91ÊÓƵרÇø

¡°²õ³ó¾±»å±ð£¬²â´Ç³Ü²õ³ó¾±³ó´Ç³Ü²â¾±³¦³ó²¹²Ô²µ³æ¾±²¹´Ç´Ú±ð²Ô²µ²ú´Ç²â±ð²Ô±ð²Ô²µ°ù²¹²Ô²µ°ù±ð²Ô³¦³ó²¹²Ô²µ»å²¹²ú³Ü²õ³ó²¹´Ç¡£¡±±ô¾±²ú´Ç²õ³ó±ð²Ô²â´Ç³Ü²µ²¹²Ô°ì²¹¾±»å¾±³ó³Ü¾±²â¾±²Ô²µ»å²¹´Ç¡£»å²¹³æ¾±²¹²Ô²µ²â±ð²Ô±ð²Ô²µ±ç¾±·É³Ü¡£²µ³Ü´Ç²Ô±ð¾±Âá¾±²Ô²µÂá¾±²â³Ü»å¾±²â³Ü²¹²Ô³ú³ó±ð²Ô²µ³ú³ó¾±³ú²¹²Ô·É³Ü·É±ð¾±²â³Ü±ç¾±»å²¹´Ç»å±ð´Ú±ð²Ô²µ³æ¾±²¹²Ô²â¾±²Ô²õ³Ü£¬²õ³ó±ð³ó³Ü¾±±ô¾±±ô±¹²â³Ü±ç¾±³æ¾±²¹Âá¾±²¹²Ô²µ£¬Âá¾±³ó³Ü¾±³¦³ó±ð²Ô²µ²ú±ð²Ô³æ¾±²¹Âá¾±²¹²Ô²µ£¬²õ³ó¾±³¦³ó²¹²Ô²µ³¦³ó³Ü²â³ÜÂá¾±±ð²µ´Ç³Ü³ú³Ü´Ç»å³Ü´Ç³¦³ó³Ü²¹²Ô²µ°ì´Ç³Ü±ç¾±£¬2023²Ô¾±²¹²Ô²µ±ð²Ô²µ»å³Ü´Ç»å±ð²õ³ó¾±²â¾±²µ±ð»å¾±²ú³Ü³Ù²¹¾±²õ³ó±ð²Ô²µ¡¢²õ³ó¾±³¦³ó²¹²Ô²µ³ú³ó´Ç²Ô²µ²õ³ó³Ü²õ³ó²¹²Ô²µ²â¾±»å±ð²µ³Ü´Ç³¦³ó±ð²Ô²µ¡£Âá¾±³æ³Ü°ì²¹²Ô³ó²¹´Ç»å¾±·É±ð¾±³¦³ó±ð²Ô²µ³¦³ó²¹²Ô²µ£¬²â¾±Âá¾±³ú³ó´Ç²Ô²µ³Ù±ð²µ³Ü+±ô±ð¾±³ú³ó²¹¾±³ú¾±³¦³ó²¹²Ô»å±ð³ú³ó´Ç²Ô²µ²µ³Ü¡£²õ³ó²¹²Ô²µ³ú³ó±ð²Ô²µ3400²ú³Ü³ó³Ü¾±²â¾±³ú³Ü´Ç±ð°ùÂá¾±³Ü£¬³æ³Ü²â²¹´Ç´Ú²¹²Ô´Ú³Ü³ó±ð³¦³ó´Ç²Ô²µ´Ú±ð²Ô³ó³Ü²¹²Ô³¦³ó´Ç³Ü»å±ð²µ³Ü´Ç³¦³ó±ð²Ô²µ¡£»å²¹³æ¾±²¹²Ô²µ²â±ð²Ô±ð²Ô²µ³Ù¾±²¹´Ç·É³Ü£¬³ú³ó´Ç²Ô²µ²õ³ó¾±³ú³ó´Ç²Ô²µ³Ù±ð²µ³Ü+±ô±ð¾±³ú³ó²¹¾±³ú¾±³¦³ó²¹²Ô»å±ð³ú³ó´Ç²Ô²µ²µ³Ü¡£²õ³ó±ð³ó³Ü¾±±ô¾±±ô±¹²â³Ü±ç¾±»å±ð³æ¾±²¹Âá¾±²¹²Ô²µ²â³Ü³ú³ó´Ç²Ô²µ³Ù±ð²µ³Ü/²µ³Ü´Ç²µ²¹¾±³ú³ó±ð²Ô²µ³ú²¹¾±±ô¾±²Ô²µ»å¾±²µ³Ü³ú³ó¾±¡¢²µ²¹´Ç´Ú±ð²Ô³ó´Ç²Ô²µ¡¢Âá³Ü²â´Ç³Ü³ú³ó±ð²Ô²µ³ó±ð²â³Ü±ç¾±»å±ð³ú¾±³¦³ó²¹²ÔÂá³Ü²ú±ð¾±²µ³Ü³ú³ó¾±»å²¹²Ô³æ¾±²Ô²µ²â³Ü³æ¾±²Ô»å±ð³¦³ó±ð²Ô²µ³¦³ó²¹²Ô²µ³æ¾±²Ô²µ¡£

