{"id":267055,"date":"2025-05-26T06:00:18","date_gmt":"2025-05-25T22:00:18","guid":{"rendered":"\/\/m.iemloyee.com\/?p=267055"},"modified":"2025-05-25T23:16:48","modified_gmt":"2025-05-25T15:16:48","slug":"nature-chemistry%e6%88%90%e6%9e%9c%e6%98%be%e7%a4%ba%ef%bc%9a%e5%a4%a7%e8%af%ad%e8%a8%80%e6%a8%a1%e5%9e%8b%e5%9c%a8%e5%8c%96%e5%ad%a6%e4%b8%8a%e5%ae%b9%e6%98%93%e5%a4%aa%e8%87%aa%e4%bf%a1","status":"publish","type":"post","link":"\/\/m.iemloyee.com\/?p=267055","title":{"rendered":"Nature Chemistry\u6210\u679c\u663e\u793a\uff1a\u5927\u8bed\u8a00\u6a21\u578b\u5728\u5316\u5b66\u4e0a\u5bb9\u6613\u592a\u81ea\u4fe1"},"content":{"rendered":"

\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u56e0\u5176\u80fd\u591f\u5904\u7406\u4eba\u7c7b\u8bed\u8a00\u5e76\u6267\u884c\u672a\u660e\u786e\u8bad\u7ec3\u4efb\u52a1\u800c\u53d7\u5230\u5e7f\u6cdb\u5173\u6ce8\u3002\u7136\u800c\uff0c\u76ee\u524d\u5bf9LLM\u7684\u5316\u5b66\u80fd\u529b\u53ea\u6709\u6709\u9650\u7684\u7cfb\u7edf\u6027\u4e86\u89e3\uff0c\u8fd9\u9700\u8981\u8fdb\u4e00\u6b65\u6539\u8fdb\u6a21\u578b\u4ee5\u51cf\u5c11\u6f5c\u5728\u5371\u5bb3\u3002\u5f17\u91cc\u5fb7\u91cc\u5e0c\u00b7\u5e2d\u52d2\u5927\u5b66\u8036\u62ff\u5206\u6821\u7684Kevin Maik Jablon\u63d0\u51fa\u4e86\u4e00\u4e2a\u540d\u4e3aChemBench\u7684\u81ea\u52a8\u5316\u6846\u67b6\uff0c\u7528\u4e8e\u8bc4\u4f30\u5f53\u524d\u6700\u5148\u8fdb\u7684LLMs\u5728\u5316\u5b66\u9886\u57df\u7684\u77e5\u8bc6\u548c\u63a8\u7406\u80fd\u529b\u3002\u4ed6\u4eec\u6574\u7406\u4e86\u8d85\u8fc72700\u4e2a\u95ee\u9898-\u7b54\u6848\u5bf9\uff0c\u8bc4\u4f30\u4e86\u591a\u4e2a\u5f00\u6e90\u548c\u95ed\u6e90\u7684LLMs\uff0c\u53d1\u73b0\u6700\u597d\u7684\u6a21\u578b\u5728\u5e73\u5747\u8868\u73b0\u4e0a\u8d85\u8fc7\u4e86\u4eba\u7c7b\u5316\u5b66\u4e13\u5bb6\u3002\u4f46\u6a21\u578b\u5728\u4e00\u4e9b\u57fa\u7840\u4efb\u52a1\u4e0a\u4ecd\u7136\u5b58\u5728\u56f0\u96be\uff0c\u5e76\u4e14\u5bb9\u6613\u7ed9\u51fa\u8fc7\u4e8e\u81ea\u4fe1\u7684\u9884\u6d4b\u3002ChemBench\u4e0d\u4ec5\u4e3aLLMs\u5728\u5316\u5b66\u9886\u57df\u7684\u4f18\u5316\u63d0\u4f9b\u4e86\u91cf\u5316\u57fa\u51c6\uff0c\u8fd8\u63ed\u793a\u4e86\u5f53\u524d\u6a21\u578b\u7684\u6f5c\u529b\u4e0e\u5c40\u9650\u3002\u7814\u7a76\u6210\u679c\u4ee5\u201cA framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertise of chemists\u201d\u4e3a\u9898\u53d1\u8868\u4e8eNature Chemistry\u3002<\/p>\n

\"\"<\/p>\n

\u4ed6\u4eec\u7684\u5de5\u4f5c\u4e3b\u8981\u521b\u65b0\u70b9\u5728\u4e8e\uff1a<\/strong><\/p>\n

