Counting papers with at most 25 authors, so collaboration author lists do not stand in for the people here.
1 paper
Akira Okutomi
High-confidence errors in large language models are often treated as evidence of fragile internal inference. We study a different possibility: stable miscalibration, where a confid…