Researcher — paper.dinq.me

2026 顶会论文

作者位次 #4/4NeurIPS 2026Poster

What to Forget in Unlearning? Forget Set Curation for Language Models

Animesh Jha · Arpandeep Khatua · Youssef Allouah · Sanmi Koyejo

安全OpenReview

作者位次 #5/5NeurIPS 2026Poster

Differential Item Functioning as an Item-Level Diagnostic for LLM Benchmarks

Hansol Lee · Jason B Cho · Nick Haber · Benjamin Domingue · Sanmi Koyejo

评测与基准OpenReview

作者位次 #6/7NeurIPS 2026Poster

Benchmarks as Measurement Instruments: Quantifying Signal and Noise for More Efficient AI Evaluations Under Distribution Shift

Michael Hardy · Anka Reuel-Lamparth · Jodi Casabianca · Hansol Lee · Benjamin Domingue · Sanmi Koyejo · Mykel J. Kochenderfer

评测与基准效率与系统OpenReview

作者位次 #6/8NeurIPS 2026Poster

Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation

Yuta Kobayashi · Pradyun Ramesh · Muhammad Ahmed Chaudhry · Vincent Jeanselme · Judy Wawira · Sanmi Koyejo · Kathleen M Capaccione · Shalmali Joshi

对齐与偏好OpenReview

作者位次 #4/4NeurIPS 2026Poster

Strategic Evaluation: Incentivizing AI Capability Coverage with Private Benchmarks

Sang Truong · Serena Wang · Nick Haber · Sanmi Koyejo

评测与基准OpenReview

作者位次 #15/16NeurIPS 2026Poster

AI Construct Lexis: An Ontology of the Hidden Assumptions in AI Evaluation

Olawale Salaudeen · Florian E. Dorner · Tom Sühr · Sang Truong · Haoran Zhang · Patrick Bissett · James Fox · Marzyeh Ghassemi · Tom Hartvigsen · Peter Hase · Augustin Kelava · Suhas Mahesh · Margaret Mitchell · Jamelle Watson-Daniels · Sanmi Koyejo · Angelina Wang

评测与基准OpenReview

作者位次 #10/11NeurIPS 2026Poster

TherapyGym: Evaluating and Aligning Clinical Fidelity and Safety in Therapy Chatbots

Fangrui Huang · Souhad Chbeir · Arpandeep Khatua · Sheng Wang · Sijun Tan · Kenan Ye · Lillian A Bailey · Merryn Daniel · Ryan Louie · Sanmi Koyejo · Ehsan Adeli

对齐与偏好安全评测与基准OpenReview

作者位次 #3/3NeurIPS 2026Poster· General Machine Learning / Everything Else

Is Backpropagation Optimal? When Synthetic Gradients Improve Sample Efficiency

Yibo Jacky Zhang · Zeyu Tang · Sanmi Koyejo

效率与系统OpenReview

作者位次 #47/48NeurIPS 2026Spotlight

EVALUATION CARDS: An Interpretive Layer for AI Evaluation Reporting

Avijit Ghosh · Anka Reuel-Lamparth · Wm. Matthew Kennedy · Jenny Chim · Yilin Huang · Srishti Yadav · Anastassia Kornilova · Damian Stachura · Kevin Klyman · Jennifer Mickel · Felix Friedrich · Max Lamparth · Jan Batzner · Anoop Mishra · Jeba Sania · Eliya Habba · Nathan Heath · Shalaleh Rismani · Usman Gohar · Andrea Loehr · David Manheim · Ruchira Dhar · Sree Harsha Nelaturu · Aarush Sinha · Leshem Choshen · Yixiong Hao · Yanan Long · Andrew Tran · Drishti Sharma · Ishan Khire · Amit Saha · Subramanyam Sahoo · Michael Hardy · Michael A. Riegler · Kabir Manghnani · Michelle Lin · Nancy Jiang · Asaf Yehudai · Nuno Moniz · Yacine Jernite · Zeerak Talat · Stella Biderman · Mubashara Akhtar · Jessica Ji · Aris Hofmann · Mykel J. Kochenderfer · Sanmi Koyejo · Irene Solaiman

可解释性评测与基准OpenReview

作者位次 #3/5NeurIPS 2026Poster· Computer Vision / Image and Video Generation

Steering Away from Memorization: Reachability-Constrained Reinforcement Learning for Text-to-Image Diffusion

Sathwik Karnik · Juyeop Kim · Sanmi Koyejo · Jong-Seok Lee · Somil Bansal

多模态可解释性强化学习与后训练OpenReview

作者位次 #110/121NeurIPS 2026Poster

Harbor Adapters and Harbor-Mix: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation

