{"id":2881,"date":"2024-01-15T11:42:17","date_gmt":"2024-01-15T03:42:17","guid":{"rendered":"https:\/\/www.1ai.net\/?p=2881"},"modified":"2024-01-15T11:46:44","modified_gmt":"2024-01-15T03:46:44","slug":"%e9%98%bf%e9%87%8c%e4%ba%91%ef%bc%9a%e5%8f%91%e5%b8%83%e7%ac%ac%e5%85%ab%e4%bb%a3%e4%bc%81%e4%b8%9a%e7%ba%a7%e5%ae%9e%e4%be%8b%ef%bc%8cai%e6%8e%a8%e7%90%86%e6%8f%90%e9%80%9f7%e5%80%8d","status":"publish","type":"post","link":"https:\/\/www.1ai.net\/en\/2881.html","title":{"rendered":"LLM AutoEval: AI platform automatically evaluates LLM in Google Colab"},"content":{"rendered":"<p>\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u9886\u57df\uff0c\u8bed\u8a00\u6a21\u578b\u7684\u8bc4\u4f30\u5bf9\u4e8e\u5f00\u53d1\u4eba\u5458\u63a8\u52a8\u8bed\u8a00\u7406\u89e3\u548c\u751f\u6210\u7684\u8fb9\u754c\u81f3\u5173\u91cd\u8981\u3002<a href=\"https:\/\/www.1ai.net\/en\/tag\/llm\" title=\"\u3010\u67e5\u770b\u542b\u6709[LLM]\u6807\u7b7e\u7684\u6587\u7ae0\u3011\" target=\"_blank\" rel=\"noopener\">LLM<\/a> AutoEval\u662f\u4e00\u6b3e\u65e8\u5728\u7b80\u5316\u548c\u52a0\u901f\u8bed\u8a00\u6a21\u578b\uff08LLMs\uff09\u8bc4\u4f30\u8fc7\u7a0b\u7684\u5de5\u5177\uff0c\u4e13\u4e3a\u5bfb\u6c42\u5feb\u901f\u9ad8\u6548\u8bc4\u4f30LLM\u6027\u80fd\u7684\u5f00\u53d1\u8005\u5b9a\u5236\u3002<\/p>\n<p class=\"article-content__img\"><img loading=\"lazy\" decoding=\"async\" class=\"alignnone size-full wp-image-2895\" src=\"https:\/\/www.1ai.net\/wp-content\/uploads\/2024\/01\/6384091454041066331867658.png\" alt=\"\" width=\"827\" height=\"395\" \/><\/p>\n<p>LLM AutoEval\u5177\u6709\u4ee5\u4e0b\u5173\u952e\u7279\u70b9:<\/p>\n<p>1. **\u81ea\u52a8\u5316\u8bbe\u7f6e\u548c\u6267\u884c:** LLM AutoEval\u901a\u8fc7\u4f7f\u7528RunPod\u7b80\u5316\u8bbe\u7f6e\u548c\u6267\u884c\u8fc7\u7a0b\uff0c\u63d0\u4f9b\u65b9\u4fbf\u7684Colab\u7b14\u8bb0\u672c\uff0c\u5b9e\u73b0\u65e0\u7f1d\u90e8\u7f72\u3002<\/p>\n<p>2. **\u53ef\u5b9a\u5236\u7684\u8bc4\u4f30\u53c2\u6570:** \u5f00\u53d1\u8005\u53ef\u4ee5\u901a\u8fc7\u9009\u62e9\u4e24\u4e2a\u57fa\u51c6\u5957\u4ef6 &#8211; nous\u6216openllm\uff0c\u5fae\u8c03\u4ed6\u4eec\u7684\u8bc4\u4f30\u3002\u8fd9\u63d0\u4f9b\u4e86\u5bf9LLM\u6027\u80fd\u7684\u7075\u6d3b\u8bc4\u4f30\u3002<\/p>\n<p>3. **\u6458\u8981\u751f\u6210\u548cGitHub Gist\u4e0a\u4f20:** LLM AutoEval\u751f\u6210\u8bc4\u4f30\u7ed3\u679c\u7684\u6458\u8981\uff0c\u5feb\u901f\u5c55\u793a\u6a21\u578b\u7684\u6027\u80fd\u3002\u8be5\u6458\u8981\u968f\u540e\u65b9\u4fbf\u5730\u4e0a\u4f20\u81f3GitHub Gist\uff0c\u4ee5\u4fbf\u8f7b\u677e\u5206\u4eab\u548c\u53c2\u8003\u3002<\/p>\n<p>LLM