Manuscript

Cycle-Consistent Search: Gold-Free Reinforcement Learning for Search Agents
Reinforcement Learning (RL) has shown strong potential for optimizing search agents in complex information retrieval tasks. However, …
Sohyun An, Shuibenyang Yuan, Hayeon Lee, Cho-Jui Hsieh, Alexander Min
Cite Paper Code
FRESCO: Benchmarking and Optimizing Re-rankers for Evolving Semantic Conflict in Retrieval-Augmented Generation
Retrieval-Augmented Generation (RAG) is a key approach to mitigating the temporal staleness of large language models (LLMs) by …
Sohyun An, Hayeon Lee, Shuibenyang Yuan, Chun-Cheng Jason Chen, Vijai Mohan, Cho-Jui Hsieh, Alexander Min
Cite Paper Code
T-MAP: Red-Teaming LLM Agents with Trajectory-aware Evolutionary Search
While prior red-teaming efforts have focused on eliciting harmful text outputs from large language models (LLMs), such approaches fail …
Hyomin Lee, Sangwoo Park, Yumin Choi, Sohyun An, Seanie Lee, Sung Ju Hwang
Cite Paper Code
DialectGen: Benchmarking and Improving Dialect Robustness in Multimodal Generation
Contact languages like English exhibit rich regional variations in the form of dialects, which are often used by dialect speakers …
Yu Zhou*, Sohyun An*, Haikang Deng*, Da Yin, Clark Peng, Cho-Jui Hsieh, Kai-Wei Chang, Nanyun Peng
Cite Paper Code
Unlabeled Data Improves Fine-Grained Image Zero-shot Classification with Multimodal LLMs
Despite Multimodal Large Language Models (MLLMs) showing promising results on general zero-shot image classification tasks, …
Yunqi Hong, Sohyun An, Andrew Bai, Neil Y.C. Lin, Cho-Jui Hsieh
Cite Paper Code
Don't Think Longer, Think Wisely: Optimizing Thinking Dynamics for Large Reasoning Models
While recent success of large reasoning models (LRMs) significantly advanced LLMs’ reasoning capability by optimizing the final …
Sohyun An, Ruochen Wang, Tianyi Zhou, Cho-Jui Hsieh
Cite Paper
Optimal Neural Architecture Generation with Diffusion Models
Existing NAS methods suffer from either an excessive amount of time for repetitive sampling and training of many task-irrelevant …
Sohyun An
Cite Paper
One Prompt is not Enough: Automated Construction of a Mixture-of-Expert Prompts
Large Language Models (LLMs) exhibit strong generalization capabilities to novel tasks when prompted with language instructions and …
Ruochen Wang*, Sohyun An*, Minhao Cheng, Tianyi Zhou, Sung Ju Hwang, Cho-Jui Hsieh
Cite Paper Code
DiffusionNAG: Predictor-guided Neural Architecture Generation with Diffusion Models
Existing NAS methods suffer from either an excessive amount of time for repetitive sampling and training of many task-irrelevant …
Sohyun An*, Hayeon Lee*, Jaehyeong Jo, Seanie Lee, Sung Ju Hwang
Cite Paper Code
Meta-prediction Model for Distillation-Aware NAS on Unseen Datasets
Distillation-aware Neural Architecture Search (DaNAS) aims to search for an optimal student architecture that obtains the best …
Hayeon Lee*, Sohyun An*, Minseon Kim, Sung Ju Hwang
Cite Paper Code