- \n
- \u4f20\u7edf\u7684\u8bc4\u4f30\u65b9\u6cd5\uff0c\u5982ROUGE\u548cBLEU\uff0c\u4e3b\u8981\u57fa\u4e8e\u8bcd\u6c47\u5c42\u9762\u7684\u5339\u914d\uff0c\u65e0\u6cd5\u5145\u5206\u53cd\u6620\u8bed\u4e49\u7684\u4e00\u81f4\u6027\u3002<\/li>\n\n\n\n
- \u57fa\u4e8eLLM\u7684\u8bc4\u4f30\u5668\uff1a\n
- \n
- \u591a\u88ab\u7528\u4f5c\u6587\u672c\u8bc4\u4f30\u7684zero-shot\u8bc4\u4f30\u5668\uff0c\u5927\u591a\u6570\u7814\u7a76\u4e00\u79cd\u76f4\u63a5\u7684\u65b9\u6cd5\uff0c\u63d0\u4f9b\u7279\u5b9a\u4efb\u52a1\u7684\u8bf4\u660e\u6765\u8ba9LLM\u8bc4\u4f30\uff0c\u662f\u4e00\u79cd\u65e0\u53c2\u8003\u5ea6\u91cf\u3002<\/li>\n\n\n\n
- \u73b0\u6709\u7814\u7a76\u901a\u5e38\u91c7\u7528\u5355\u4e00\u4ee3\u7406\u7684\u65b9\u5f0f\uff0c\u5bb9\u6613\u53d7\u5230\u6a21\u578b\u504f\u89c1\u7684\u5f71\u54cd\u3002\u8fd9\u79cd\u504f\u89c1\u9650\u5236\u4e86\u8bc4\u4f30\u7684\u6709\u6548\u6027\u548c\u7a33\u5b9a\u6027\uff0c\u56e0\u6b64\u6709\u5fc5\u8981\u5f15\u5165\u66f4\u52a0\u5ba2\u89c2\u7684\u8bc4\u4f30\u6846\u67b6\u3002<\/li>\n<\/ul>\n<\/li>\n<\/ul>\n\n\n\n
\u4e09\u3001\u672c\u6587\u521b\u65b0\u70b9<\/h3>\n\n\n\n
- \n
- \u8bba\u6587\u63d0\u51fa\u4e86\u4e00\u4e2a\u65b0\u7684NLG\u8bc4\u4f30\u6846\u67b6\uff0c\u540d\u4e3aDEBATE\uff08Devil’s Advocate-Based Assessment and Text Evaluation\uff09\u8be5\u6846\u67b6\u5f15\u5165\u4e86\u591a\u4ee3\u7406\u8bc4\u5206\u7cfb\u7edf\uff0c\u5e76\u4e14\u901a\u8fc7\u5f15\u5165\u201c\u9b54\u9b3c\u4ee3\u8a00\u4eba\u201d\u7684\u89d2\u8272\u6765\u6279\u5224\u5176\u4ed6\u4ee3\u7406\u7684\u8bc4\u4f30\u7ed3\u679c\uff0c\u4ece\u800c\u51cf\u5c11\u5355\u4e00\u4ee3\u7406\u7684\u504f\u89c1\u95ee\u9898\u3002<\/li>\n\n\n\n
- DEBATE\u53ef\u4ee5\u901a\u8fc7\u591a\u6b21\u8fed\u4ee3\u8fa9\u8bba\uff0c\u6df1\u5165\u6316\u6398\u6587\u672c\u4e4b\u95f4\u7684\u8bed\u4e49\u5173\u8054<\/li>\n\n\n\n
- DEBATE\u901a\u8fc7\u591a\u4ee3\u7406\u95f4\u7684\u4e89\u8bba\u548c\u6279\u8bc4\uff0c\u63d0\u9ad8\u4e86\u8bc4\u4f30\u7684\u53ef\u9760\u6027\u548c\u4e0e\u4eba\u7c7b\u8bc4\u5206\u7684\u76f8\u5173\u6027\u3002<\/li>\n<\/ol>\n\n\n\n
\u56db\u3001\u7b97\u6cd5<\/h3>\n\n\n\n
DEBATE\u6846\u67b6\u4e2d\u5305\u542b\u4e09\u4e2a\u4e3b\u8981\u89d2\u8272\uff1a\u6307\u6325\u5b98\uff08Commander\uff09\u3001\u8bc4\u5206\u8005\uff08Scorer\uff09\u548c\u6279\u8bc4\u8005\uff08Critic\uff09\u3002\u6307\u6325\u5b98\u8d1f\u8d23\u7ec4\u7ec7\u8fa9\u8bba\uff0c\u8bc4\u5206\u8005\u6839\u636e\u4efb\u52a1\u6307\u4ee4\u8fdb\u884c\u8bc4\u4f30\uff0c\u6279\u8bc4\u8005\u5219\u626e\u6f14\u201c\u9b54\u9b3c\u4ee3\u8a00\u4eba\u201d\uff0c\u6279\u5224\u8bc4\u5206\u8005\u7684\u8bc4\u5206\u7ed3\u679c\uff0c\u5e2e\u52a9\u5e73\u8861\u8bc4\u4f30\u8fc7\u7a0b\u3002<\/p>\n\n\n\n
