语言大模型 / LLM - arXivDaily 专题

2306.00721 2023-10-13 cs.SD cs.AI eess.AS 57%

UnDiff: Unsupervised Voice Restoration with Unconditional Diffusion Model

Anastasiia Iashchenko, Pavel Andreev, Ivan Shchekotov, Nicholas Babaev, Dmitry Vetrov

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.AI

Comments Accepted to Interspeech 2023

2309.03925 2023-09-11 q-bio.QM cs.LG 57%

Beyond attention: deriving biologically interpretable insights from weakly-supervised multiple-instance learning models

Willem Bonnaffé, CRUK ICGC Prostate Group, Freddie Hamdy, Yang Hu, Ian Mills, Jens Rittscher, Clare Verrill, Dan J. Woodcock

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2308.09850 2023-08-22 cs.LG cs.CR 57%

Backdoor Mitigation by Correcting the Distribution of Neural Activations

Xi Li, Zhen Xiang, David J. Miller, George Kesidis

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2307.00360 2023-08-16 cs.CL 57%

BatGPT: A Bidirectional Autoregessive Talker from Generative Pre-trained Transformer

Zuchao Li, Shitou Zhang, Hai Zhao, Yifei Yang, Dongjie Yang

专题命中后训练与偏好优化：language model(abstract)；分类 cs.CL

2308.04617 2023-08-10 cs.LG cs.CR 57%

Improved Activation Clipping for Universal Backdoor Mitigation and Test-Time Detection

Hang Wang, Zhen Xiang, David J. Miller, George Kesidis

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2307.15801 2023-08-03 cs.RO cs.AI 57%

Primitive Skill-based Robot Learning from Human Evaluative Feedback

Ayano Hiranaka, Minjune Hwang, Sharon Lee, Chen Wang, Li Fei-Fei, Jiajun Wu, Ruohan Zhang

专题命中后训练与偏好优化：RLHF(abstract)；分类 cs.AI

2302.02596 2023-06-27 cs.LG 57%

Ten Lessons We Have Learned in the New "Sparseland": A Short Handbook for Sparse Neural Network Researchers

Shiwei Liu, Zhangyang Wang

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2306.01800 2023-06-06 cs.CY cs.AI 57%

The ethical ambiguity of AI data enrichment: Measuring gaps in research ethics norms and practices

Will Hawkins, Brent Mittelstadt

专题命中后训练与偏好优化：RLHF(abstract)；分类 cs.AI

Comments 10 pages

Journal ref 2023 ACM Conference on Fairness, Accountability, and Transparency

2303.03320 2023-05-30 cs.LG cs.CR cs.DC 57%

Learning to Backdoor Federated Learning

Henger Li, Chen Wu, Sencun Zhu, Zizhan Zheng

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2304.12298 2023-04-25 cs.CR cs.AI 57%

BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT

Jiawen Shi, Yixin Liu, Pan Zhou, Lichao Sun

专题命中后训练与偏好优化：language model(abstract)；分类 cs.AI

Comments This paper is accepted as a poster in NDSS2023

2304.10707 2023-04-24 stat.ML cs.LG 57%

Persistently Trained, Diffusion-assisted Energy-based Models

Xinwei Zhang, Zhiqiang Tan, Zhijian Ou

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments main text 8 pages

2211.11417 2023-04-03 cs.CV cs.GR cs.LG 57%

DyNCA: Real-time Dynamic Texture Synthesis Using Neural Cellular Automata

Ehsan Pajouheshgar, Yitao Xu, Tong Zhang, Sabine Süsstrunk

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments Link to the demo: https://dynca.github.io/

2204.02471 2023-02-14 cs.RO cs.LG cs.SY eess.SY 57%

Configuration Path Control

Sergey Pankov

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments 12 pages, 3 figures, accepted for publication

Journal ref Int. J. Control Autom. Syst. 21, 306-317 (2023)

2212.10938 2022-12-22 cs.CL 57%

Critic-Guided Decoding for Controlled Text Generation

Minbeom Kim, Hwanhee Lee, Kang Min Yoo, Joonsuk Park, Hwaran Lee, Kyomin Jung

专题命中后训练与偏好优化：language model(abstract)；分类 cs.CL

Comments 11 pages, 6 figures

2211.11602 2022-11-22 cs.LG cs.HC cs.MA 57%

Improving Multimodal Interactive Agents with Reinforcement Learning from Human Feedback

Josh Abramson, Arun Ahuja, Federico Carnevale, Petko Georgiev, Alex Goldin, Alden Hung, Jessica Landon, Jirka Lhotka, Timothy Lillicrap, Alistair Muldal, George Powell, Adam Santoro, Guy Scully, Sanjana Srivastava, Tamara von Glehn, Greg Wayne, Nathaniel Wong, Chen Yan, Rui Zhu

专题命中后训练与偏好优化：RLHF(abstract)；分类 cs.LG

2211.06519 2022-11-15 cs.LG 57%

The Expertise Problem: Learning from Specialized Feedback

Oliver Daniels-Koch, Rachel Freedman

专题命中后训练与偏好优化：RLHF(abstract)；分类 cs.LG

Comments Accepted to the ML Safety Workshop, NeurIPS 2022

2209.02201 2022-09-07 cs.NE cs.LG 57%

What to Prune and What Not to Prune at Initialization

Maham Haroon

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2206.12963 2022-06-28 cs.CV cs.LG 57%

