Research
A paper is listed when one of its authors is affiliated with the lab. Data: OpenAlex.
Hierarchical Compression of Vision-Language Model BenchmarksUpstage · arXiv (Cornell University)Sep 29CultureConverse: A Multilingual Multi-turn Simulation Harness for Culturally Grounded Assistance in East and Southeast AsiaMicrosoft, Upstage · arXiv (Cornell University)Aug 28Let's Scale Step by Step: Compute-Efficient Hyperparameter Transfer for Large-Scale Mixture-of-ExpertsUpstage · arXiv (Cornell University)Aug 20LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RLNVIDIA, Upstage · arXiv (Cornell University)Jul 5Breaking Failure Cascades: Step-Aware Reinforcement Learning for Medical Multimodal ReasoningUpstage · arXiv (Cornell University)Jun 30Ko-WideSearch: A Korean Breadth-Search Benchmark for Exhaustive Set Enumeration by Web AgentsUpstage · arXiv (Cornell University)Jun 25OpenBioRQ: Unsolved Biomedical Research Questions for AgentsUpstage · arXiv (Cornell University)Jun 20Multi-Step Likelihood-Ratio Correction for Reinforcement Learning with Verifiable RewardsUpstage · arXiv (Cornell University)May 20KoALa-Bench: Evaluating Large Audio Language Models on Korean Speech Understanding and FaithfulnessUpstage · arXiv (Cornell University)Mar 30ISD-Agent-Bench: A Comprehensive Benchmark for Evaluating LLM-based Instructional Design AgentsUpstage · arXiv (Cornell University)Feb 11Rewarding How Models Think Pedagogically: Integrating Pedagogical Reasoning and Thinking Rewards for LLMs in EducationUpstage · arXiv (Cornell University)Jan 21TREX: Tokenizer Regression for Optimal Data MixtureUpstage · arXiv (Cornell University)Jan 20TREX: Tokenizer Regression for Optimal Data MixtureUpstage · arXiv (Cornell University)Jan 20OpenLearnLM Benchmark: A Unified Framework for Evaluating Knowledge, Skill, and Attitude in Educational Large Language ModelsUpstage · arXiv (Cornell University)Jan 20User-Oriented Multi-Turn Dialogue Generation with Tool Use at scaleUpstage · arXiv (Cornell University)Jan 13From National Curricula to Cultural Awareness: Constructing Open-Ended Culture-Specific Question Answering DatasetUpstage · arXiv (Cornell University)Jan 8TReX: Tokenizer Regression for Optimal Data MixtureUpstage · Conference of the European Chapter of the Association for Computational LinguisticsJan 1Alignment Data Map for Efficient Preference Data Selection and DiagnosisUpstage · Findings of the Association for Computational Linguistics: ACL (Findings of ACL)Jan 1Bounded Hyperbolic Tangent: A Stable and Efficient Alternative to Pre-Layer Normalization in Large Language ModelsUpstage · arXiv (Cornell University)Dec 26, 2025MIDUS: Memory-Infused Depth Up-ScalingUpstage · arXiv (Cornell University)Dec 15, 2025Thinking Sparks!: Emergent Attention Heads in Reasoning Models During Post TrainingUpstage · arXiv (Cornell University)Sep 30, 2025Understanding LLM Development Through Longitudinal Study: Insights from the Open Ko-LLM LeaderboardUpstage · Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL)2 citationsJan 1, 2025ZERA: Zero-init Instruction Evolving Refinement Agent – From Zero Instructions to Structured Prompts via Principle-based OptimizationUpstage · Conference on Empirical Methods in Natural Language Processing (EMNLP)1 citationJan 1, 2025Rethinking KenLM: Good and Bad Model Ensembles for Efficient Text Quality Filtering in Large Web CorporaUpstage · Annual Meeting of the Association for Computational Linguistics (ACL)1 citationJan 1, 2025Open Ko-LLM Leaderboard2: Bridging Foundational and Practical Evaluation for Korean LLMsUpstage · Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL)1 citationJan 1, 2025