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
DOJ drops probe into Powell
Soldier held over raid bet
Israel-Lebanon ceasefire
China to send pandas to US
Trump plans pool makeover
Arrest over mass shooting plot
Canada C‑3 citizenship law
Newsmax case moved to FL
2 Chinese nationals charged
Manager charged over fraud
Probes missing scientists
Mall of Louisiana shooting
Former Vikings star dies
Unveils deal with Regeneron
Venice Film Festival jury pres
To lead US Ryder Cup team?
Brings back cereal box toys
Synagogue attack plot arrests
Heated socks recalled
Chiefs assistant coach charged
Tornado hits city in Oklahoma
Urges US, Iran to resume talks
Lindor placed on injured list
DOJ watchdog launches probe
Boosts spending plan to $25B
OK’s $106B loan to help UKR
To lay off 10% of employees
Defamation claims dismissed
Walker released by Phillies
Returns to Thrive Capital
3rd US carrier arrives in ME
Joining Raiders w/ No. 1 pick
反馈