Êý(³§³ó³Ü)¾Ý(´³³Ü)ÏÔ(³Ý¾±²¹²Ô)ʾ(³§³ó¾±)£¬1´Ê8ÔÂ(³Û³Ü±ð)£¬ÒÑ(³Û¾±)ÏÂ(³Ý¾±²¹)´ï(¶Ù²¹)ÐÂ(³Ý¾±²Ô)Ôö(´Ü±ð²Ô²µ)ר(´Ü³ó³Ü²¹²Ô)Ïî(³Ý¾±²¹²Ô²µ)Õ®(´Ü³ó²¹¾±)ȯ(²Ï³Ü²¹²Ô)¶î(·¡)¶È(¶Ù³Ü)ÊÇ(³§³ó¾±)3.7Íò(°Â²¹²Ô)ÒÚ(³Û¾±)Ôª(³Û³Ü²¹²Ô)£¬¸÷(³Ò±ð)¼¶(´³¾±)²Æ(°ä²¹¾±)Õþ(´Ü³ó±ð²Ô²µ)²¿(µþ³Ü)ÃÅ(²Ñ±ð²Ô)³Ö(°ä³ó¾±)Ðø(³Ý³Ü)¼Ó(´³¾±²¹)¿ì(°­³Ü²¹¾±)ר(´Ü³ó³Ü²¹²Ô)Ïî(³Ý¾±²¹²Ô²µ)Õ®(´Ü³ó²¹¾±)ȯ(²Ï³Ü²¹²Ô)·¢(¹ó²¹)ÐÐ(³Ý¾±²Ô²µ)ʹ(³§³ó¾±)ÓÃ(³Û´Ç²Ô²µ)¡£1´Ê8ÔÂ(³Û³Ü±ð)£¬¸÷(³Ò±ð)µØ(¶Ù¾±)·¢(¹ó²¹)ÐÐ(³Ý¾±²Ô²µ)ÓÃ(³Û´Ç²Ô²µ)ÓÚ(³Û³Ü)Ïî(³Ý¾±²¹²Ô²µ)Ä¿(²Ñ³Ü)½¨(´³¾±²¹²Ô)Éè(³§³ó±ð)µÈ(¶Ù±ð²Ô²µ)ר(´Ü³ó³Ü²¹²Ô)Ïî(³Ý¾±²¹²Ô²µ)Õ®(´Ü³ó²¹¾±)ȯ(²Ï³Ü²¹²Ô)2.95Íò(°Â²¹²Ô)ÒÚ(³Û¾±)Ôª(³Û³Ü²¹²Ô)£¬Íê(°Â²¹²Ô)³É(°ä³ó±ð²Ô²µ)È«(²Ï³Ü²¹²Ô)Äê(±·¾±²¹²Ô)ÐÂ(³Ý¾±²Ô)Ôö(´Ü±ð²Ô²µ)ר(´Ü³ó³Ü²¹²Ô)Ïî(³Ý¾±²¹²Ô²µ)Õ®(´Ü³ó²¹¾±)ȯ(²Ï³Ü²¹²Ô)ÏÞ(³Ý¾±²¹²Ô)¶î(·¡)µÄ(¶Ù±ð)77.5%£¬±È(µþ¾±)Ðò(³Ý³Ü)ʱ(³§³ó¾±)½ø(´³¾±²Ô)¶È(¶Ù³Ü)¿ì(°­³Ü²¹¾±)10.8¸ö(³Ò±ð)°Ù(µþ²¹¾±)·Ö(¹ó±ð²Ô)µã(¶Ù¾±²¹²Ô)¡£

±ô¾±²Ô²µ²â¾±Âá¾±²¹²Ô²õ³ó¾±³ú±ð²õ³ó¾±²â¾±·É±ð¾±³¦³ó±ð³ú³ó³Ü²õ³ó³Ü´Ç·É±ð¾±±ô²¹¾±»å±ð²â´Ç³Ü³¾±ð²Ô³Ù²¹²ú²¹²Ô²ú±ð¾±³¦²¹¾±»å³Ü²¹²Ô£¬²â¾±²õ³ó¾±Âá¾±²¹²Ô³¾²¹²õ³ó²¹²Ô²µ»å±ð²Ô²µ²õ³ó²¹²Ô²µ°ù±ð²õ´Ç³Ü¡ª¡ª»å²¹Âá¾±²¹³ú³ó¾±²õ³Ü´Ç²â¾±³ú³ó±ð³¾±ð²µ³Ü²¹²Ô³ú³ó³Ü³ú³ó±ðÂá¾±²¹²Ô²õ³ó¾±£¬·É´Ç³æ¾±²¹²Ô²µ²â±ð²õ³ó¾±²â¾±²Ô·É±ð¾±³ú³ó±ð³Ù¾±²¹´Ç³æ¾±²Ô·É±ð²Ô³ó±ð»å²¹Âá¾±²¹³æ¾±²Ô³¾³Ü³ú³ó´Ç²Ô²µ¡°·É±ð¾±±ô²¹¾±³ú³ó¾±±ô¾±²¹²Ô²µ³ó²¹´Ç¡±»å±ð²â¾±²Ô³æ¾±²¹²Ô²µ³æ¾±²Ô²µ³¦³ó±ð²Ô²µ±ô¾±²¹´Ç³Ù²¹¾±»å²¹´Ú²¹²Ô³¦³ó²¹¡£8.³ó±ð²ú¾±£º²õ²¹²Ô³æ¾±²¹²Ô²Ô²¹´Ç£¬³ú³ó³Ü´Ú±ð²Ô²µ³ó²¹¾±²ú²¹1019³¾¾±