Lin Shi · Haowei Lin · Zixuan Zhu · Xiaoyue Zhou · Xiang Li · Xiangning Lin · Yaxuan Deng · Han Xu · Yuangang Li · Shanda Li · Zizhao Chen · Hanwen Xing · Harsh Raj · Bo Chen · Quan Shi · Steven Dillmann · Yipeng Gao · Puneesh Khanna · Ruofan Lu · Chao Zhou · Michael Yang · Robert Zhang · Siyuan Chai · Jiayu Chang · Yizhao Chen · Xiaokun Chen · Yiwei Dai · Wenting Yang · Hange Liu · Minghao Liu · Zihan Wang · Adnan E Assadi · Benedikt Stroebl · Estefany Kelly Buchanan · Han Meng · Junwei He · Longxuan YU · Radin Shayanfar · Yukyung Lee · Zhikang Dong · Allen G Hart · Anjiang Wei · Anurag Kashyap · Arpandeep Khatua · Audrey J Zheng · Chengrui Ma · David Heineman · dubing · Hai-Anh Trinh · Haishuo Fang · Hefan Zhang · Hui Shen · Issa Sugiura · Jiankai Sun · Jiechao Gao · Junhong Lin · Junnan Li · Kai Yang · Lei Hsiung · Maoyu Wang · Mengze Tang · Nabil Omi · Negin Raoof · Nicholas Edwards · Ruohao Guo · Orfeas Menis Mastromichalakis · Ryan Ji · Przemysław Hejman · Qi Qi · Qunshu Lin · Richard Zhuang · Rui Yang · Ruichen Zheng · Ryan Marten · Shaghayegh Fazliani · Shizheng Hou · Sicong Jiang · Sijie Li · Song Bian · Terry Yue Zhuo · Tianqing Wu · Tom Tang · Wanjia Zhao · Weihao Xuan · Wenhua Liang · Xian Liu · Xin Lan · Ruby Zhang · Xuandong Zhao · Yanchuan Tang · Yifan Jiang · Yijiang Li · Yitong Guan · Yizhi Li · YONGHUI LIU · Yuheng Tang · Yujun Mao · Yunfei Zhao · Yuxin Wang · Yuxuan Tang · Zhenheng Tang · Zhifei Li · Ziruo Wang · Ziyu She · Kaiyuan Liu · Iheb Chaabane · Yuxin Tang · Xiangyi Li · Soroush Vosoughi · Sanmi Koyejo · Di He · Etash Guha · Benjamin Feuer · Andy Konwinski · Boxuan Li · Leon L Chen · Alex Dimakis · Nicholas Carlini · Mike Merrill · Ludwig Schmidt · Alex Shaw

智能体与工具评测与基准数据与预训练OpenReview

作者位次 #30/33NeurIPS 2026Poster

$\chi$-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?

Haolin Chen · Deon Metelski · Leon Qi · Tao Xia · Joonyul Lee · Steve Brown · Kevin Riley · Frank Wang · T. Y Liu · Hank Capps · Zeyu Tang · Xiangchen Song · Lingjing Kong · Fan Feng · Tianyi Zeng · Zhiwei Liu · Zixian Ma · Hang Jiang · Fangli Geng · Yuan Yuan · Chenyu You · Qingsong Wen · Hua Wei · Yanjie Fu · Yue Zhao · Carl Yang · Biwei Huang · Kun Zhang · Caiming Xiong · Sanmi Koyejo · Eric Xing · Philip S. Yu · Weiran Yao

智能体与工具社会与人本OpenReview

作者位次 #7/9NeurIPS 2026Poster· Data-centric AI / Everything Else

AI Evaluation Should Require Standardized Item-Level Data Releases

Han Jiang · Susu Zhang · Dongyao Zhu · Yuzhuo Bai · Sang Truong · Xiaoyuan Yi · Sanmi Koyejo · Xing Xie · Ziang Xiao

评测与基准OpenReview

作者位次 #5/5NeurIPS 2026Poster

Welfare, Improvability, and Variance: A Principal-Agent Approach to Optimal Benchmark Item Aggregation

Andreas Haupt · Justin Hartenstein · Anka Reuel-Lamparth · Mykel J. Kochenderfer · Sanmi Koyejo

智能体与工具评测与基准OpenReview

作者位次 #5/5NeurIPS 2026Poster· Theory / Reinforcement Learning and Planning

Behaving Better, Thinking Worse: Sycophancy Across Post-Training Stages

Sonnet Xu · Kritika Singh · Sheharbano Jafry · Roxana Daneshjou · Sanmi Koyejo

强化学习与后训练推理对齐与偏好OpenReview

Sanmi Koyejo

获取完整研究档案

联系方式

2026 顶会论文

What to Forget in Unlearning? Forget Set Curation for Language Models

Differential Item Functioning as an Item-Level Diagnostic for LLM Benchmarks

Benchmarks as Measurement Instruments: Quantifying Signal and Noise for More Efficient AI Evaluations Under Distribution Shift

Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation

Strategic Evaluation: Incentivizing AI Capability Coverage with Private Benchmarks

AI Construct Lexis: An Ontology of the Hidden Assumptions in AI Evaluation

TherapyGym: Evaluating and Aligning Clinical Fidelity and Safety in Therapy Chatbots

Is Backpropagation Optimal? When Synthetic Gradients Improve Sample Efficiency

EVALUATION CARDS: An Interpretive Layer for AI Evaluation Reporting

Steering Away from Memorization: Reachability-Constrained Reinforcement Learning for Text-to-Image Diffusion

Harbor Adapters and Harbor-Mix: Infrastructure and a Curated Meta-Dataset for Large-Scale Agentic Evaluation

$\chi$-Bench: Can AI Agents Automate End-to-End, Long-Horizon, Policy-Rich Healthcare Workflows?

AI Evaluation Should Require Standardized Item-Level Data Releases

Welfare, Improvability, and Variance: A Principal-Agent Approach to Optimal Benchmark Item Aggregation

Behaving Better, Thinking Worse: Sycophancy Across Post-Training Stages

常见合作者