Large Language Models are Miscalibrated In-Context Learners (ACL 2025 Findings) - View it on GitHub