{"id":5160,"date":"2024-11-23T10:12:37","date_gmt":"2024-11-23T02:12:37","guid":{"rendered":"https:\/\/nullthought.net\/?p=5160"},"modified":"2024-12-25T14:29:40","modified_gmt":"2024-12-25T06:29:40","slug":"temporalbench%ef%bc%9a%e5%9f%ba%e4%ba%8e%e7%bb%86%e7%b2%92%e5%ba%a6%e6%97%b6%e5%ba%8f%e7%90%86%e8%a7%a3%e7%9a%84%e5%a4%9a%e6%a8%a1%e6%80%81%e8%a7%86%e9%a2%91%e6%a8%a1%e5%9e%8b%e5%9f%ba%e5%87%86","status":"publish","type":"post","link":"https:\/\/nullthought.net\/?p=5160","title":{"rendered":"TemporalBench\uff1a\u57fa\u4e8e\u7ec6\u7c92\u5ea6\u65f6\u5e8f\u7406\u89e3\u7684\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u57fa\u51c6\u6d4b\u8bd5"},"content":{"rendered":"\n<p>\u8bba\u6587<strong><a href=\"https:\/\/www.arxiv.org\/abs\/2410.10818v2\" target=\"_blank\" rel=\"noreferrer noopener\">TemporalBench: Benchmarking Fine-grained Temporal Understanding for Multimodal Video Models<\/a><\/strong>\uff08\u300aTemporalBench: \u57fa\u4e8e\u7ec6\u7c92\u5ea6\u65f6\u5e8f\u7406\u89e3\u7684\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u57fa\u51c6\u6d4b\u8bd5\u300b\uff09\u6df1\u5165\u63a2\u8ba8\u4e86\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u5728\u65f6\u5e8f\u7406\u89e3\u4e0a\u7684\u8868\u73b0\uff0c\u63d0\u51fa\u4e86\u4e00\u4e2a\u540d\u4e3aTemporalBench\u7684\u65b0\u57fa\u51c6\u3002TemporalBench\u53ef\u7cfb\u7edf\u5316\u5730\u6d4b\u8bd5\u6a21\u578b\u5728\u6355\u6349\u7ec6\u8282\u3001\u7406\u89e3\u590d\u6742\u65f6\u5e8f\u52a8\u6001\u4e0a\u7684\u8868\u73b0\u3002\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u5f53\u524d\u7684\u591a\u6a21\u6001\u6a21\u578b\u5728\u8fd9\u65b9\u9762\u4ecd\u7136\u8fdc\u8fdc\u843d\u540e\u4e8e\u4eba\u7c7b\uff0c\u5c24\u5176\u662f\u5728\u957f\u89c6\u9891\u7406\u89e3\u548c\u7ec6\u7c92\u5ea6\u52a8\u4f5c\u63a8\u7406\u4e0a\u3002\u8bba\u6587\u8fd8\u63d0\u51fa\u4e86\u591a\u91cd\u4e8c\u5143\u51c6\u786e\u6027\uff08MBA\uff09\u4f5c\u4e3a\u65b0\u7684\u8bc4\u4f30\u6807\u51c6\uff0c\u4ee5\u6d88\u9664\u591a\u9879\u9009\u62e9\u95ee\u7b54\u4e2d\u7684\u504f\u5dee\uff0c\u771f\u6b63\u6d4b\u8bd5\u6a21\u578b\u7684\u65f6\u5e8f\u7406\u89e3\u80fd\u529b\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u4f5c\u8005\u4e3aMu Cai, Reuben Tan, Jianrui Zhang, Bocheng Zou, Kai Zhang, Feng Yao, Fangrui Zhu, Jing Gu, Yiwu Zhong, Yuzhang Shang, Yao Dou, Jaden Park, Jianfeng Gao, Yong Jae Lee, Jianwei Yang\uff0c\u6765\u81eaUniversity of Wisconsin-Madison, Microsoft Research, Ohio State University, University of California(San