#Rubricon

1 stories · 1 in 7 days · 1 in 30 days

HealthBench 1 LLM评估 1 Qwen3.5 1 RLHF 1

Stories, products and papers tagged "Rubricon", newest first.

Oct 30

Papers Builders10:01

Chinese contentRubricon：自动生成与精炼评分标准，用于 LLM 评估与训练

训练和评估没有标准答案的任务时，人工写 rubric 很贵，这篇论文给出全自动生成方案，对齐度比现有方法高 9%

#Rubricon #LLM评估 #Qwen3.5 #HealthBench

arXiv cs.AI@Ulyana Tkachenko et al. (4)Source