paper-conference

Reject Only Critical Tokens: Pivot-Aware Speculative Decoding
Speculative decoding enforces exact target-distribution matching, which can unnecessarily reduce token acceptance and limit speedups. …
Amir Ziashahabi, Yavuz Faruk Bakman, Duygu Nur Yaldiz, Mostafa El-Khamy, Sai Praneeth Karimireddy, Salman Avestimehr
PDF Code Source Document
TruthTorchLM: A Comprehensive Library for Predicting Truthfulness in LLM Outputs
TruthTorchLM is an open-source library for estimating the truthfulness of LLM outputs. It unifies over 30 truthfulness methods spanning …
Duygu Nur Yaldiz, Yavuz Faruk Bakman, Sungmin Kang, Alperen Ozis, Hayrettin Eren Yildiz, Mitash Ashish Shah, Zhiqi Huang, Anoop Kumar, Alfy Samuel, Daben Liu, Sai Praneeth Karimireddy, Salman Avestimehr
PDF Code Source Document DOI
Uncertainty as Feature Gaps: Epistemic Uncertainty Quantification of LLMs in Contextual Question-Answering
We study uncertainty quantification for contextual question answering and propose a principled epistemic uncertainty measure derived …
Yavuz Faruk Bakman, Sungmin Kang, Zhiqi Huang, Duygu Nur Yaldiz, Catarina Belem, Chenyang Zhu, Anoop Kumar, Alfy Samuel, Daben Liu, Sai Praneeth Karimireddy, Salman Avestimehr
PDF Source Document
Reconsidering LLM Uncertainty Estimation Methods in the Wild
This paper studies practical deployment challenges for LLM uncertainty estimation beyond standard short-form QA evaluation. It analyzes …
Yavuz Faruk Bakman, Duygu Nur Yaldiz, Sungmin Kang, Tuo Zhang, Baturalp Buyukates, Salman Avestimehr, Sai Praneeth Karimireddy
PDF Code Source Document DOI
Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs
In this work, we introduce the Learnable Response Scoring Function (LARS) for Uncertainty Estimation (UE) in generative Large Language …
Duygu Nur Yaldiz, Yavuz Faruk Bakman, Baturalp Buyukates, Anil Ramakrishna, Chenyang Tao, Dimitrios Dimitriadis, Jieyu Zhao, Salman Avestimehr
PDF Cite
Do Not Design, Learn: A Trainable Scoring Function for Uncertainty Estimation in Generative LLMs
CroMo-Mixup: Augmenting Cross-Model Representations for Continual Self-Supervised Learning
Continual self-supervised learning (CSSL) learns a series of tasks sequentially on the unlabeled data. Two main challenges of con- …
Erum Mushtaq, Duygu Nur Yaldiz, Yavuz Faruk Bakman, J Ding, Chenyang Tao, Dimitrios Dimitriadis, Salman Avestimehr
PDF Cite Code
CroMo-Mixup: Augmenting Cross-Model Representations for Continual Self-Supervised Learning
Do LLMs Recognize me, When I is not me: Assessment of LLMs Understanding of Turkish Indexical Pronouns in Indexical Shift Contexts
In this work, we introduce the Learnable Response Scoring Function (LARS) for Uncertainty Estimation (UE) in generative Large Language …
Metehan Oguz, Y. Umut Ciftci, Yavuz Faruk Bakman
PDF Cite
Do LLMs Recognize me, When I is not me: Assessment of LLMs Understanding of Turkish Indexical Pronouns in Indexical Shift Contexts
MARS: Meaning-Aware Response Scoring for Uncertainty Estimation in Generative LLMs
Generative Large Language Models (LLMs) are widely utilized for their excellence in vari- ous tasks. However, their tendency to produce …
Yavuz Faruk Bakman, Duygu Nur Yaldiz, Baturalp Buyukates, Chenyang Tao, Dimitrios Dimitriadis, Salman Avestimehr
PDF Cite Code
MARS: Meaning-Aware Response Scoring for Uncertainty Estimation in Generative LLMs
Federated Orthogonal Training: Mitigating Global Catastrophic Forgetting in Continual Federated Learning
Current literature in FL mostly focuses on single-task learning. However, over time, new tasks may appear in the clients and the global …
Yavuz Faruk Bakman, Duygu Nur Yaldiz, Yahya H. Ezzeldin, Salman Avestimehr
PDF Cite Code
Federated Orthogonal Training: Mitigating Global Catastrophic Forgetting in Continual Federated Learning
Federated Alternate Training (FAT): Leveraging Unannotated Data Silos in Federated Segmentation for Medical Imaging
Federated Learning (FL) aims to train a machine learning (ML) model in a distributed fashion to strengthen data privacy with limited …
Erum Mushtaq, Yavuz Faruk Bakman, Jie Ding, Salman Avestimehr
PDF Cite DOI
Federated Alternate Training (FAT): Leveraging Unannotated Data Silos in Federated Segmentation for Medical Imaging