Diego), Northeastern University, University of California(Santa Cruz), Chinese University of Hong Kong, Illinois Institute of Technology, Georgia Institute of Technology\u7b49\u3002<\/p>\n\n\n\n<figure class=\"wp-block-image aligncenter size-full\"><img loading=\"lazy\" decoding=\"async\" width=\"1019\" height=\"541\" src=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-22.png\" alt=\"\" class=\"wp-image-5161\" srcset=\"https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-22.png 1019w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-22-300x159.png 300w, https:\/\/nullthought.net\/wp-content\/uploads\/2024\/11\/image-22-768x408.png 768w\" sizes=\"auto, (max-width: 1019px) 100vw, 1019px\" \/><figcaption class=\"wp-element-caption\"><strong><a href=\"https:\/\/www.arxiv.org\/abs\/2410.10818v2\" target=\"_blank\" rel=\"noreferrer noopener\">TemporalBench: Benchmarking Fine-grained Temporal Understanding for Multimodal Video Models<\/a><\/strong><\/figcaption><\/figure>\n\n\n\n<h4 class=\"wp-block-heading\">1. \u7814\u7a76\u80cc\u666f\u4e0e\u52a8\u673a<\/h4>\n\n\n\n<p>\u5728\u5f53\u524d\u7684\u4eba\u5de5\u667a\u80fd\u5e94\u7528\u4e2d\uff0c\u89c6\u9891\u7406\u89e3\u662f\u81f3\u5173\u91cd\u8981\u7684\uff0c\u6db5\u76d6\u4e86\u4ece\u6d3b\u52a8\u8bc6\u522b\u3001\u957f\u65f6\u5e8f\u52a8\u4f5c\u9884\u6d4b\u5230\u81ea\u52a8\u9a7e\u9a76\u548c\u673a\u5668\u4eba\u611f\u77e5\u7b49\u9886\u57df\u3002<strong>\u73b0\u6709\u7684\u89c6\u9891\u7406\u89e3\u57fa\u51c6\u5927\u591a\u7f3a\u4e4f\u7ec6\u7c92\u5ea6\u7684\u65f6\u5e8f\u6807\u6ce8\uff0c\u5bfc\u81f4\u8fd9\u4e9b\u57fa\u51c6\u66f4\u50cf\u662f\u56fe\u50cf\u7406\u89e3\u7684\u6269\u5c55\uff0c\u5ffd\u89c6\u4e86\u89c6\u9891\u672c\u8d28\u4e0a\u7684\u65f6\u95f4\u7ef4\u5ea6<\/strong>\u3002\u4f8b\u5982\uff0c\u5f88\u591a\u73b0\u6709\u7684\u57fa\u51c6\u53ea\u9700\u8981\u6a21\u578b\u8bc6\u522b\u89c6\u9891\u4e2d\u7684\u9759\u6001\u573a\u666f\uff0c\u751a\u81f3\u5355\u5e27\u753b\u9762\u5373\u53ef\u56de\u7b54\u5927\u90e8\u5206\u95ee\u9898\uff0c\u8fd9\u65e0\u6cd5\u771f\u5b9e\u53cd\u6620\u6a21\u578b\u5bf9\u65f6\u95f4\u52a8\u6001\u7684\u7406\u89e3\u80fd\u529b\u3002\u901a\u8fc7\u5bf9\u73b0\u6709\u6570\u636e\u96c6\u7684\u5206\u6790\uff0c\u4f5c\u8005\u53d1\u73b0\u8fd9\u4e9b\u57fa\u51c6\u901a\u5e38\u5b58\u5728\u201c\u5355\u5e27\u504f\u5dee\u201d\uff08Single Frame