AutoEval\u63d0\u4f9b\u4e86\u7528\u6237\u53cb\u597d\u7684\u754c\u9762\uff0c\u53ef\u5b9a\u5236\u7684\u8bc4\u4f30\u53c2\u6570\uff0c\u6ee1\u8db3\u5f00\u53d1\u8005\u5728\u8bc4\u4f30\u8bed\u8a00\u6a21\u578b\u6027\u80fd\u65f6\u7684\u591a\u6837\u5316\u9700\u6c42\u3002\u4e24\u4e2a\u57fa\u51c6\u5957\u4ef6\uff0cnous\u548copenllm\uff0c\u63d0\u4f9b\u4e86\u4e0d\u540c\u7684\u4efb\u52a1\u5217\u8868\u8fdb\u884c\u8bc4\u4f30\u3002nous\u5957\u4ef6\u5305\u62ec\u8bf8\u5982AGIEval\u3001GPT4ALL\u3001TruthfulQA\u548cBigbench\u7b49\u4efb\u52a1\uff0c\u63a8\u8350\u7528\u4e8e\u5168\u9762\u8bc4\u4f30\u3002<\/p>\n<p>\u53e6\u4e00\u65b9\u9762\uff0copenllm\u5957\u4ef6\u5305\u542b\u4efb\u52a1\uff0c\u5982ARC\u3001HellaSwag\u3001MMLU\u3001Winogrande\u3001GSM8K\u548cTruthfulQA\uff0c\u5229\u7528vllm\u5b9e\u73b0\u589e\u5f3a\u901f\u5ea6\u3002\u5f00\u53d1\u8005\u53ef\u4ee5\u4eceHugging Face\u4e2d\u9009\u62e9\u7279\u5b9a\u7684\u6a21\u578bID\uff0c\u9009\u62e9<span class=\"spamTxt\">\u9996\u9009<\/span>GPU\uff0c\u6307\u5b9aGPU\u6570\u91cf\uff0c\u8bbe\u7f6e\u5bb9\u5668\u78c1\u76d8\u5927\u5c0f\uff0c\u9009\u62e9\u5728RunPod\u4e0a\u4f7f\u7528\u793e\u533a\u6216\u5b89\u5168\u4e91\uff0c\u5e76\u5207\u6362\u5bf9\u4e8e\u50cfPhi\u8fd9\u6837\u7684\u6a21\u578b\u7684\u4fe1\u4efb\u8fdc\u7a0b\u4ee3\u7801\u6807\u5fd7\u3002\u6b64\u5916\uff0c\u5f00\u53d1\u8005\u8fd8\u53ef\u4ee5\u6fc0\u6d3b\u8c03\u8bd5\u6a21\u5f0f\uff0c\u5c3d\u7ba1\u4e0d\u5efa\u8bae\u5728\u8bc4\u4f30\u540e\u4fdd\u6301Pod\u5904\u4e8e\u6d3b\u52a8\u72b6\u6001\u3002<\/p>\n<p>\u4e3a\u4e86\u5728LLM AutoEval\u4e2d\u5b9e\u73b0\u65e0\u7f1d\u7684\u4ee4\u724c\u96c6\u6210\uff0c\u7528\u6237\u5fc5\u987b\u4f7f\u7528Colab\u7684Secrets\u9009\u9879\u5361\uff0c\u5728\u90a3\u91cc\u521b\u5efa\u4e24\u4e2a\u540d\u4e3arunpod\u548cgithub\u7684\u79d8\u5bc6\uff0c\u5206\u522b\u5305\u542bRunPod\u548cGitHub\u6240\u9700\u7684\u4ee4\u724c\u3002<\/p>\n<p>\u4e24\u4e2a\u57fa\u51c6\u5957\u4ef6\uff0cnous\u548copenllm\uff0c\u6ee1\u8db3\u4e0d\u540c\u7684\u8bc4\u4f30\u9700\u6c42:<\/p>\n<p>1. Nous\u5957\u4ef6:*\u5f00\u53d1\u8005\u53ef\u4ee5\u5c06\u5176LLM\u7ed3\u679c\u4e0eOpenHermes-2.5-Mistral-7B\u3001Nous-Hermes-2-SOLAR-10.7B\u6216Nous-Hermes-2-Yi-34B\u7b49\u6a21\u578b\u8fdb\u884c\u6bd4\u8f83\u3002Teknium\u7684LLM-Benchmark-Logs\u53ef\u4f5c\u4e3a\u8bc4\u4f30\u6bd4\u8f83\u7684\u6709\u4ef7\u503c\u53c2\u8003\u3002<\/p>\n<p>2. Open LLM\u5957\u4ef6:\u8be5\u5957\u4ef6\u5141\u8bb8\u5f00\u53d1\u8005\u5c06\u5176\u6a21\u578b\u4e0e\u5217\u5728Open LLM\u6392\u884c\u699c\u4e0a\u7684\u6a21\u578b\u8fdb\u884c\u57fa\u51c6\u6d4b\u8bd5\uff0c\u4fc3\u8fdb\u793e\u533a\u5185\u66f4\u5e7f\u6cdb\u7684\u6bd4\u8f83\u3002<\/p>\n<p>\u5728LLM AutoEval\u4e2d\u8fdb\u884c\u6545\u969c\u6392\u9664\u5f97\u5230\u4e86\u5bf9\u5e38\u89c1\u95ee\u9898\u7684\u660e\u786e\u6307\u5bfc\u3002\u4f8b\u5982\uff0c\u201cError: File does not