Self-Healing Robust Neural Networks via Closed-Loop Control

Zhuotong Chen, Qianxiao Li, Zheng Zhang

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments 48 pages, 5 figures

2104.13312 2022-06-22 cs.LG cs.CY 57%

Multi-fairness under class-imbalance

Arjun Roy, Vasileios Iosifidis, Eirini Ntoutsi

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2206.02126 2022-06-07 cs.LG 57%

Learning Dynamics and Generalization in Reinforcement Learning

Clare Lyle, Mark Rowland, Will Dabney, Marta Kwiatkowska, Yarin Gal

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2112.07673 2022-06-01 hep-ph cs.LG 57%

Machine learning a manifold

Sean Craven, Djuna Croon, Daniel Cutting, Rachel Houtz

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments 7 pages, 2 figures. Version published in PRD. (SC+RH) + DC^2 propose mape + epsilon^2

Journal ref Phys.Rev.D 105 (2022) 9, 096030

2204.07064 2022-04-15 cs.SD cs.LG eess.AS stat.ML 57%

Streamable Neural Audio Synthesis With Non-Causal Convolutions

Antoine Caillon, Philippe Esling

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2009.10054 2022-04-12 cs.CV cs.LG 57%

Regularizing Attention Networks for Anomaly Detection in Visual Question Answering

Doyup Lee, Yeongjae Cheon, Wook-Shin Han

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments 16 pages, 7 figures, Accepted by AAAI-21

2106.03761 2022-03-31 cs.CV cs.LG stat.ML 57%

FairCal: Fairness Calibration for Face Verification

Tiago Salvador, Stephanie Cairns, Vikram Voleti, Noah Marshall, Adam Oberman

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments Accepted at ICLR 2022

2104.08281 2022-01-05 physics.ao-ph cs.LG 57%

Controlled abstention neural networks for identifying skillful predictions for classification problems

Elizabeth A. Barnes, Randal J. Barnes

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments submitted to the Journal of Advances in Earth System Modeling. arXiv admin note: substantial text overlap with arXiv:2104.08236

2104.08236 2022-01-05 cs.LG physics.ao-ph 57%

Controlled abstention neural networks for identifying skillful predictions for regression problems

Elizabeth A. Barnes, Randal J. Barnes

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments submitted to the Journal of Advances of Earth System Modeling

2112.03350 2021-12-08 cs.CR cs.LG 57%

Test-Time Detection of Backdoor Triggers for Poisoned Deep Neural Networks

Xi Li, Zhen Xiang, David J. Miller, George Kesidis

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

2111.02529 2021-11-05 cs.LG 57%

Shift Happens: Adjusting Classifiers

Theodore James Thibault Heiser, Mari-Liis Allikivi, Meelis Kull

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments ECML PKDD 2019 conference paper, 16 pages

Journal ref ECML PKDD 2019. Lecture Notes in Computer Science, vol 11907. Springer, Cham (2020)

2111.01256 2021-11-03 cs.LG 57%

Reverse engineering recurrent neural networks with Jacobian switching linear dynamical systems

Jimmy T. H. Smith, Scott W. Linderman, David Sussillo

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments 23 pages, 9 figures

2110.13220 2021-10-27 cs.LG stat.ML 57%

Demystifying and Generalizing BinaryConnect

Tim Dockhorn, Yaoliang Yu, Eyyüb Sari, Mahdi Zolnouri, Vahid Partovi Nia

专题命中后训练与偏好优化：post-training(abstract)；分类 cs.LG

Comments NeurIPS 2021

1. 后训练与偏好优化 4541 篇

UnDiff: Unsupervised Voice Restoration with Unconditional Diffusion Model

Beyond attention: deriving biologically interpretable insights from weakly-supervised multiple-instance learning models

Backdoor Mitigation by Correcting the Distribution of Neural Activations

BatGPT: A Bidirectional Autoregessive Talker from Generative Pre-trained Transformer

Improved Activation Clipping for Universal Backdoor Mitigation and Test-Time Detection

Primitive Skill-based Robot Learning from Human Evaluative Feedback

Ten Lessons We Have Learned in the New "Sparseland": A Short Handbook for Sparse Neural Network Researchers

The ethical ambiguity of AI data enrichment: Measuring gaps in research ethics norms and practices

Learning to Backdoor Federated Learning

BadGPT: Exploring Security Vulnerabilities of ChatGPT via Backdoor Attacks to InstructGPT

Persistently Trained, Diffusion-assisted Energy-based Models

DyNCA: Real-time Dynamic Texture Synthesis Using Neural Cellular Automata

Configuration Path Control

Critic-Guided Decoding for Controlled Text Generation

Improving Multimodal Interactive Agents with Reinforcement Learning from Human Feedback

The Expertise Problem: Learning from Specialized Feedback

What to Prune and What Not to Prune at Initialization

Self-Healing Robust Neural Networks via Closed-Loop Control

Multi-fairness under class-imbalance

Learning Dynamics and Generalization in Reinforcement Learning

Machine learning a manifold

Streamable Neural Audio Synthesis With Non-Causal Convolutions

Regularizing Attention Networks for Anomaly Detection in Visual Question Answering

FairCal: Fairness Calibration for Face Verification

Controlled abstention neural networks for identifying skillful predictions for classification problems

Controlled abstention neural networks for identifying skillful predictions for regression problems

Test-Time Detection of Backdoor Triggers for Poisoned Deep Neural Networks

Shift Happens: Adjusting Classifiers

Reverse engineering recurrent neural networks with Jacobian switching linear dynamical systems

Demystifying and Generalizing BinaryConnect