Bias\uff09\uff0c\u5373\u6a21\u578b\u53ef\u4ee5\u901a\u8fc7\u5206\u6790\u5355\u5e27\u9759\u6001\u4fe1\u606f\u6765\u83b7\u5f97\u9ad8\u5206\uff0c\u800c\u65e0\u9700\u7406\u89e3\u89c6\u9891\u4e2d\u7684\u65f6\u5e8f\u53d8\u5316\u3002\u8fd9\u4e9b\u95ee\u9898\u4fc3\u4f7f\u7814\u7a76\u8005\u4eec\u5f00\u53d1\u4e86TemporalBench\uff0c\u4ee5\u7cfb\u7edf\u5316\u5730\u8bc4\u4f30\u591a\u6a21\u6001\u6a21\u578b\u5bf9\u65f6\u95f4\u52a8\u6001\u7684\u7406\u89e3\u80fd\u529b\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">2. TemporalBench \u7684\u6784\u5efa\u4e0e\u7279\u70b9<\/h4>\n\n\n\n<p>TemporalBench\u662f\u4e00\u4e2a\u65b0\u5f00\u53d1\u7684\u57fa\u51c6\uff0c\u4e13\u95e8\u7528\u4e8e\u8bc4\u4f30\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u5728\u7ec6\u7c92\u5ea6\u65f6\u5e8f\u7406\u89e3\u65b9\u9762\u7684\u80fd\u529b\u3002\u5176\u6838\u5fc3\u8d21\u732e\u5305\u62ec\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u4e30\u5bcc\u7684\u65f6\u5e8f\u6807\u6ce8<\/strong>\uff1aTemporalBench\u5305\u542b\u7ea610,000\u4e2a\u95ee\u7b54\u5bf9\uff0c\u8fd9\u4e9b\u95ee\u7b54\u5bf9\u662f\u57fa\u4e8e\u7ea62,000\u4e2a\u89c6\u9891\u7247\u6bb5\u7684\u9ad8\u8d28\u91cf\u4eba\u5de5\u6ce8\u91ca\u751f\u6210\u7684\u3002\u8fd9\u4e9b\u6ce8\u91ca\u7ec6\u81f4\u63cf\u8ff0\u4e86\u89c6\u9891\u4e2d\u7684\u52a8\u4f5c\u9891\u6b21\u3001\u8fd0\u52a8\u5e45\u5ea6\u3001\u4e8b\u4ef6\u987a\u5e8f\u7b49\u7ec6\u7c92\u5ea6\u4fe1\u606f\u200b\u3002<\/li>\n\n\n\n<li><strong>\u591a\u6837\u5316\u7684\u4efb\u52a1\u652f\u6301<\/strong>\uff1aTemporalBench\u652f\u6301\u591a\u79cd\u4efb\u52a1\u7684\u8bc4\u4f30\uff0c\u5305\u62ec\u89c6\u9891\u95ee\u7b54\u3001\u89c6\u9891\u5b57\u5e55\u751f\u6210\u3001\u957f\u89c6\u9891\u7406\u89e3\u7b49\uff0c\u4e14\u6db5\u76d6\u4e0d\u540c\u6a21\u578b\uff0c\u5982\u591a\u6a21\u6001\u89c6\u9891\u5d4c\u5165\u6a21\u578b\u548c\u6587\u672c\u751f\u6210\u6a21\u578b\u3002\u8fd9\u6837\uff0c\u7814\u7a76\u4eba\u5458\u53ef\u4ee5\u5229\u7528\u8fd9\u4e9b\u4efb\u52a1\u5168\u65b9\u4f4d\u5730\u8bc4\u4f30\u6a21\u578b\u5728\u65f6\u5e8f\u7406\u89e3\u4e0a\u7684\u8868\u73b0\u3002<\/li>\n\n\n\n<li><strong>\u7ec6\u7c92\u5ea6\u7684\u65f6\u95f4\u63a8\u7406\u80fd\u529b\u8bc4\u4f30<\/strong>\uff1aTemporalBench\u901a\u8fc7\u7cbe\u5fc3\u8bbe\u8ba1\u7684\u95ee\u7b54\u5bf9\uff0c\u6d4b\u8bd5\u6a21\u578b\u5bf9\u89c6\u9891\u4e2d\u8fde\u7eed\u52a8\u4f5c\u548c\u7ec6\u5fae\u53d8\u5316\u7684\u7406\u89e3\u80fd\u529b\u3002\u4f8b\u5982\uff0c\u8981\u6c42\u6a21\u578b\u533a\u522b\u201c\u5207\u4e09\u6b21\u751f\u59dc\u201d\u4e0e\u201c\u5207\u4e24\u6b21\u751f\u59dc\u201d\u7684\u5dee\u5f02\u3002\u8fd9\u79cd\u7ec6\u8282\u4e0d\u4ec5\u8003\u5bdf\u6a21\u578b\u5bf9\u5355\u4e2a\u52a8\u4f5c\u7684\u8bc6\u522b\u80fd\u529b\uff0c\u8fd8\u8003\u5bdf\u5176\u5bf9\u52a8\u4f5c\u95f4\u590d\u6742\u65f6\u5e8f\u5173\u7cfb\u7684\u7406\u89e3\u200b\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">3. \u591a\u6a21\u6001\u89c6\u9891\u7406\u89e3\u4e2d\u7684\u6311\u6218<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">3.1 \u5355\u5e27\u504f\u5dee\u4e0e\u8bed\u8a00\u504f\u5dee<\/h5>\n\n\n\n<p>\u4f5c\u8005\u6307\u51fa\uff0c\u73b0\u6709\u57fa\u51c6\u5b58\u5728\u660e\u663e\u7684\u5355\u5e27\u504f\u5dee\u548c\u8bed\u8a00\u504f\u5dee\u3002\u5355\u5e27\u504f\u5dee\u6307\u7684\u662f\u6a21\u578b\u53ea\u9700\u4ece\u89c6\u9891\u4e2d\u7684\u67d0\u4e00\u9759\u6001\u5e27\u63d0\u53d6\u4fe1\u606f\uff0c\u5c31\u80fd\u6b63\u786e\u56de\u7b54\u5927\u90e8\u5206\u95ee\u9898\uff0c\u800c\u4e0d\u9700\u8981\u8003\u8651\u89c6\u9891\u4e2d\u7684\u52a8\u6001\u53d8\u5316\u3002\u4f8b\u5982\uff0c\u5728\u4e00\u4e9b\u89c6\u9891\u95ee\u7b54\u4efb\u52a1\u4e2d\uff0c\u53ea\u9700\u77e5\u9053\u89c6\u9891\u4e2d\u7684\u573a\u666f\u5143\u7d20\uff08\u4f8b\u5982\u4e00\u4e2a\u4eba\u5728\u505a\u996d\uff09\u5c31\u8db3\u4ee5\u63a8\u65ad\u7b54\u6848\uff0c\u800c\u65e0\u9700\u4e86\u89e3\u4e8b\u4ef6\u7684\u53d1\u751f\u987a\u5e8f\u6216\u52a8\u4f5c\u7684\u7ec6\u8282\u3002<\/p>\n\n\n\n<p>\u8bed\u8a00\u504f\u5dee\u5219\u4f53\u73b0\u5728\uff0c\u73b0\u6709\u7684\u591a\u6a21\u6001\u6a21\u578b\u5f80\u5f80\u57fa\u4e8e\u8bed\u8a00\u6a21\u578b\u8fdb\u884c\u8bad\u7ec3\uff0c\u4f7f\u5f97\u5b83\u4eec\u5728\u9047\u5230\u89c6\u9891\u4e2d\u7684\u7ec6\u8282\u95ee\u9898\u65f6\u4f1a\u66f4\u591a\u4f9d\u8d56\u8bed\u8a00\u7406\u89e3\u800c\u4e0d\u662f\u65f6\u5e8f\u63a8\u7406\u3002\u8fd9\u610f\u5473\u7740\uff0c\u6a21\u578b\u53ef\u80fd\u53ea\u662f\u57fa\u4e8e\u5bf9\u95ee\u9898\u548c\u7b54\u6848\u9009\u9879\u7684\u8bed\u8a00\u6a21\u5f0f\u5339\u914d\u6765\u505a\u51fa\u56de\u7b54\uff0c\u800c\u4e0d\u662f\u771f\u6b63\u7406\u89e3\u89c6\u9891\u4e2d\u7684\u5185\u5bb9\u200b\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">3.2 \u7ec6\u7c92\u5ea6\u89c6\u9891\u7406\u89e3\u7684\u590d\u6742\u6027<\/h5>\n\n\n\n<p>\u7ec6\u7c92\u5ea6\u89c6\u9891\u7406\u89e3\u9700\u8981\u6a21\u578b\u80fd\u591f\u6355\u6349\u5230\u89c6\u9891\u4e2d\u7684\u7ec6\u5fae\u53d8\u5316\u548c\u590d\u6742\u7684\u65f6\u5e8f\u4f9d\u8d56\u5173\u7cfb\u3002\u4f8b\u5982\uff0cTemporalBench\u4e2d\u5305\u542b\u5927\u91cf\u63cf\u8ff0\u4eba\u7c7b\u52a8\u4f5c\u7684\u7247\u6bb5\uff0c\u8fd9\u4e9b\u7247\u6bb5\u7684\u65f6\u5e8f\u52a8\u6001\u96be\u4ee5\u7528\u5355\u4e00\u753b\u9762\u6355