Paper 'Uncertainty as Feature Gaps: Epistemic Uncertainty Quantification of LLMs in Contextual Question-Answering' accepted to NeurIPS ReliableML Workshop (2025)
Paper 'Reject Only Critical Tokens: Pivot-Aware Speculative Decoding' accepted to NeurIPS Efficient Reasoning Workshop (2025)
Open-source library TruthTorchLM accepted to EMNLP 2025 (2025)
Paper 'Reconsidering Reconsidering LLM Uncertainty Estimation Methods in the Wild' accepted to ACL 2025 (2025)
Paper 'Un-considering Contextual Information: Assessing LLMs’ Understanding of Indexical Elements' accepted to ACL 2025 Findings (2025)
Paper 'CroMo-Mixup: Augmenting Cross-Model Representations for Continual Self-Supervised Learning' accepted to ECCV 2024 (2024)
Paper 'Do LLMs Recognize me, When I is not me' accepted to ACL Turkic Languages 2024 Workshop (2024)
Paper 'MARS: Meaning-Aware Response Scoring for Uncertainty Estimation in Generative LLMs' accepted to ACL 2024 (2024)
First-author paper 'Federated Orthogonal Training: Mitigating Global Catastrophic Forgetting in Continual Federated Learning' accepted to ICLR 2024 (2024)
Officially released TruthTorchLM (May 2025), an open-source library for assessing LLM truthfulness