1.\u9996\u4e2a\u5316\u5b66\u9886\u57df\u591a\u7ef4\u5ea6\u8bc4\u4f30\u6846\u67b6<\/strong>\uff1a\u586b\u8865\u4e86\u73b0\u6709\u57fa\u51c6\uff08\u5982BigBench\u3001LM Eval Harness\uff09\u5728\u5316\u5b66\u4e13\u4e1a\u4efb\u52a1\u4e0a\u7684\u7a7a\u767d\uff0c\u652f\u6301\u5f00\u653e\u6027\u548c\u5de5\u5177\u589e\u5f3a\u578b\u7cfb\u7edf\u7684\u8bc4\u4f30\u3002<\/p>\n

2.\u6570\u636e\u96c6\u6784\u5efa\u65b9\u6cd5\u521b\u65b0<\/strong>\uff1a\u7ed3\u5408\u624b\u52a8\u6574\u7406\uff08\u6559\u79d1\u4e66\u3001\u8003\u8bd5\u9898\uff09\u4e0e\u534a\u81ea\u52a8\u751f\u6210\uff08\u5316\u5b66\u6570\u636e\u5e93\u884d\u751f\u95ee\u9898\uff09\uff0c\u786e\u4fdd\u8986\u76d6\u5e7f\u5ea6\u548c\u8d28\u91cf\u3002\u5f15\u5165\u8bed\u4e49\u6807\u6ce8\uff08\u5982SMILES\u5b57\u7b26\u4e32\u3001\u65b9\u7a0b\u5f0f\u6807\u7b7e\uff09\uff0c\u9002\u914d\u79d1\u5b66\u6587\u672c\u5904\u7406\u9700\u6c42\u3002<\/p>\n

3.\u4eba\u7c7b-\u6a21\u578b\u5bf9\u6bd4\u5206\u6790<\/strong>\uff1a\u9996\u6b21\u7cfb\u7edf\u5316\u5bf9\u6bd4LLMs\u4e0e\u5316\u5b66\u4e13\u5bb6\u7684\u8868\u73b0\uff0c\u63ed\u793a\u6a21\u578b\u5728\u7279\u5b9a\u4efb\u52a1\u4e0a\u7684\u4f18\u52bf\uff08\u5982\u6559\u79d1\u4e66\u95ee\u9898\uff09\u4e0e\u52a3\u52bf\uff08\u5982\u7ed3\u6784\u63a8\u7406\uff09\u3002<\/p>\n

4.\u81ea\u4fe1\u5ea6\u8bc4\u4f30\u4e0e\u6821\u51c6\u7814\u7a76<\/strong>\uff1a\u901a\u8fc7\u63d0\u793a\u6a21\u578b\u81ea\u6211\u8bc4\u4f30\u4fe1\u5fc3\u6c34\u5e73\uff0c\u53d1\u73b0\u5176\u81ea\u4fe1\u5ea6\u4e0e\u7b54\u6848\u6b63\u786e\u7387\u8131\u8282\uff0c\u4e3a\u5b89\u5168\u5e94\u7528\u4e2d\u7684\u4e0d\u786e\u5b9a\u6027\u7ba1\u7406\u63d0\u4f9b\u6d1e\u89c1\u3002<\/p>\n

\"\"<\/p>\n

\u56fe1\uff1aChemBench\u6846\u67b6\u6982\u8ff0\u3002<\/p>\n

\"\"<\/p>\n

\u56fe2\uff1a\u4e3b\u9898\u548c\u6240\u9700\u6280\u80fd\u7684\u5206\u5e03\u3002<\/p>\n

\"\"<\/p>\n

\u56fe3\uff1aChemBench-Mini\u4e0a\u6a21\u578b\u548c\u4eba\u7c7b\u7684\u6027\u80fd\u3002<\/p>\n

\"\"<\/p>\n

\u56fe4\uff1aChemBench-Mini\u4e0a\u4e0d\u540c\u4e3b\u9898\u7684\u6a21\u578b\u548c\u4eba\u7c7b\u7684\u8868\u73b0\u3002<\/p>\n

\"\"<\/p>\n

\u56fe5\uff1a\u7f6e\u4fe1\u4f30\u8ba1\u503c\u7684\u53ef\u9760\u6027\u548c\u5206\u5e03\u3002<\/p>\n

