{"id":174,"date":"2024-07-25T16:15:36","date_gmt":"2024-07-25T08:15:36","guid":{"rendered":"https:\/\/www.onetts.com\/ai\/?p=174"},"modified":"2024-08-01T14:34:56","modified_gmt":"2024-08-01T06:34:56","slug":"glm-4-9b","status":"publish","type":"post","link":"https:\/\/www.onetts.com\/ai\/models\/glm-4-9b\/","title":{"rendered":"GLM-4-9B"},"content":{"rendered":"<p data-spm-anchor-id=\"a2c6h.13066512.0.i3.3cf036afHVA7eI\">GLM-4-9B \u662f\u7531\u667a\u8c31 AI \u63a8\u51fa\u7684\u6700\u65b0\u4e00\u4ee3\u5f00\u6e90\u9884\u8bad\u7ec3\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5c5e\u4e8e GLM-4 \u7cfb\u5217\u3002\u8be5\u6a21\u578b\u5728\u8bed\u4e49\u7406\u89e3\u3001\u6570\u5b66\u63a8\u7406\u3001\u4ee3\u7801\u751f\u6210\u548c\u77e5\u8bc6\u638c\u63e1\u7b49\u65b9\u9762\u8868\u73b0\u51fa\u8272\u3002GLM-4-9B \u7cfb\u5217\u6a21\u578b\u5305\u62ec\u57fa\u7840\u7248 GLM-4-9B\u3001\u5bf9\u8bdd\u7248 GLM-4-9B-Chat\u3001\u957f\u6587\u672c\u7248 GLM-4-9B-Chat-1M \u4ee5\u53ca\u591a\u6a21\u6001\u7248 GLM-4V-9B\u3002\u8fd9\u4e9b\u6a21\u578b\u4e0d\u4ec5\u652f\u6301\u591a\u8f6e\u5bf9\u8bdd\uff0c\u8fd8\u5177\u5907\u7f51\u9875\u6d4f\u89c8\u3001\u4ee3\u7801\u6267\u884c\u3001\u81ea\u5b9a\u4e49\u5de5\u5177\u8c03\u7528\u548c\u957f\u6587\u672c\u63a8\u7406\u7b49\u9ad8\u7ea7\u529f\u80fd\u3002<\/p>\n<h2 id=\"-\">\u6a21\u578b\u8bc4\u6d4b<\/h2>\n<p>GLM-4-9B \u7cfb\u5217\u6a21\u578b\u5728\u591a\u4e2a\u8bc4\u6d4b\u4efb\u52a1\u4e2d\u8868\u73b0\u4f18\u5f02\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u5173\u952e\u8bc4\u6d4b\u7ed3\u679c\uff1a<\/p>\n<ul>\n<li><strong>\u5bf9\u8bdd\u6a21\u578b\u5178\u578b\u4efb\u52a1<\/strong>: \u5728 AlignBench\u3001MT-Bench\u3001IFEval\u3001MMLU\u3001C-Eval \u7b49\u4efb\u52a1\u4e2d\uff0cGLM-4-9B-Chat \u5747\u8868\u73b0\u51fa\u8d85\u8d8a Llama-3-8B-Instruct \u7684\u6027\u80fd\u3002<\/li>\n<li><strong>\u57fa\u5ea7\u6a21\u578b\u5178\u578b\u4efb\u52a1<\/strong>: GLM-4-9B \u5728 MMLU\u3001C-Eval\u3001GPQA\u3001GSM8K\u3001MATH \u7b49\u4efb\u52a1\u4e2d\u8868\u73b0\u7a81\u51fa\u3002<\/li>\n<li><strong>\u957f\u6587\u672c\u80fd\u529b<\/strong>: \u5728 1M \u7684\u4e0a\u4e0b\u6587\u957f\u5ea6\u4e0b\u8fdb\u884c\u5927\u6d77\u635e\u9488\u5b9e\u9a8c\uff0cGLM-4-9B-Chat \u663e\u793a\u51fa\u5353\u8d8a\u7684\u957f\u6587\u672c\u5904\u7406\u80fd\u529b\u3002<\/li>\n<li><strong>\u591a\u8bed\u8a00\u80fd\u529b<\/strong>: \u5728 M-MMLU\u3001FLORES\u3001MGSM\u3001XWinograd\u3001XStoryCloze\u3001XCOPA \u7b49\u591a\u8bed\u8a00\u6570\u636e\u96c6\u4e0a\uff0cGLM-4-9B-Chat \u8868\u73b0\u4f18\u4e8e Llama-3-8B-Instruct\u3002<\/li>\n<li><strong>\u5de5\u5177\u8c03\u7528\u80fd\u529b<\/strong>: \u5728 Berkeley Function Calling Leaderboard \u4e0a\uff0cGLM-4-9B-Chat \u663e\u793a\u51fa\u8f83\u9ad8\u7684\u5de5\u5177\u8c03\u7528\u51c6\u786e\u6027\u3002<\/li>\n<li><strong>\u591a\u6a21\u6001\u80fd\u529b<\/strong>: GLM-4V-9B \u5728 MMBench-EN-Test\u3001MMBench-CN-Test\u3001SEEDBench_IMG \u7b49\u591a\u4e2a\u89c6\u89c9\u7406\u89e3\u4efb\u52a1\u4e2d\u8868\u73b0\u4f18\u5f02\u3002<\/li>\n<\/ul>\n<h2 id=\"-\">\u90e8\u7f72\u4f7f\u7528<\/h2>\n<h4 