\u6349\uff0c\u6a21\u578b\u5fc5\u987b\u7406\u89e3\u52a8\u4f5c\u662f\u5982\u4f55\u9010\u6b65\u5c55\u5f00\u7684\uff0c\u5982\u201c\u4e00\u4e2a\u4eba\u7528\u53f3\u624b\u62ff\u8d77\u684c\u4e0a\u7684\u7269\u54c1\uff0c\u518d\u7528\u5de6\u624b\u62c6\u5f00\u5305\u88c5\u201d\u3002\u8fd9\u79cd\u5bf9\u590d\u6742\u52a8\u6001\u7684\u7406\u89e3\u662f\u73b0\u6709\u5927\u90e8\u5206\u89c6\u9891\u7406\u89e3\u6a21\u578b\u6240\u4e0d\u5177\u5907\u7684\uff0c\u56e0\u6b64TemporalBench\u65e8\u5728\u586b\u8865\u8fd9\u4e00\u7a7a\u767d\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">4. \u6570\u636e\u96c6\u6784\u5efa\u4e0e\u6ce8\u91ca\u65b9\u6cd5<\/h4>\n\n\n\n<p>TemporalBench\u7684\u6570\u636e\u96c6\u7531\u4eba\u5de5\u7cbe\u5fc3\u6ce8\u91ca\uff0c\u89c6\u9891\u7247\u6bb5\u4e3b\u8981\u53d6\u81ea\u4e8e\u73b0\u6709\u7684\u89c6\u9891\u57fa\u51c6\uff0c\u5305\u62ec\u7a0b\u5e8f\u89c6\u9891\uff08\u5982COIN\uff09\u3001\u4eba\u7c7b\u6d3b\u52a8\u89c6\u9891\uff08\u5982ActivityNet\u3001Charades\uff09\u3001\u81ea\u6211\u89c6\u89d2\u89c6\u9891\uff08\u5982EgoExo4D\uff09\u3001\u7535\u5f71\u63cf\u8ff0\uff08\u5982MPI Movie Description\uff09\u7b49\u3002\u6570\u636e\u96c6\u6784\u5efa\u8fc7\u7a0b\u5305\u62ec\u4ee5\u4e0b\u51e0\u4e2a\u9636\u6bb5\uff1a<\/p>\n\n\n\n<ul class=\"wp-block-list\">\n<li><strong>\u89c6\u9891\u91c7\u96c6<\/strong>\uff1a\u4ece\u591a\u4e2a\u5df2\u6709\u7684\u6570\u636e\u96c6\u4e2d\u91c7\u6837\u89c6\u9891\u7247\u6bb5\uff0c\u5171\u8ba1\u7ea62,000\u4e2a\u7247\u6bb5\u3002\u6bcf\u4e2a\u89c6\u9891\u7247\u6bb5\u90fd\u7ecf\u8fc7\u7cbe\u5fc3\u6311\u9009\uff0c\u786e\u4fdd\u6db5\u76d6\u591a\u79cd\u4eba\u7c7b\u6d3b\u52a8\u548c\u4e0d\u540c\u573a\u666f\u200b\u3002<\/li>\n\n\n\n<li><strong>\u6b63\u5411\u6ce8\u91ca\u751f\u6210<\/strong>\uff1a\u7531\u7ecf\u9a8c\u4e30\u5bcc\u7684AMT\u5de5\u4eba\u5bf9\u89c6\u9891\u8fdb\u884c\u521d\u6b65\u6ce8\u91ca\uff0c\u63d0\u4f9b\u7ec6\u81f4\u7684\u63cf\u8ff0\uff0c\u5305\u62ec\u6bcf\u4e2a\u52a8\u4f5c\u7684\u5f00\u59cb\u548c\u7ed3\u675f\u65f6\u95f4\u3001\u52a8\u4f5c\u7684\u5177\u4f53\u7ec6\u8282\u7b49\u3002\u4e4b\u540e\uff0c\u8bba\u6587\u7684\u4f5c\u8005\u5bf9\u8fd9\u4e9b\u6ce8\u91ca\u8fdb\u884c\u6821\u5bf9\u548c\u5b8c\u5584\uff0c\u786e\u4fdd\u63cf\u8ff0\u51c6\u786e\u3001\u8be6\u5c3d\u200b\u3002<\/li>\n\n\n\n<li><strong>\u8d1f\u5411\u6ce8\u91ca\u751f\u6210<\/strong>\uff1a\u4e3a\u4e86\u751f\u6210\u5177\u6709\u6311\u6218\u6027\u7684\u591a\u9879\u9009\u62e9\u9898\uff0c\u4f5c\u8005\u5229\u7528GPT-4\u548c\u5176\u4ed6\u5927\u578b\u8bed\u8a00\u6a21\u578b\u751f\