English
全部
搜索
图片
视频
地图
资讯
Copilot
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
生物通
3 个月
评估大语言模型(LLMs)在可解释的深度强化学习(explainable deep ...
本文评估了CoT、MCTS增强和SFT三种方法在生成强化学习解释中的效果,发现MCTS显著提升大模型在复杂环境(如Lunar Lander)的解释质量,而SFT对中小模型更有效。通过LLMs作为评判者,验证了自动化评估框架与人工评估高度一致(Cohen's κ=0.77,Spearman ρ=0.88)。
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
Judge blocks subpoenas
6 US service members killed
Trump: US attacks Kharg Island
Race data demand blocked
NK fires missiles toward sea
Charges dismissed for teens
Ex-NY trooper found guilty
DOJ drops prosecution
Faces 3 felony charges
Won’t run for US Senate
WI legislator pleads guilty
Blast rocks Tehran
Brazil's ex-president in ICU
Breaks 63-yr-old NBA record
SLU coach agrees to extension
Cuba confirms talks w/ US
Top DEA fugitive captured
Shooter released from prison
EPA to ease pollution limits
Russian attack on Kyiv region
Adobe to settle US lawsuit
Kennedy Center head to exit
Taiwan OKs US arms deal
US job openings rise
Iconic NY news anchor dies
MA ICE report portal launched
UVA hit with bomb threat
US reaches WBC semifinals
Los Angeles asks for probe
Trump endorses Hern
NBA’s Silver visits Portland
Epstein jail guard to testify
反馈