73b0\u4e86\u4f18\u79c0\u7684\u6570\u5b66\u95ee\u9898\u89e3\u51b3\u80fd\u529b\u3002<\/li>\n<li><strong>\u4e2d\u6587\u4efb\u52a1<\/strong>\uff1a\u5728 C-Eval\u3001CMMLU \u7b49\u4e2d\u6587\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u53d6\u5f97\u4e86\u9ad8\u5206\u3002<\/li>\n<li><strong>\u591a\u8bed\u8a00\u4efb\u52a1<\/strong>\uff1a\u5728 Multi-Exam\u3001Multi-Understanding \u7b49\u591a\u8bed\u8a00\u6570\u636e\u96c6\u4e0a\u8868\u73b0\u7a81\u51fa\u3002<\/li>\n<\/ul>\n<h2>\u90e8\u7f72\u4f7f\u7528<\/h2>\n<p>\u90e8\u7f72 Qwen2-72B \u6a21\u578b\u6d89\u53ca\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<ol>\n<li><strong>\u73af\u5883\u51c6\u5907<\/strong>\uff1a\u786e\u4fdd\u5b89\u88c5\u4e86 Python \u548c PyTorch\u3002\u63a8\u8350\u4f7f\u7528 Python 3.6 \u53ca\u4ee5\u4e0a\u7248\u672c\u3002<\/li>\n<li><strong>\u5b89\u88c5\u4f9d\u8d56<\/strong>\uff1a\u901a\u8fc7 pip \u5b89\u88c5 Hugging Face transformers \u5e93\uff1a<code class=\"inlineCode___uUrJ9\">pip install transformers&gt;=4.37.0<\/code>\u3002<\/li>\n<li><strong>\u4e0b\u8f7d\u6a21\u578b<\/strong>\uff1a\u4f7f\u7528 transformers \u5e93\u63d0\u4f9b\u7684\u63a5\u53e3\u4e0b\u8f7d Qwen2-72B \u6a21\u578b\u3002<\/li>\n<li><strong>\u6a21\u578b\u52a0\u8f7d<\/strong>\uff1a\u5728\u4ee3\u7801\u4e2d\u52a0\u8f7d\u6a21\u578b\uff0c\u51c6\u5907\u8fdb\u884c\u63a8\u7406\u6216\u8fdb\u4e00\u6b65\u7684\u8bad\u7ec3\u3002<\/li>\n<li><strong>\u5e94\u7528\u6a21\u578b<\/strong>\uff1a\u6839\u636e\u9700\u6c42\u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u6587\u672c\u751f\u6210\u3001\u95ee\u9898\u56de\u7b54\u6216\u5176\u4ed6 NLP \u4efb\u52a1\u3002<\/li>\n<\/ol>\n<h2>\u5e38\u89c1\u95ee\u9898<\/h2>\n<p><strong>Q: \u5982\u4f55\u907f\u514d\u5728\u5b89\u88c5 transformers \u5e93\u65f6\u9047\u5230 KeyError?<\/strong> A: \u786e\u4fdd\u5b89\u88c5\u4e86\u6b63\u786e\u7248\u672c\u7684 transformers \u5e93\uff0c\u4f7f\u7528 <code class=\"inlineCode___uUrJ9\">pip install transformers&gt;=4.37.0<\/code> \u547d\u4ee4\u5b89\u88c5\u3002<\/p>\n<p><strong>Q: Qwen2-72B \u652f\u6301\u54ea\u4e9b\u7c7b\u578b\u7684\u4efb\u52a1\uff1f<\/strong> A: Qwen2-72B \u652f\u6301\u5e7f\u6cdb\u7684\u4efb\u52a1\uff0c\u5305\u62ec\u8bed\u8a00\u7406\u89e3\u3001\u8bed\u8a00\u751f\u6210\u3001\u7f16\u7a0b\u3001\u6570\u5b66\u95ee\u9898\u89e3\u51b3\u3001\u591a\u8bed\u8a00\u5904\u7406\u7b49\u3002<\/p>\n<p><strong>Q: \u5982\u4f55\u5728 Qwen2-72B \u4e0a\u8fdb\u884c\u5fae\u8c03\uff1f<\/strong> A: \u53ef\u4ee5\u4f7f\u7528\u76d1\u7763\u5fae\u8c03\uff08SFT\uff09\u3001\u57fa\u4e8e\u4eba\u7c7b\u53cd\u9988\u7684\u5f3a\u5316\u5b66\u4e60\uff08RLHF\uff09\u6216\u7ee7\u7eed\u9884\u8bad\u7ec3\u7b49\u65b9\u6cd5\u5bf9 Qwen2-72B \u8fdb\u884c\u5fae\u8c03\u3002<\/p>\n<h2>\u76f8\u5173\u8d44\u6e90<\/h2>\n<ul>\n<li><strong>\u5b98\u65b9\u6587\u6863<\/strong>\uff1a<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/modelscope.cn\/models\/qwen\/Qwen2-72B\" rel=\"noopener\">Qwen2-72B Documentation<\/a><\/li>\n<li><strong>GitHub \u4ed3\u5e93<\/strong>\uff1a<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/github.com\/qwen\/Qwen2\" target=\"_blank\" rel=\"noopener\">Qwen2 GitHub<\/a><\/li>\n<li><strong>\u6280\u672f\u535a\u5ba2<\/strong>\uff1a<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/blog.qwen.ai\/qwen2\" target=\"_blank\" rel=\"noopener\">Qwen2 Technical Blog<\/a><\/li>\n<li><strong>\u6a21\u578b\u4e0b\u8f7d<\/strong>\uff1a\u901a\u8fc7 Hugging Face Model Hub \u4e0b\u8f7d Qwen2-72B \u6a21\u578b\u3002<\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"<p>Qwen2-72B \u662f Qwen2 \u7cfb\u5217\u4e2d\u7684\u65d7\u8230\u6a21\u578b\uff0c\u4e00\u4e2a\u5177\u6709 72 \u4ebf\u53c2\u6570\u7684\u5927\u578b\u8bed\u8a00\u6a21\u578b\u3002\u5b83\u57fa\u4e8e Transformer \u67b6\u6784\uff0c\u91c7\u7528 SwiGLU \u6fc0\u6d3b\u51fd\u6570\u3001\u6ce8\u610f\u529b QKV \u504f\u7f6e\u548c\u5206\u7ec4\u67e5\u8be2\u6ce8\u610f\u529b\u7b49\u5148\u8fdb\u6280\u672f\u3002Qwen2-72B \u65e8\u5728\u63d0\u4f9b\u5353\u8d8a\u7684\u81ea\u7136\u8bed\u8a00\u5904\u7406\u80fd\u529b\uff0c\u5305\u62ec\u8bed\u8a00\u7406\u89e3\u3001\u751f\u6210\u3001\u591a\u8bed\u8a00\u5904\u7406\u3001\u7f16\u7a0b\u3001\u6570\u5b66\u63a8\u7406\u7b49\u3002 \u6a21\u578b\u8bc4\u6d4b Qwen2-72B \u5728\u591a\u4e2a\u57fa\u51c6\u6d4b\u8bd5\u4e2d\u8868\u73b0\u51fa\u8272\uff0c\u8d85\u8d8a\u4e86\u4f17\u591a\u5f00\u6e90\u6a21\u578b\uff0c\u5e76\u5728\u4e0e\u4e13\u6709\u6a21\u578b\u7684\u7ade\u4e89\u4e2d\u5c55\u73b0\u4e86\u5176\u7ade\u4e89\u529b\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u5173\u952e\u7684\u8bc4\u6d4b\u7ed3\u679c\uff1a \u82f1\u8bed\u4efb\u52a1\uff1a\u5728 MMLU\u3001MMLU-Pro\u3001GPQA \u7b49\u6570\u636e\u96c6\u4e0a\u53d6\u5f97<\/p>\n","protected":false},"author":1,"featured_media":92,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[],"collection":[44],"company":[6],"rank":[],"class_list":["post-91","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-llm","collection-qwen2","company-alibaba"],"_links":{"self":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/91","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/comments?post=91"}],"version-history":[{"count":2,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/91\/revisions"}],"predecessor-version":[{"id":94,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/91\/revisions\/94"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/media\/92"}],"wp:attachment":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/media?parent=91"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/categories?post=91"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/tags?post=91"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/collection?post=91"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/company?post=91"},{"taxonomy":"rank","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/rank?post=91"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}