一作ACL 2026Findings· Findings of ACL
Large Language Models Are Overconfident in Their Own Responses
Mario Sanz-Guerrero · Manuel Mager · Katharina von der Wense
摘要
Prior work has shown that instruction-tuned large language models (LLMs) are less well calibrated than their base pre-trained counterparts. However, little is known about the frequently used chat template’s effect on the calibration of conversational LLMs. In this work, we investigate the mechani...