u6210\u8d1f\u5411\u63cf\u8ff0\uff0c\u8fd9\u4e9b\u63cf\u8ff0\u5728\u7ec6\u8282\u4e0a\u7565\u6709\u4e0d\u540c\uff0c\u4f8b\u5982\u5c06\u201c\u5207\u4e09\u6b21\u751f\u59dc\u201d\u6539\u4e3a\u201c\u5207\u4e24\u6b21\u751f\u59dc\u201d\uff0c\u6216\u5c06\u52a8\u4f5c\u7684\u987a\u5e8f\u8c03\u6574\uff0c\u4ee5\u9a8c\u8bc1\u6a21\u578b\u662f\u5426\u80fd\u771f\u6b63\u7406\u89e3\u89c6\u9891\u4e2d\u7684\u65f6\u5e8f\u52a8\u6001\u3002<\/li>\n<\/ul>\n\n\n\n<h4 class=\"wp-block-heading\">5. \u8bc4\u4f30\u6307\u6807\u4e0e\u5b9e\u9a8c\u7ed3\u679c<\/h4>\n\n\n\n<h5 class=\"wp-block-heading\">5.1 \u591a\u91cd\u4e8c\u5143\u51c6\u786e\u6027\uff08Multiple Binary Accuracy, MBA\uff09<\/h5>\n\n\n\n<p>\u8bba\u6587\u4e2d\u63d0\u51fa\u4e86\u591a\u91cd\u4e8c\u5143\u51c6\u786e\u6027\uff08MBA\uff09\u6765\u53d6\u4ee3\u4f20\u7edf\u7684\u591a\u9879\u9009\u62e9\u8bc4\u4f30\u65b9\u6cd5\u3002\u5728\u4f20\u7edf\u7684\u591a\u9879\u9009\u62e9\u95ee\u7b54\u4e2d\uff0c\u6a21\u578b\u5f80\u5f80\u53ef\u4ee5\u901a\u8fc7\u6392\u9664\u4e0d\u592a\u53ef\u80fd\u7684\u9009\u9879\u6765\u731c\u6d4b\u6b63\u786e\u7b54\u6848\uff0c\u7279\u522b\u662f\u5f53\u6240\u6709\u9519\u8bef\u9009\u9879\u53ea\u662f\u5bf9\u6b63\u786e\u7b54\u6848\u7684\u7ec6\u5fae\u6539\u52a8\u65f6\uff0c\u6a21\u578b\u53ef\u4ee5\u68c0\u6d4b\u8fd9\u4e9b\u7ec6\u5fae\u53d8\u5316\u6765\u627e\u5230\u201c\u6700\u5408\u7406\u201d\u7684\u63cf\u8ff0\u3002\u4e3a\u6b64\uff0c\u4f5c\u8005\u63d0\u51fa\u5c06\u591a\u9879\u9009\u62e9\u95ee\u9898\u62c6\u89e3\u4e3a\u591a\u4e2a\u4e8c\u5143\u9009\u62e9\u95ee\u9898\uff0c\u4ee5\u66f4\u516c\u5e73\u5730\u8bc4\u4f30\u6a21\u578b\u5bf9\u89c6\u9891\u65f6\u5e8f\u7684\u7406\u89e3\u80fd\u529b\u3002\u5b9e\u9a8c\u7ed3\u679c\u8868\u660e\uff0c\u6a21\u578b\u5728\u8fd9\u79cd\u65b0\u7684\u8bc4\u4f30\u6807\u51c6\u4e0b\u7684\u8868\u73b0\u663e\u8457\u4e0b\u964d\uff0c\u8fd9\u8868\u660e\u73b0\u6709\u6a21\u578b\u7684\u65f6\u5e8f\u63a8\u7406\u80fd\u529b\u4ecd\u7136\u6709\u5f88\u5927\u63d0\u5347\u7a7a\u95f4\u200b\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">5.2 \u89c6\u9891\u6a21\u578b\u7684\u8868\u73b0<\/h5>\n\n\n\n<p>\u5b9e\u9a8c\u7ed3\u679c\u663e\u793a\uff0c\u73b0\u6709\u7684\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u5728TemporalBench\u4e0a\u7684\u8868\u73b0\u8fdc\u4f4e\u4e8e\u4eba\u7c7b\u6c34\u5e73\u3002\u4f8b\u5982\uff0cGPT-4\u5728\u77ed\u89c6\u9891\u7684\u591a\u91cd\u4e8c\u5143\u95ee\u7b54\u4efb\u52a1\u4e2d\u7684\u51c6\u786e\u7387\u4e3a38.5%\uff0c\u800c\u4eba\u7c7b\u7684\u5e73\u5747\u8868\u73b0\u4e3a67.9%\u3002