id=\"-\">\u8be6\u7ec6\u6b65\u9aa4<\/h4>\n<ol>\n<li><strong>\u5b89\u88c5\u4f9d\u8d56<\/strong>: \u786e\u4fdd\u7cfb\u7edf\u5b89\u88c5\u4e86 Python \u548c\u5fc5\u8981\u7684\u5e93\u3002\u53ef\u4ee5\u901a\u8fc7\u4ee5\u4e0b\u547d\u4ee4\u5b89\u88c5 transformers \u5e93\uff1a\n<pre><code class=\"lang-bash\">pip <span class=\"token function\">install<\/span> transformers\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u4e0b\u8f7d\u6a21\u578b<\/strong>: \u4ece <a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/huggingface.co\/THUDM\/glm-4-9b\">Huggingface<\/a> \u6216 <a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/modelscope.cn\/models\/ZhipuAI\/glm-4-9b\">ModelScope<\/a> \u4e0b\u8f7d GLM-4-9B \u6a21\u578b\u3002\n<pre><code class=\"lang-python\"><span class=\"token keyword\">import<\/span> torch\r\n<span class=\"token keyword\">from<\/span> transformers <span class=\"token keyword\">import<\/span> AutoModelForCausalLM<span class=\"token punctuation\">,<\/span> AutoTokenizer\r\n\r\ntokenizer <span class=\"token operator\">=<\/span> AutoTokenizer<span class=\"token punctuation\">.<\/span>from_pretrained<span class=\"token punctuation\">(<\/span><span class=\"token string\">\"THUDM\/glm-4-9b-chat\"<\/span><span class=\"token punctuation\">)<\/span>\r\nmodel <span class=\"token operator\">=<\/span> AutoModelForCausalLM<span class=\"token punctuation\">.<\/span>from_pretrained<span class=\"token punctuation\">(<\/span><span class=\"token string\">\"THUDM\/glm-4-9b-chat\"<\/span><span class=\"token punctuation\">)<\/span>\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u914d\u7f6e\u786c\u4ef6<\/strong>: \u6839\u636e\u6a21\u578b\u9700\u6c42\u914d\u7f6e\u786c\u4ef6\uff0c\u5982 GPU \u6216 CPU\u3002\u786e\u4fdd\u8bbe\u5907\u652f\u6301\u6a21\u578b\u8fd0\u884c\u3002<\/li>\n<li><strong>\u8fd0\u884c\u6a21\u578b<\/strong>: \u4f7f\u7528\u4ee5\u4e0b\u4ee3\u7801\u8fdb\u884c\u6a21\u578b\u63a8\u7406\uff1a\n<pre><code class=\"lang-python\">device <span class=\"token operator\">=<\/span> <span class=\"token string\">\"cuda\"<\/span> <span class=\"token keyword\">if<\/span> torch<span class=\"token punctuation\">.<\/span>cuda<span class=\"token punctuation\">.<\/span>is_available<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span> <span class=\"token keyword\">else<\/span> <span class=\"token string\">\"cpu\"<\/span>\r\nmodel<span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>eval<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span>\r\n\r\nquery <span class=\"token operator\">=<\/span> <span class=\"token string\">\"\u4f60\u597d\"<\/span>\r\ninputs <span class=\"token operator\">=<\/span> tokenizer<span class=\"token punctuation\">.<\/span>encode<span class=\"token punctuation\">(<\/span>query<span class=\"token punctuation\">,<\/span> return_tensors<span class=\"token operator\">=<\/span><span class=\"token string\">\"pt\"<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">.<\/span>to<span class=\"token punctuation\">(<\/span>device<span class=\"token punctuation\">)<\/span>\r\n<span class=\"token keyword\">with<\/span> torch<span class=\"token punctuation\">.