仓库议题
changkun/trustcalib
Trust calibration for agentic tool use as preference learning: a GP-probit allow/ask/block policy gateway framed as Preferential Bayesian Optimization, with the paper and a reproducible simulation.
议题
此仓库没有开放的已索引议题。
仓库议题
Trust calibration for agentic tool use as preference learning: a GP-probit allow/ask/block policy gateway framed as Preferential Bayesian Optimization, with the paper and a reproducible simulation.
此仓库没有开放的已索引议题。