Hair-Trigger Alignment: Black-Box Evaluation Cannot Guarantee Post-Update Alignment.

Yavuz Faruk Bakman

Yavuz Faruk Bakman Duygu Nur Yaldiz Salman Avestimehr Sai Praneeth Karimireddy Hair-Trigger Alignment: Black-Box Evaluation Cannot Guarantee Post-Update Alignment. 2026 January abs/2601.22313 CoRR https://doi.org/10.48550/arXiv.2601.22313 db/journals/corr/corr2601.html#abs-2601-22313 streams/journals/corr

Metehan Oguz Yavuz Faruk Bakman Duygu Nur Yaldiz Un-considering Contextual Information: Assessing LLMs' Understanding of Indexical Elements. 23410-23427 2025 ACL (Findings) https://doi.org/10.18653/v1/2025.findings-acl.1203 https://aclanthology.org/2025.findings-acl.1203/ conf/acl/2025f db/conf/acl/acl2025f.html#OguzBY25 Yavuz Faruk Bakman Duygu Nur Yaldiz Sungmin Kang Tuo Zhang Baturalp Buyukates Salman Avestimehr Sai Praneeth Karimireddy Reconsidering LLM Uncertainty Estimation Methods in the Wild. 29531-29556 2025 ACL (1) https://doi.org/10.18653/v1/2025.acl-long.1429 https://aclanthology.org/2025.acl-long.1429/ conf/acl/2025-1 db/conf/acl/acl2025-1.html#BakmanYKZBAK25 Erum Mushtaq Zalan Fabian Yavuz Faruk Bakman Anil Ramakrishna Mahdi Soltanolkotabi Salman Avestimehr HARMONY: Hidden Activation Representations and Model Output-Aware Uncertainty Estimation for Vision-Language Models. 1663-1668 2025 CVPR Workshops https://openaccess.thecvf.com/content/CVPR2025W/TMM-OpenWorld/html/Mushtaq_HARMONY_Hidden_Activation_Representations_and_Model_Output-Aware_Uncertainty_Estimation_for_CVPRW_2025_paper.html conf/cvpr/2025w db/conf/cvpr/cvprw2025.html#MushtaqFBRSA25 Duygu Nur Yaldiz Yavuz Faruk Bakman Sungmin Kang Alperen Özis Hayrettin Eren Yildiz Mitash Ashish Shah Zhiqi Huang Anoop Kumar Alfy Samuel Daben Liu Sai Praneeth Karimireddy Salman Avestimehr TruthTorchLM: A Comprehensive Library for Predicting Truthfulness in LLM Outputs. 717-728 2025 EMNLP (System Demonstrations) https://doi.org/10.18653/v1/2025.emnlp-demos.54 conf/emnlp/2025d db/conf/emnlp/emnlp2025d.html#YaldizBKOYSHKSL25 Duygu Nur Yaldiz Yavuz Faruk Bakman Baturalp Buyukates Chenyang Tao Anil Ramakrishna Dimitrios Dimitriadis Jieyu Zhao Salman Avestimehr Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs. 691-713 2025 NAACL (Findings) https://doi.org/10.18653/v1/2025.findings-naacl.41 https://aclanthology.org/2025.findings-naacl.41/ conf/naacl/2025f db/conf/naacl/naacl2025f.html#YaldizBBTRDZA25 streams/conf/naacl

Metehan Oguz Yavuz Faruk Bakman Duygu Nur Yaldiz Un-considering Contextual Information: Assessing LLMs' Understanding of Indexical Elements. 2025 June abs/2506.01089 CoRR https://doi.org/10.48550/arXiv.2506.01089 db/journals/corr/corr2506.html#abs-2506-01089 streams/journals/corr

Yavuz Faruk Bakman Duygu Nur Yaldiz Sungmin Kang Tuo Zhang Baturalp Buyukates Salman Avestimehr Sai Praneeth Karimireddy Reconsidering LLM Uncertainty Estimation Methods in the Wild. 2025 June abs/2506.01114 CoRR https://doi.org/10.48550/arXiv.2506.01114 db/journals/corr/corr2506.html#abs-2506-01114 streams/journals/corr