![\"\"](\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2024\/10\/\u56fe\u7247-1024x369.png\")
<\/figure>\n\n\n\n\u6279\u8bc4\u5bb6\u7684prompt\u4e3a\uff1a<\/p>\n\n\n\n
![\"\"](\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2024\/10\/\u56fe\u7247-1.png\")
<\/figure>\n\n\n\n\u4e94\u3001\u5b9e\u9a8c<\/h3>\n\n\n\n
1. \u57fa\u51c6\u6570\u636e\u96c6<\/h4>\n\n\n\n
- \n
- SummEval\uff1a\u7528\u4e8e\u6458\u8981\u751f\u6210\u7684\u8bc4\u4f30\uff0c\u8003\u5bdf\u6587\u672c\u7684\u6d41\u7545\u6027\u3001\u4e00\u81f4\u6027\u3001\u8fde\u8d2f\u6027\u548c\u76f8\u5173\u6027\u3002DEBATE\u5728\u591a\u4e2a\u7ef4\u5ea6\u4e0a\u8d85\u8fc7\u4e86\u73b0\u6709\u6700\u4f18\u65b9\u6cd5\uff0c\u663e\u793a\u51fa\u66f4\u9ad8\u7684\u4eba\u7c7b\u76f8\u5173\u6027\u3002<\/li>\n\n\n\n
- Topical-Chat\uff1a\u7528\u4e8e\u5bf9\u8bdd\u751f\u6210\u7684\u8bc4\u4f30\uff0c\u8003\u5bdf\u5bf9\u8bdd\u7684\u81ea\u7136\u6027\u3001\u8fde\u8d2f\u6027\u3001\u5438\u5f15\u529b\u548c\u57fa\u7840\u6027\u3002DEBATE\u540c\u6837\u53d6\u5f97\u4e86\u663e\u8457\u7684\u63d0\u5347\uff0c\u5c24\u5176\u5728\u81ea\u7136\u6027\u548c\u8fde\u8d2f\u6027\u4e0a\u8868\u73b0\u7a81\u51fa\u3002<\/li>\n<\/ul>\n\n\n\n
2. \u5bf9\u6bd4evalator<\/h4>\n\n\n\n
- \n
- \u4f20\u7edf\u7684\u8bc4\u4f30\u5668ROUGE-L (2004)\uff1b
<\/li>\n\n\n\n - \u57fa\u4e8e\u9884\u8bad\u7ec3\u8bed\u8a00\u6a21\u578b\u7684\u8bc4\u4f30\u5668\uff0cBERTScore \uff082022\uff09\u3001MoverScore \uff082019\uff09\u3001BARTScore \uff082021\uff09\u548cUniEval (2022)\uff1b
<\/li>\n\n\n\n - \u6700\u8fd1\u57fa\u4e8ellm\u7684\u8bc4\u4f30\u5668\uff0cGPTScore (2023),G-Eval \uff082023\uff09\u548cChatEval \uff082023\uff09\u3002
<\/li>\n\n\n\n - \u8fd8\u5305\u62ecMultiAgent\uff0c\u8fd9\u662f\u4e00\u4e2a\u7c7b\u4f3c\u4e8eDEBATE\u7684\u6846\u67b6\uff0c\u4f46\u6279\u8bc4\u5bb6\u88ab\u5206\u914d\u4e86\u4e00\u4e2a\u4e2d\u7acb\u7684\u8fa9\u8bba\u89d2\u8272\uff0c\u6807\u8bb0\u4e3aPlain\u3002<\/li>\n<\/ul>\n\n\n\n
3. \u5b9e\u9a8c\u7ed3\u679c<\/h4>\n\n\n\n
![\"\"](\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2024\/10\/\u56fe\u7247-2.png\")
<\/figure>\n\n\n\n![\"\"](\"https:\/\/www.15zhi.net\/blog\/wp-content\/uploads\/2024\/10\/\u56fe\u7247-3-1024x386.png\")
<\/figure>\n\n\n\n<\/p>\n\n\n\n
\u516d\u3001\u603b\u7ed3\u4e0e\u601d\u8003<\/h3>\n\n\n\n
- \n
