2 papers
cs.CL2026
Controlling the Diversity-Quality Trade-Off of LLMs by Merging Models Before and After Alignment
Yasuaki Sumita, Tomoharu Iwata, Toshiyuki Tanaka
Alignment is known to change the output characteristics of large language models (LLMs). Base models tend to generate more diverse outputs, whereas aligned models tend to produce h…
cs.CL2024
Cognitive Biases in Large Language Models: A Survey and Mitigation Experiments
Yasuaki Sumita, Koh Takeuchi, Hisashi Kashima
Large Language Models (LLMs) are trained on large corpora written by humans and demonstrate high performance on various tasks. However, as humans are susceptible to cognitive biase…