Nien-Shao Wang Duygu Nur Yaldiz Yavuz Faruk Bakman Sai Praneeth Karimireddy Conformal Prediction Adaptive to Unknown Subpopulation Shifts. 2025 June abs/2506.05583 CoRR https://doi.org/10.48550/arXiv.2506.05583 db/journals/corr/corr2506.html#abs-2506-05583 streams/journals/corr

Duygu Nur Yaldiz Yavuz Faruk Bakman Sungmin Kang Alperen Özis Hayrettin Eren Yildiz Mitash Ashish Shah Zhiqi Huang Anoop Kumar Alfy Samuel Daben Liu Sai Praneeth Karimireddy Salman Avestimehr TruthTorchLM: A Comprehensive Library for Predicting Truthfulness in LLM Outputs. 2025 July abs/2507.08203 CoRR https://doi.org/10.48550/arXiv.2507.08203 db/journals/corr/corr2507.html#abs-2507-08203 streams/journals/corr

Yavuz Faruk Bakman Sungmin Kang Zhiqi Huang Duygu Nur Yaldiz Catarina G. Belém Chenyang Zhu 0011 Anoop Kumar Alfy Samuel Salman Avestimehr Daben Liu Sai Praneeth Karimireddy Uncertainty as Feature Gaps: Epistemic Uncertainty Quantification of LLMs in Contextual Question-Answering. 2025 October abs/2510.02671 CoRR https://doi.org/10.48550/arXiv.2510.02671 db/journals/corr/corr2510.html#abs-2510-02671 streams/journals/corr

Sungmin Kang Yavuz Faruk Bakman Duygu Nur Yaldiz Baturalp Buyukates Salman Avestimehr Uncertainty Quantification for Hallucination Detection in Large Language Models: Foundations, Methodology, and Future Directions. 2025 October abs/2510.12040 CoRR https://doi.org/10.48550/arXiv.2510.12040 db/journals/corr/corr2510.html#abs-2510-12040 streams/journals/corr

Erum Mushtaq Zalan Fabian Yavuz Faruk Bakman Anil Ramakrishna Mahdi Soltanolkotabi Salman Avestimehr HARMONY: Hidden Activation Representations and Model Output-Aware Uncertainty Estimation for Vision-Language Models. 2025 October abs/2510.22171 CoRR https://doi.org/10.48550/arXiv.2510.22171 db/journals/corr/corr2510.html#abs-2510-22171 streams/journals/corr

Amir Ziashahabi Yavuz Faruk Bakman Duygu Nur Yaldiz Mostafa El-Khamy Sai Praneeth Karimireddy Salman Avestimehr Reject Only Critical Tokens: Pivot-Aware Speculative Decoding. 2025 November abs/2511.00351 CoRR https://doi.org/10.48550/arXiv.2511.00351 db/journals/corr/corr2511.html#abs-2511-00351 streams/journals/corr

Yavuz Faruk Bakman Duygu Nur Yaldiz Baturalp Buyukates Chenyang Tao Dimitrios Dimitriadis Salman Avestimehr MARS: Meaning-Aware Response Scoring for Uncertainty Estimation in Generative LLMs. 7752-7767 2024 ACL (1) https://doi.org/10.18653/v1/2024.acl-long.419 https://aclanthology.org/2024.acl-long.419 https://www.wikidata.org/entity/Q131458085 conf/acl/2024-1 db/conf/acl/acl2024-1.html#BakmanYBTDA24 Erum Mushtaq Duygu Nur Yaldiz Yavuz Faruk Bakman Jie Ding 0002 Chenyang Tao Dimitrios Dimitriadis Salman Avestimehr CroMo-Mixup: Augmenting Cross-Model Representations for Continual Self-Supervised Learning. 311-328 2024 ECCV (80) https://doi.org/10.1007/978-3-031-72989-8_18 conf/eccv/2024-80 db/conf/eccv/eccv2024-80.html#MushtaqYBDTDA24 streams/conf/eccv Yavuz Faruk Bakman Duygu Nur Yaldiz Yahya H. Ezzeldin Salman Avestimehr Federated Orthogonal Training: Mitigating Global Catastrophic Forgetting in Continual Federated Learning. 2024 ICLR https://openreview.net/forum?id=nAs4LdaP9Y conf/iclr/2024 db/conf/iclr/iclr2024.html#BakmanYEA24 Yavuz Faruk Bakman Duygu Nur Yaldiz Baturalp Buyukates Salman Avestimehr Chenyang Tao Dimitrios Dimitriadis Predicting Uncertainty of Generative LLMs with MARS: Meaning-Aware Response Scoring. 2033-2037 2024 ISIT https://doi.org/10.1109/ISIT57864.2024.10619136 conf/isit/2024 db/conf/isit/isit2024.html#BakmanYBATD24

