实体
HaluEval
HaluEval
PulseAugur coverage of HaluEval — every cluster mentioning HaluEval across labs, papers, and developer communities, ranked by signal.
总计 · 30天
2
90 天内 2
发布 · 30天
0
90 天内 0
论文 · 30天
2
90 天内 2
层级分布 · 90 天
最近 · 第 1/1 页 · 共 2 条
-
New research reveals limits of spectral diagnostics in understanding LLM hallucinations
Researchers have developed a new diagnostic framework to understand how large language models hallucinate by analyzing their self-attention mechanisms. The proposed method, which focuses on the "transport" properties of…
-
New framework uses multiple LLMs to reduce hallucination and bias
Researchers have developed a new framework called Council Mode designed to mitigate hallucinations and biases in Large Language Models. This approach involves querying multiple diverse LLMs simultaneously and then synth…