English
全部
搜索
图片
视频
地图
资讯
更多
购物
航班
旅游
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
最佳匹配
最新
新浪网
1 个月
腾讯纯文本LLM训视觉encoder,拿捏图表长视频,达到开源小模型SOTA!
打破多模态视觉+语言拼接套路! 腾讯开源Penguin-VL,直接用纯文本LLM训视觉编码器。 这项研究跳出了先有传统视觉 backbone,再接语言模型的常规路径,直接从text-only LLM初始化vision encoder。 并在2B/8B紧凑参数规模下的文档理解、长视频时序定位等复杂任务中表现出 ...
当前正在显示可能无法访问的结果。
隐藏无法访问的结果
今日热点
Louisiana mass shooting
Lead prosecutor removed?
Flight evacuated in Pittsburgh
US, Cuban officials met
Launches ballistic missiles
‘Downton Abbey’ star dies
Obama, Mamdani meet in NYC
Today in history: 1897
Meets WH chief of staff
Rejects merger talks w/ United
Sabah fire razes 200 homes
Charged with insurance fraud
Applegate hospitalized in LA
DOJ to target egg producers
Tentative deal averts strike
Basketball Hall of Famer dies
2 soldiers hurt in bear attack
Federal judge blocks merger
Sues CA prison system
Blocked in RI voter data bid
Giants, Bengals agree to trade
US renews Russian oil waiver
Robot wins half-marathon
Teases UFO file release
Shooting near Iowa campus
Iran fires on tankers
Trump signs drug review order
Shooting in Kyiv
Pope Leo: No Trump debate
Skydiver hits VA scoreboard
Titanic life jacket auctioned
Murphy receives AFI honor
反馈