Yavuz Faruk Bakman Duygu Nur Yaldiz Baturalp Buyukates Chenyang Tao Dimitrios Dimitriadis Salman Avestimehr MARS: Meaning-Aware Response Scoring for Uncertainty Estimation in Generative LLMs. 2024 abs/2402.11756 CoRR https://doi.org/10.48550/arXiv.2402.11756 db/journals/corr/corr2402.html#abs-2402-11756

Metehan Oguz Yusuf Umut Ciftci Yavuz Faruk Bakman Do LLMs Recognize me, When I is not me: Assessment of LLMs Understanding of Turkish Indexical Pronouns in Indexical Shift Contexts. 2024 abs/2406.05569 CoRR https://doi.org/10.48550/arXiv.2406.05569 db/journals/corr/corr2406.html#abs-2406-05569

Duygu Nur Yaldiz Yavuz Faruk Bakman Baturalp Buyukates Chenyang Tao Anil Ramakrishna Dimitrios Dimitriadis Salman Avestimehr Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs. 2024 abs/2406.11278 CoRR https://doi.org/10.48550/arXiv.2406.11278 db/journals/corr/corr2406.html#abs-2406-11278 streams/journals/corr

Erum Mushtaq Duygu Nur Yaldiz Yavuz Faruk Bakman Jie Ding 0002 Chenyang Tao Dimitrios Dimitriadis Salman Avestimehr CroMo-Mixup: Augmenting Cross-Model Representations for Continual Self-Supervised Learning. 2024 abs/2407.12188 CoRR https://doi.org/10.48550/arXiv.2407.12188 db/journals/corr/corr2407.html#abs-2407-12188 streams/journals/corr

Erum Mushtaq Yavuz Faruk Bakman Jie Ding 0002 Salman Avestimehr Federated Alternate Training (Fat): Leveraging Unannotated Data Silos in Federated Segmentation for Medical Imaging. 1-5 2023 ISBI https://doi.org/10.1109/ISBI53787.2023.10230533 conf/isbi/2023 db/conf/isbi/isbi2023.html#MushtaqBDA23

Erum Mushtaq Yavuz Faruk Bakman Jie Ding 0002 Salman Avestimehr Federated Alternate Training (FAT): Leveraging Unannotated Data Silos in Federated Segmentation for Medical Imaging. 2023 abs/2304.09327 CoRR https://doi.org/10.48550/arXiv.2304.09327 db/journals/corr/corr2304.html#abs-2304-09327

Yavuz Faruk Bakman Duygu Nur Yaldiz Yahya H. Ezzeldin Salman Avestimehr Federated Orthogonal Training: Mitigating Global Catastrophic Forgetting in Continual Federated Learning. 2023 abs/2309.01289 CoRR https://doi.org/10.48550/arXiv.2309.01289 db/journals/corr/corr2309.html#abs-2309-01289

Amir Salman AvestimehrSalman Avestimehr Catarina G. Belém Baturalp Buyukates Yusuf Umut Ciftci Dimitrios Dimitriadis Jie Ding 0002 Mostafa El-Khamy Yahya H. Ezzeldin Zalan Fabian Zhiqi Huang Sungmin Kang Sai Praneeth Karimireddy Anoop Kumar Daben Liu Erum Mushtaq Metehan Oguz Alperen Özis Anil Ramakrishna Alfy Samuel Mitash Ashish Shah Mahdi Soltanolkotabi Chenyang Tao Nien-Shao Wang Duygu Nur Yaldiz Hayrettin Eren Yildiz Tuo Zhang Jieyu Zhao Chenyang Zhu 0011 Amir Ziashahabi