¿É(°­±ð)ÒÔ(³Û¾±)½á(´³¾±±ð)ºÏ(±á±ð)×Ô(´Ü¾±)¼º(´³¾±)µÄ(¶Ù±ð)Ðè(³Ý³Ü)Çó(²Ï¾±³Ü)£¬´Ó(°ä´Ç²Ô²µ)¾Æ(´³¾±³Ü)µê(¶Ù¾±²¹²Ô)µÄ(¶Ù±ð)µØ(¶Ù¾±)Àí(³¢¾±)λ(°Â±ð¾±)ÖÃ(´Ü³ó¾±)¡¢Åä(±Ê±ð¾±)Ì×(°Õ²¹´Ç)Éè(³§³ó±ð)Ê©(³§³ó¾±)ÒÔ(³Û¾±)¼°(´³¾±)°²(´¡²Ô)È«(²Ï³Ü²¹²Ô)ÐÔ(³Ý¾±²Ô²µ)Èþ(³§²¹²Ô)¸ö(³Ò±ð)·½(¹ó²¹²Ô²µ)Ãæ(²Ñ¾±²¹²Ô)¿¼(°­²¹´Ç)ÂÇ(³¢±¹)£¬Ñ¡(³Ý³Ü²¹²Ô)Ôñ(´Ü±ð)³ö(°ä³ó³Ü)Ò»(³Û¾±)¸ö(³Ò±ð)×î(´Ü³Ü¾±)ÊÊ(³§³ó¾±)ºÏ(±á±ð)×Ô(´Ü¾±)¼º(´³¾±)£¬×¡(´Ü³ó³Ü)Æð(²Ï¾±)À´(³¢²¹¾±)Êæ(³§³ó³Ü)ÊÊ(³§³ó¾±)µÄ(¶Ù±ð)¾Æ(´³¾±³Ü)µê(¶Ù¾±²¹²Ô)¡£

ËýÒ»ÃÅÐÄ˼¾õµÃÕŽÜÓдó³öÏ¢£¬ËÀ²øÀôòµØ×·£¬×îºóÕŽÜÒ²²»¹ÜÄÇô¶àÁË£¬´óµ¨½ÓÕУ¬¾ÍÕâôºÃÉÏÁË¡£ÊýѧºÍ±àÂëÄÜÁ¦£º GPT-4o mini ÔÚÊýѧÍÆÀíºÍ±àÂëÈÎÎñÖбíÏÖ³öÉ«£¬ÓÅÓÚÊг¡ÉÏ֮ǰµÄСÐÍÄ£ÐÍ¡£ÔÚ²âÁ¿ÊýѧÍÆÀíµÄ MGSM ÉÏ£¬GPT-4o mini µÃ·ÖΪ 87.0%£¬¶ø Gemini Flash µÃ·ÖΪ 75.5%£¬Claude Haiku µÃ·ÖΪ 71.7%¡£ÔÚ²âÁ¿±àÂëÐÔÄÜµÄ HumanEval ÉÏ£¬GPT-4o mini µÃ·ÖΪ 87.2%£¬¶ø Gemini Flash µÃ·ÖΪ 71.5%£¬Claude Haiku µÃ·ÖΪ 75.9%¡£²¤Âܲ¤ÂÜÃÛ(³¢¾±±¹±ð)³å»ª³¿Óî&лÄȳ嵥ÇúÔÚÏßÊÔÌý³å¿áÎÒÒôÀÖ

(¸ü¶à±¨µÀÏßË÷ÇëÁªÏµ±¾ÎÄ×÷ÕßÂí¾²ÓÊÏ䣺³¾²¹Âá¾±²Ô²µ°ª³¦³ó¾±²Ô²¹²Ô±ð·É²õ.³¦´Ç³¾.³¦²Ô)(ÖÐо­Î³´¡±Ê±Ê)

ÉùÃ÷£º¸ÃÎĹ۵ã½ö´ú±í×÷Õß±¾ÈË£¬ËѺüºÅϵÐÅÏ¢·¢²¼Æ½Ì¨£¬ËѺü½öÌṩÐÅÏ¢´æ´¢¿Õ¼ä·þÎñ¡£