\u66f4\u957f\u7684\u89c6\u9891\u7406\u89e3\u4efb\u52a1\u5bf9\u73b0\u6709\u6a21\u578b\u63d0\u51fa\u4e86\u66f4\u5927\u6311\u6218\uff0c\u6a21\u578b\u7684\u8868\u73b0\u666e\u904d\u8f83\u5dee\uff0c\u7279\u522b\u662f\u5728\u7406\u89e3\u591a\u4e2a\u8fde\u7eed\u7684\u590d\u6742\u52a8\u4f5c\u65f6\uff0c\u6a21\u578b\u7684\u8868\u73b0\u8fdc\u4e0d\u53ca\u4eba\u7c7b\u200b\u3002<\/p>\n\n\n\n<p>\u6b64\u5916\uff0c\u4f5c\u8005\u8fd8\u53d1\u73b0\uff0c\u589e\u52a0\u8f93\u5165\u89c6\u9891\u7684\u5e27\u6570\u5e76\u6ca1\u6709\u663e\u8457\u63d0\u5347\u6a21\u578b\u7684\u8868\u73b0\uff0c\u8fd9\u8868\u660e\u6a21\u578b\u5728\u7ec6\u7c92\u5ea6\u7684\u52a8\u4f5c\u7406\u89e3\u4e0a\u5b58\u5728\u74f6\u9888\u3002\u6a21\u578b\u5728\u89c6\u9891\u957f\u5ea6\u589e\u52a0\u65f6\u7684\u8868\u73b0\u4e5f\u4e0d\u7406\u60f3\uff0c\u8fd9\u8bf4\u660e\u7406\u89e3\u957f\u89c6\u9891\u4e2d\u7684\u7ec6\u5fae\u65f6\u5e8f\u53d8\u5316\u4ecd\u662f\u4e00\u4e2a\u6781\u5177\u6311\u6218\u6027\u7684\u4efb\u52a1\u3002<\/p>\n\n\n\n<h5 class=\"wp-block-heading\">5.3 \u6a21\u578b\u5bf9\u4e0d\u540c\u7ec6\u7c92\u5ea6\u4efb\u52a1\u7684\u8868\u73b0<\/h5>\n\n\n\n<p>\u5728\u7ec6\u7c92\u5ea6\u4efb\u52a1\u7684\u8bc4\u4f30\u4e2d\uff0c\u6a21\u578b\u5728\u533a\u5206\u52a8\u4f5c\u987a\u5e8f\u3001\u9891\u7387\u4ee5\u53ca\u52a8\u4f5c\u65b9\u5411\u7b49\u65b9\u9762\u8868\u73b0\u8f83\u5f31\u3002\u7279\u522b\u662f\u5728\u6d89\u53ca\u52a8\u4f5c\u9891\u7387\uff08\u5982\u8bc6\u522b\u67d0\u4e2a\u52a8\u4f5c\u6267\u884c\u4e86\u51e0\u6b21\uff09\u7684\u95ee\u9898\u4e0a\uff0c\u73b0\u6709\u6a21\u578b\u5f80\u5f80\u65e0\u6cd5\u51c6\u786e\u533a\u5206\u4e0d\u540c\u9891\u6b21\u7684\u52a8\u4f5c\uff0c\u8868\u660e\u6a21\u578b\u7f3a\u4e4f\u5bf9\u91cd\u590d\u4e8b\u4ef6\u7684\u8bb0\u5fc6\u548c\u7406\u89e3\u80fd\u529b\u200b\u3002<\/p>\n\n\n\n<h4 class=\"wp-block-heading\">6. TemporalBench \u7684\u610f\u4e49\u4e0e\u672a\u6765\u5c55\u671b<\/h4>\n\n\n\n<p>TemporalBench\u4e3a\u591a\u6a21\u6001\u89c6\u9891\u7406\u89e3\u63d0\u4f9b\u4e86\u4e00\u4e2a\u66f4\u5177\u6311\u6218\u6027\u7684\u8bc4\u4f30\u57fa\u51c6\uff0c\u80fd\u591f\u66f4\u5168\u9762\u5730\u6d4b\u8bd5\u6a21\u578b\u5728\u7ec6\u7c92\u5ea6\u65f6\u5e8f\u7406\u89e3\u4e0a\u7684\u80fd\u529b\u3002\u901a\u8fc7\u8be6\u7ec6\u7684\u65f6\u5e8f\u6ce8\u91ca\uff0cTemporalBench\u5f25\u8865\u4e86\u73b0\u6709\u89c6\u9891\u57fa\u51c6\u5728\u65f6\u5e8f\u52a8\u6001\u8bc4\u4f30\u4e0a\u7684\u4e0d\u8db3\uff0c\u7279\u522b\u662f\u5b83\u80fd\u6709\u6548\u533a\u5206\u6a21\u578b\u662f\u5426\u771f\u6b63\u7406\u89e3\u4e86\u89c