7 papers
Multi-Agent Debate and Visual Information Extraction for SeePhys Pro: A 1st-Place Technical Report from ICML 2026 AI4Math Track 3 Challenge
Jiseok Kwak, Suhyeon Jo, Taewoo Kim +3
This technical report presents our approach to Challenge Track~3: SeePhys Pro at the 3rd AI for Math Workshop, where the task is to answer college-level physics questions whose sta…
Lookahead Sample Reward Guidance for Test-Time Scaling of Diffusion Models
Yeongmin Kim, Donghyeok Shin, Byeonghu Na +3
Diffusion models have demonstrated strong generative performance; however, generated samples often fail to fully align with human intent. This paper studies an efficient test-time…
Distillation of Large Language Models via Concrete Score Matching
Yeongmin Kim, Donghyeok Shin, Mina Kang +2
Large language models (LLMs) deliver remarkable performance but are costly to deploy, motivating knowledge distillation (KD) for efficient inference. Existing KD objectives typical…
AMiD: Knowledge Distillation for LLMs with -mixture Assistant Distribution
Donghyeok Shin, Yeongmin Kim, Suhyeon Jo +2
Autoregressive large language models (LLMs) have achieved remarkable improvement across many tasks but incur high computational and memory costs. Knowledge distillation (KD) mitiga…
Semantic-aware Wasserstein Policy Regularization for Large Language Model Alignment
Byeonghu Na, Hyungho Na, Yeongmin Kim +4
Large language models (LLMs) are commonly aligned with human preferences using reinforcement learning from human feedback (RLHF). In this method, LLM policies are generally optimiz…
Preference Optimization by Estimating the Ratio of the Data Distribution
Yeongmin Kim, Heesun Bae, Byeonghu Na +1
Direct preference optimization (DPO) is widely used as a simple and stable method for aligning large language models (LLMs) with human preferences. This paper investigates a genera…