\u8be5\u7814\u7a76\u6210\u679c\u663e\u793a\uff1a<\/p>\n

\u4e00\u65b9\u9762\uff0c\u7814\u7a76\u6210\u679c\u5f3a\u8c03\u4e86\u5927\u8bed\u8a00\u6a21\u578b\u5728\u5316\u5b66\u79d1\u5b66\u4e2d\u7684\u5f3a\u5927\u80fd\u529b\uff1a\u9886\u5148\u7684\u6a21\u578b\u5728\u8bb8\u591a\u4e3b\u9898\u7684\u5177\u4f53\u5316\u5b66\u95ee\u9898\u4e0a\u8d85\u8d8a\u4e86\u9886\u57df\u4e13\u5bb6\u3002\u53e6\u4e00\u65b9\u9762\uff0c\u4ecd\u5b58\u5728\u663e\u8457\u7684\u5c40\u9650\u6027\u3002\u5bf9\u4e8e\u975e\u5e38\u76f8\u5173\u7684\u8bdd\u9898\uff0c\u6a21\u578b\u63d0\u4f9b\u7684\u7b54\u6848\u662f\u9519\u8bef\u7684\u3002\u6b64\u5916\uff0c\u8bb8\u591a\u6a21\u578b\u65e0\u6cd5\u53ef\u9760\u5730\u4f30\u8ba1\u81ea\u8eab\u7684\u5c40\u9650\u6027\u3002<\/p>\n

\u7136\u800c\uff0c\u6a21\u578b\u5728\u8bc4\u4f30\u4e2d\u7684\u6210\u529f\u6216\u8bb8\u66f4\u591a\u5730\u63ed\u793a\u4e86\u6211\u4eec\u7528\u6765\u8bc4\u4f30\u6a21\u578b\u548c\u5316\u5b66\u5bb6\u7684\u95ee\u9898\u7684\u5c40\u9650\u6027\uff0c\u800c\u4e0d\u662f\u6a21\u578b\u672c\u8eab\u3002\u4f8b\u5982\uff0c\u867d\u7136\u6a21\u578b\u5728\u8bb8\u591a\u6559\u79d1\u4e66\u95ee\u9898\u4e0a\u8868\u73b0\u826f\u597d\uff0c\u4f46\u5728\u9700\u8981\u66f4\u591a\u5173\u4e8e\u5316\u5b66\u7ed3\u6784\u63a8\u7406\u7684\u95ee\u9898\uff08\u5982\u5f02\u6784\u4f53\u6570\u91cf\u6216\u6838\u78c1\u5171\u632f\u5cf0\uff09\u4e0a\u5374\u663e\u5f97\u529b\u4e0d\u4ece\u5fc3\u3002<\/p>\n

\u9274\u4e8e\u6a21\u578b\u5728\u7814\u7a76\u4e2d\u4f18\u4e8e\u666e\u901a\u4eba\u7c7b\uff0c\u7814\u7a76\u8005\u4efb\u52a1\u9700\u8981\u91cd\u65b0\u601d\u8003\u5982\u4f55\u6559\u6388\u548c\u8003\u8bd5\u5316\u5b66\u3002\u6279\u5224\u6027\u601d\u7ef4\u8d8a\u6765\u8d8a\u91cd\u8981\uff0c\u800c\u6b7b\u8bb0\u786c\u80cc\u6216\u8bb0\u5fc6\u4e8b\u5b9e\u4ecd\u7136\u662f\u5927\u8bed\u8a00\u6a21\u578b\u5c06\u7ee7\u7eed\u8d85\u8d8a\u4eba\u7c7b\u7684\u9886\u57df\uff08\u5f53\u8bad\u7ec3\u5728\u6b63\u786e\u7684\u8bad\u7ec3\u8bed\u6599\u5e93\u65f6\uff09\u3002<\/p>\n

\u7814\u7a76\u6210\u679c\u8fd8\u7a81\u663e\u4e86\u8bc4\u4f30\u6846\u67b6\u5e7f\u5ea6\u4e0e\u6df1\u5ea6\u4e4b\u95f4\u7684\u5fae\u5999\u6743\u8861\u3002\u4e0d\u540c\u4e3b\u9898\u4e0a\u7684\u6a21\u578b\u6027\u80fd\u5206\u6790\u663e\u793a\uff0c\u6a21\u578b\u5728\u6240\u6d4b\u8bd5\u7684\u5b50\u9886\u57df\u4e2d\u7684\u8868\u73b0\u5dee\u5f02\u5f88\u5927\u3002\u7136\u800c\uff0c\u5373\u4f7f\u5728\u540c\u4e00\u4e3b\u9898\u5185\uff0c\u6a21\u578b\u7684\u8868\u73b0\u4f1a\u56e0\u95ee\u9898\u7c7b\u578b\u548c\u56de\u7b54\u6240\u9700\u63a8\u7406\u7684\u4e0d\u540c\u800c\u5927\u76f8\u5f84\u5ead\u3002\u76ee\u524d\u5316\u5b66\u5927\u8bed\u8a00\u6a21\u578b\u7684\u8bc4\u4f30\u6846\u67b6\u4e3b\u8981\u8bbe\u8ba1\u7528\u4e8e\u8861\u91cf\u6a21\u578b\u5728\u7279\u5b9a\u5c5e\u6027\u9884\u6d4b\u4efb\u52a1\u4e0a\u7684\u8868\u73b0\u3002\u5b83\u4eec\u65e0\u6cd5\u7528\u4e8e\u8bc4\u4f30\u63a8\u7406\u6216\u4e3a\u79d1\u5b66\u5e94\u7528\u6784\u5efa\u7684\u7cfb\u7edf\u3002<\/p>\n