exist\u201d\u60c5\u666f\u63d0\u793a\u7528\u6237\u6fc0\u6d3b\u8c03\u8bd5\u6a21\u5f0f\u5e76\u91cd\u65b0\u8fd0\u884c\u8bc4\u4f30\uff0c\u4fbf\u4e8e\u68c0\u67e5\u65e5\u5fd7\u4ee5\u8bc6\u522b\u548c\u7ea0\u6b63\u4e0e\u7f3a\u5c11\u7684JSON\u6587\u4ef6\u76f8\u5173\u7684\u95ee\u9898\u3002\u5728\u201c700Killed\u201d\u9519\u8bef\u7684\u60c5\u51b5\u4e0b\uff0c\u8b66\u544a\u7528\u6237\u786c\u4ef6\u53ef\u80fd\u4e0d\u8db3\uff0c\u7279\u522b\u662f\u5728\u5c1d\u8bd5\u5728\u50cfRTX3070\u8fd9\u6837\u7684GPU\u4e0a\u8fd0\u884cOpen LLM\u57fa\u51c6\u5957\u4ef6\u65f6\u3002\u6700\u540e\uff0c\u5bf9\u4e8e\u8fc7\u65f6\u7684CUDA\u9a71\u52a8\u7a0b\u5e8f\u7684\u4e0d\u5e78\u60c5\u51b5\uff0c\u5efa\u8bae\u7528\u6237\u542f\u52a8\u65b0\u7684pod\u4ee5\u786e\u4fddLLM AutoEval\u5de5\u5177\u7684\u517c\u5bb9\u6027\u548c\u5e73\u7a33\u8fd0\u884c\u3002<\/p>\n<p>LM AutoEval\u662f\u4e00\u6b3e\u4e3a\u5f00\u53d1\u8005\u5728\u590d\u6742\u7684LLM\u8bc4\u4f30\u9886\u57df\u4e2d\u822a\u884c\u63d0\u4f9b\u5e2e\u52a9\u7684\u6709\u524d\u9014\u7684\u5de5\u5177\u3002\u4f5c\u4e3a\u4e00\u4e2a\u4e3a\u4e2a\u4eba\u4f7f\u7528\u800c\u8bbe\u8ba1\u7684\u4e0d\u65ad\u53d1\u5c55\u7684\u9879\u76ee\uff0c\u9f13\u52b1\u5f00\u53d1\u8005\u8c28\u614e\u4f7f\u7528\uff0c\u5e76\u4e3a\u5176\u53d1\u5c55\u505a\u51fa\u8d21\u732e\uff0c\u786e\u4fdd\u5728\u81ea\u7136\u8bed\u8a00\u5904\u7406\u793e\u533a\u4e2d\u6301\u7eed\u589e\u957f\u548c\u5b9e\u7528\u6027\u3002<\/p>\n<p>\u9879\u76ee\u7f51\u5740:https:\/\/github.com\/mlabonne\/llm-autoeval?tab=readme-ov-file<\/p>\n","protected":false},"excerpt":{"rendered":"<p>In the field of natural language processing, the assessment of language models is essential to the boundaries where developers promote language understanding and generation. LLM AutoEval is a tool to simplify and accelerate the language model (LLMs) assessment process, tailored for developers seeking to assess LLM performance quickly and efficiently. 1. ** Automation set-up and implementation:** LLM AutoEval achieves seamless deployment by using RunPod to simplify the set-up and implementation process and to provide a convenient Colab notebook. ** Custom assessment parameters:** Developers can fine-tune their assessments by selecting two baseline packages \u2013 nous or openllm. That<\/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":[473],"collection":[],"class_list":["post-2881","post","type-post","status-publish","format-standard","hentry","category-news","tag-llm"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/2881","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=2881"}],"version-history":[{"count":0,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/posts\/2881\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/media?parent=2881"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/categories?post=2881"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/tags?post=2881"},{"taxonomy":"collection","embeddable":true,"href":"https:\/\/www.1ai.net\/en\/wp-json\/wp\/v2\/collection?post=2881"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}