- DEBATE\u6846\u67b6\u901a\u8fc7\u5f15\u5165\u591a\u4ee3\u7406\u7cfb\u7edf\uff0c\u5c24\u5176\u662f\u201c\u9b54\u9b3c\u4ee3\u8a00\u4eba\u201d\u7684\u6279\u8bc4\u673a\u5236\uff0c\u6709\u6548\u6539\u5584\u4e86\u673a\u5668\u751f\u6210\u6587\u672c\u7684\u8bc4\u4f30\u8d28\u91cf\uff0c\u6d88\u9664LLM\u7684\u504f\u89c1\u548c\u5c40\u9650\u6027\uff0c\u4ece\u800c\u63d0\u9ad8\u8bc4\u4f30\u7684\u51c6\u786e\u6027\u3002<\/li>\n\n\n\n
- \u8fd9\u4e00\u521b\u65b0\u6846\u67b6\u4e3a\u672a\u6765\u7684NLG\u8bc4\u4f30\u63d0\u4f9b\u4e86\u65b0\u7684\u601d\u8def\uff0c\u8bc1\u660e\u4e86\u793e\u4f1a\u79d1\u5b66\u4e2d\u7684\u8fa9\u8bba\u7b56\u7565\u5728LLM\u8bc4\u4f30\u4e2d\u7684\u6f5c\u529b\u3002<\/li>\n\n\n\n
- \u601d\u8003\uff1aDEBATE\u8bc1\u660e\u4e86\u591a\u4ee3\u7406\u534f\u4f5c\u8bc4\u4f30\u53ef\u4ee5\u663e\u8457\u63d0\u5347\u8bc4\u4f30\u7684\u8d28\u91cf\uff0c\u8fd9\u4e3a\u672a\u6765\u5f00\u53d1\u66f4\u590d\u6742\u7684\u3001\u591a\u5c42\u6b21\u7684\u8bc4\u4f30\u6846\u67b6\u63d0\u4f9b\u4e86\u53ef\u80fd\u3002\u672a\u6765\u7684\u8bc4\u4f30\u65b9\u6cd5\u53ef\u4ee5\u501f\u9274DEBATE\u7684\u591a\u4ee3\u7406\u6846\u67b6\uff0c\u7ed3\u5408\u66f4\u591a\u7684\u89d2\u8272\uff08\u5982\u7ea0\u6b63\u8005\u3001\u8865\u5145\u8005\u7b49\uff09\uff0c\u8fdb\u884c\u66f4\u52a0\u590d\u6742\u7684\u6587\u672c\u751f\u6210\u8bc4\u4f30\u3002<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"
\u4e00\u3001\u4fe1\u606f \u6765\u6e90\uff1a ACL2024\u4f5c\u8005\uff1aAlex G. Kim\u3001Keonwoo Kim\u3001Sangwon Yoon […]<\/p>\n","protected":false},"author":18,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-1261","post","type-post","status-publish","format-standard","hentry","category-events"],"_links":{"self":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1261","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/users\/18"}],"replies":[{"embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/comments?post=1261"}],"version-history":[{"count":1,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1261\/revisions"}],"predecessor-version":[{"id":1321,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/posts\/1261\/revisions\/1321"}],"wp:attachment":[{"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/media?parent=1261"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/categories?post=1261"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.15zhi.net\/blog\/wp-json\/wp\/v2\/tags?post=1261"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
- \u4f20\u7edf\u7684\u8bc4\u4f30\u5668ROUGE-L (2004)\uff1b