\u7814\u7a76\u663e\u793a\uff0c\u7cbe\u5fc3\u7b56\u5212\u7684\u57fa\u51c6\u53ef\u4ee5\u63d0\u4f9b\u66f4\u7ec6\u81f4\u7684\u7406\u89e3\uff0c\u63ed\u793a\u5927\u8bed\u8a00\u6a21\u578b\u5728\u5316\u5b66\u79d1\u5b66\u4e2d\u7684\u80fd\u529b\u3002\u91cd\u8981\u7684\u662f\uff0c\u5728\u5f00\u53d1\u66f4\u597d\u7684\u4eba\u673a\u4ea4\u4e92\u6846\u67b6\u65b9\u9762\u9700\u8981\u66f4\u591a\u5173\u6ce8\uff0c\u56e0\u4e3a\u6a21\u578b\u65e0\u6cd5\u4f30\u8ba1\u5176\u5c40\u9650\u6027\u3002<\/p>\n

\u5c3d\u7ba1\u53d1\u73b0\u6307\u51fa\u4e86\u8bb8\u591a\u6539\u8fdb\u5927\u8bed\u8a00\u6a21\u578b\u7cfb\u7edf\u7684\u9886\u57df\uff0c\u4f46\u4e5f\u8981\u8ba4\u8bc6\u5230\u660e\u786e\u5b9a\u4e49\u7684\u5ea6\u91cf\u6807\u51c6\u662f\u8bb8\u591a\u673a\u5668\u5b66\u4e60\u9886\u57df\uff0c\u5982\u8ba1\u7b97\u673a\u89c6\u89c9\u8fdb\u6b65\u7684\u5173\u952e\u3002\u867d\u7136\u5f53\u524d\u7cfb\u7edf\u53ef\u80fd\u8fdc\u672a\u8fbe\u5230\u50cf\u5316\u5b66\u5bb6\u90a3\u6837\u63a8\u7406\u7684\u7a0b\u5ea6\uff0c\u4f46\u7814\u7a76\u8005\u8ba4\u4e3aChemBench\u6846\u67b6\u5c06\u6210\u4e3a\u5b9e\u73b0\u8fd9\u4e00\u76ee\u6807\u7684\u57ab\u811a\u77f3\u3002<\/p>\n

\u8bba\u6587\u5730\u5740\uff1ahttps:\/\/www.nature.com\/articles\/s41557-025-01815-x<\/p>\n

\u00a0<\/p>\n","protected":false},"excerpt":{"rendered":"

\u5927\u578b\u8bed\u8a00\u6a21\u578b\uff08LLM\uff09\u56e0\u5176\u80fd\u591f\u5904\u7406\u4eba\u7c7b\u8bed\u8a00\u5e76\u6267\u884c\u672a\u660e\u786e\u8bad\u7ec3\u4efb\u52a1\u800c\u53d7\u5230\u5e7f\u6cdb\u5173\u6ce8\u3002\u7136\u800c\uff0c\u76ee\u524d\u5bf9LLM\u7684\u5316\u5b66\u80fd\u529b\u53ea\u6709\u6709\u9650\u7684\u7cfb\u7edf\u6027\u4e86\u89e3\uff0c\u8fd9\u9700\u8981\u8fdb\u4e00\u6b65\u6539\u8fdb\u6a21\u578b\u4ee5\u51cf\u5c11\u6f5c\u5728\u5371…<\/p>\n","protected":false},"author":2,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[215],"tags":[],"class_list":["post-267055","post","type-post","status-publish","format-standard","hentry","category-jiedu"],"_links":{"self":[{"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/posts\/267055"}],"collection":[{"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"\/\/m.iemloyee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=267055"}],"version-history":[{"count":3,"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/posts\/267055\/revisions"}],"predecessor-version":[{"id":267064,"href":"\/\/m.iemloyee.com\/index.php?rest_route=\/wp\/v2\/posts\/267055\/revisions\/267064"}],"wp:attachment":[{"href":"\/\/m.iemloyee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=267055"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"\/\/m.iemloyee.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=267055"},{"taxonomy":"post_tag","embeddable":true,"href":"\/\/m.iemloyee.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=267055"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}