English
全部
图片
视频
地图
资讯
购物
更多
航班
旅游
酒店
搜索
笔记本
Top stories
Sports
U.S.
Local
World
Science
Technology
Entertainment
Business
More
Politics
时间不限
过去 1 小时
过去 24 小时
过去 7 天
过去 30 天
按相关度排序
按时间排序
16 天
腾讯Hunyuan团队提出MM-IQ基准:多模态模型推理能力的新标准
在人类的认知能力评估中,智商测试(IQ Test)被广泛用作衡量抽象推理和逻辑思维的标准。这类测试通过剥离语言背景、特定领域知识等,更加专注于人类的核心认知能力。然而,在人工智能领域,尤其是在多模态系统当中,目前却缺乏这样一个能够全面量化这些关键认知维度的评估标准。虽然现有的多模态模型在特定任务如OCR、目标定位以及医学图像分析等领域表现优异,但它们的评估指标并未涵盖核心推理能力。
一些您可能无法访问的结果已被隐去。
显示无法访问的结果
今日热点
Sworn in as Canada's PM
Child abuse images sentence
Texas measles outbreak
Strikes deal with DOGE
Syed formally resentenced
March megastorm
Man hit by motorcade, dies
Plane engine catches fire
Coffee creamer recall
Top intelligence agency visit
Hamas to release hostage
US expels SA ambassador
Court lifts Trump's DEI block
Felony gun possession arrest
Permanently shuts route
Influencer leaves Australia
Consumer sentiment drops
Second protester arrested
Severe weather outbreak
‘Ted Lasso’ is coming back
Gold rises to new heights
Senate confirmation hearing
ISIS leader killed in Iraq
Influential WY senator dies
To cut 2,000+ jobs
Legendary sportswriter dies
Pleads not guilty
Cuba suffers power outage
Wins longest-ever Iditarod
Laceration hazard recall
Kupp signs with Seahawks
Boehler withdraws nomination
反馈