6\u9891\u4e2d\u7684\u65f6\u5e8f\u5173\u7cfb\uff0c\u800c\u4e0d\u4ec5\u4ec5\u662f\u901a\u8fc7\u8bed\u8a00\u6a21\u5f0f\u6765\u731c\u6d4b\u7b54\u6848\u3002<\/p>\n\n\n\n<p>\u8bba\u6587\u7684\u4f5c\u8005\u5e0c\u671bTemporalBench\u80fd\u63a8\u52a8\u591a\u6a21\u6001\u89c6\u9891\u6a21\u578b\u5728\u65f6\u5e8f\u63a8\u7406\u80fd\u529b\u4e0a\u7684\u8fdb\u4e00\u6b65\u53d1\u5c55\uff0c\u5305\u62ec\u66f4\u597d\u5730\u6355\u6349\u89c6\u9891\u4e2d\u7684\u65f6\u5e8f\u4f9d\u8d56\u5173\u7cfb\u3001\u6539\u8fdb\u5bf9\u957f\u89c6\u9891\u7684\u7406\u89e3\u80fd\u529b\u7b49\u3002\u6b64\u5916\uff0cTemporalBench\u8fd8\u53ef\u4ee5\u7528\u4e8e\u5176\u4ed6\u57fa\u7840\u6027\u89c6\u9891\u4efb\u52a1\u7684\u7814\u7a76\uff0c\u5982\u7ec6\u7c92\u5ea6\u65f6\u7a7a\u5b9a\u4f4d\u548c\u57fa\u4e8e\u8be6\u7ec6\u63d0\u793a\u7684\u6587\u672c\u5230\u89c6\u9891\u751f\u6210\uff08Text-to-Video Generation\uff09\u200b\u3002<\/p>\n\n\n\n<hr class=\"wp-block-separator has-alpha-channel-opacity\"\/>\n\n\n\n<p>TemporalBench on GitHub: <a href=\"https:\/\/temporalbench.github.io\/\" target=\"_blank\" rel=\"noreferrer noopener\">https:\/\/temporalbench.github.io\/<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u8bba\u6587TemporalBench: Benchmarking Fine-grained Temporal Und [&hellip;]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[8],"tags":[39,90,80],"class_list":["post-5160","post","type-post","status-publish","format-standard","hentry","category-tech","tag-ai","tag-multimodal","tag-cv"],"rttpg_featured_image_url":null,"rttpg_author":{"display_name":"NullThought","author_link":"https:\/\/nullthought.net\/?author=1"},"rttpg_comment":0,"rttpg_category":"<a href=\"https:\/\/nullthought.net\/?cat=8\" rel=\"category\">Tech<\/a>","rttpg_excerpt":"\u8bba\u6587TemporalBench: Benchmarking Fine-grained Temporal Und&hellip;","_links":{"self":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5160","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5160"}],"version-history":[{"count":1,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5160\/revisions"}],"predecessor-version":[{"id":5162,"href":"https:\/\/nullthought.net\/index.php?rest_route=\/wp\/v2\/posts\/5160\/revisions\/5162"}],"wp:attachment":[{"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5160"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5160"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/nullthought.net\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5160"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}