<\/span>no_grad<span class=\"token punctuation\">(<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">:<\/span>\r\n    outputs <span class=\"token operator\">=<\/span> model<span class=\"token punctuation\">.<\/span>generate<span class=\"token punctuation\">(<\/span>inputs<span class=\"token punctuation\">,<\/span> max_length<span class=\"token operator\">=<\/span><span class=\"token number\">50<\/span><span class=\"token punctuation\">)<\/span>\r\n<span class=\"token keyword\">print<\/span><span class=\"token punctuation\">(<\/span>tokenizer<span class=\"token punctuation\">.<\/span>decode<span class=\"token punctuation\">(<\/span>outputs<span class=\"token punctuation\">[<\/span><span class=\"token number\">0<\/span><span class=\"token punctuation\">]<\/span><span class=\"token punctuation\">,<\/span> skip_special_tokens<span class=\"token operator\">=<\/span><span class=\"token boolean\">True<\/span><span class=\"token punctuation\">)<\/span><span class=\"token punctuation\">)<\/span>\r\n<\/code><\/pre>\n<\/li>\n<li><strong>\u4f18\u5316\u548c\u8c03\u6574<\/strong>: \u6839\u636e\u5b9e\u9645\u9700\u6c42\u8c03\u6574\u6a21\u578b\u53c2\u6570\uff0c\u5982\u6700\u5927\u957f\u5ea6\u3001\u6e29\u5ea6\u7b49\uff0c\u4ee5\u83b7\u5f97\u6700\u4f73\u6027\u80fd\u3002<\/li>\n<\/ol>\n<h2 id=\"-\">\u5e38\u89c1\u95ee\u9898<\/h2>\n<p><strong>Q: \u5982\u4f55\u5728\u4e0d\u540c\u8bbe\u5907\u4e0a\u90e8\u7f72 GLM-4-9B \u6a21\u578b\uff1f<\/strong><\/p>\n<p>A: \u53ef\u4ee5\u4f7f\u7528 transformers \u5e93\u5728 CPU \u6216 GPU \u4e0a\u90e8\u7f72\u6a21\u578b\u3002\u786e\u4fdd\u5b89\u88c5\u4e86\u6b63\u786e\u7684 CUDA \u7248\u672c\u548c\u9a71\u52a8\u7a0b\u5e8f\u3002<\/p>\n<p><strong>Q: \u5982\u4f55\u5904\u7406\u6a21\u578b\u7684\u957f\u6587\u672c\u8f93\u5165\uff1f<\/strong><\/p>\n<p>A: GLM-4-9B-Chat \u652f\u6301\u6700\u5927 128K \u7684\u4e0a\u4e0b\u6587\u957f\u5ea6\uff0cGLM-4-9B-Chat-1M \u652f\u6301 1M \u7684\u4e0a\u4e0b\u6587\u957f\u5ea6\u3002\u53ef\u4ee5\u901a\u8fc7\u8c03\u6574\u6a21\u578b\u53c2\u6570\u6765\u5904\u7406\u957f\u6587\u672c\u3002<\/p>\n<p><strong>Q: \u5982\u4f55\u8fdb\u884c\u6a21\u578b\u5fae\u8c03\uff1f<\/strong><\/p>\n<p>A: \u53ef\u4ee5\u4f7f\u7528 PEFT (LORA, P-Tuning) \u6216 SFT \u5fae\u8c03\u4ee3\u7801\u5bf9\u6a21\u578b\u8fdb\u884c\u5fae\u8c03\uff0c\u4ee5\u9002\u5e94\u7279\u5b9a\u4efb\u52a1\u3002<\/p>\n<h2 id=\"-\">\u76f8\u5173\u8d44\u6e90<\/h2>\n<ul>\n<li><strong>GitHub \u4ed3\u5e93<\/strong>:\u00a0<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/github.com\/THUDM\/GLM-4\">THUDM\/GLM-4<\/a><\/li>\n<li><strong>Huggingface \u6a21\u578b\u9875\u9762<\/strong>:\u00a0<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/huggingface.co\/THUDM\/glm-4-9b\">GLM-4-9B<\/a><\/li>\n<li><strong>ModelScope \u6a21\u578b\u9875\u9762<\/strong>:\u00a0<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/www.modelscope.cn\/models\/THUDM\/glm-4-9b\">GLM-4-9B<\/a><\/li>\n<li><strong>\u6280\u672f\u62a5\u544a<\/strong>:\u00a0<a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/arxiv.org\/pdf\/2406.12793\">GLM-4 \u6280\u672f\u62a5\u544a<\/a><\/li>\n<li><strong>\u76f8\u5173\u8bba\u6587<\/strong>: <a target=\"_blank\" rel=\"noopener noreferrer nofollow\" href=\"https:\/\/arxiv.org\/abs\/2406.12793\">ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools<\/a><\/li>\n<\/ul>\n<p>\u901a\u8fc7\u8fd9\u4e9b\u8d44\u6e90\uff0c\u4f60\u53ef\u4ee5\u66f4\u6df1\u5165\u5730\u4e86\u89e3 GLM-4-9B \u6a21\u578b\u7684\u8be6\u7ec6\u4fe1\u606f\u3001\u4f7f\u7528\u65b9\u6cd5\u548c\u6700\u65b0\u8fdb\u5c55\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"<p>GLM-4-9B \u662f\u7531\u667a\u8c31 AI \u63a8\u51fa\u7684\u6700\u65b0\u4e00\u4ee3\u5f00\u6e90\u9884\u8bad\u7ec3\u5927\u8bed\u8a00\u6a21\u578b\uff0c\u5c5e\u4e8e GLM-4 \u7cfb\u5217\u3002\u8be5\u6a21\u578b\u5728\u8bed\u4e49\u7406\u89e3\u3001\u6570\u5b66\u63a8\u7406\u3001\u4ee3\u7801\u751f\u6210\u548c\u77e5\u8bc6\u638c\u63e1\u7b49\u65b9\u9762\u8868\u73b0\u51fa\u8272\u3002GLM-4-9B \u7cfb\u5217\u6a21\u578b\u5305\u62ec\u57fa\u7840\u7248 GLM-4-9B\u3001\u5bf9\u8bdd\u7248 GLM-4-9B-Chat\u3001\u957f\u6587\u672c\u7248 GLM-4-9B-Chat-1M \u4ee5\u53ca\u591a\u6a21\u6001\u7248 GLM-4V-9B\u3002\u8fd9\u4e9b\u6a21\u578b\u4e0d\u4ec5\u652f\u6301\u591a\u8f6e\u5bf9\u8bdd\uff0c\u8fd8\u5177\u5907\u7f51\u9875\u6d4f\u89c8\u3001\u4ee3\u7801\u6267\u884c\u3001\u81ea\u5b9a\u4e49\u5de5\u5177\u8c03\u7528\u548c\u957f\u6587\u672c\u63a8\u7406\u7b49\u9ad8\u7ea7\u529f\u80fd\u3002 \u6a21\u578b\u8bc4\u6d4b GLM-4-9B \u7cfb\u5217\u6a21\u578b\u5728\u591a\u4e2a\u8bc4\u6d4b\u4efb\u52a1\u4e2d\u8868\u73b0\u4f18\u5f02\u3002\u4ee5\u4e0b\u662f\u4e00\u4e9b\u5173\u952e\u8bc4\u6d4b\u7ed3\u679c\uff1a \u5bf9\u8bdd\u6a21\u578b\u5178\u578b\u4efb<\/p>\n","protected":false},"author":1,"featured_media":175,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3],"tags":[],"collection":[52],"company":[17],"rank":[54,53],"class_list":["post-174","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-llm","collection-glm-4","company-zhipuai","rank-llm-leaderboard","rank-china-llm-ranking"],"_links":{"self":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/174","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=174"}],"version-history":[{"count":1,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/174\/revisions"}],"predecessor-version":[{"id":176,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/posts\/174\/revisions\/176"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/media\/175"}],"wp:attachment":[{"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/media?parent=174"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/categories?post=174"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/tags?post=174"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/collection?post=174"},{"taxonomy":"company","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/company?post=174"},{"taxonomy":"rank","embeddable":true,"href":"https:\/\/www.onetts.com\/ai\/wp-json\/wp\/v2\/rank?post=174"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}