6 papers
Hidden Threat in Synthetic Data: Covert Targeted Bias Injection through Benign Text
Minkyung Cho, Jihyo Kim, SeungWoo Song +4
Synthetic data is increasingly used to train large language models (LLMs), yet its security implications remain poorly understood. Prior work on subliminal learning suggests that m…
Semantic Hardness Is Not Visual Hardness: Sign-Aware Hard Negative Mining for Sign Language Retrieval
Junmyeong Lee, Chan Hur, ChangSu Choi +5
Sign Language Retrieval (SLRet) enables efficient access to sign language content but remains fragile in fine-grained scenarios where visually similar signs must be distinguished.…
MentalBench: A DSM-Grounded Benchmark for Evaluating Psychiatric Diagnostic Capability of Large Language Models
Hoyun Song, Migyeong Kang, Jisu Shin +7
Large language models (LLMs) have attracted growing interest as supportive tools for psychiatric assessment and clinical decision support. However, existing mental health benchmark…
TREX: Tokenizer Regression for Optimal Data Mixture
Inho Won, Hangyeol Yoo, Minkyung Cho +3
Building effective tokenizers for multilingual Large Language Models (LLMs) requires careful control over language-specific data mixtures. While a tokenizer's compression performan…
KORMo: Korean Open Reasoning Model for Everyone
Minjun Kim, Hyeonseok Lim, Hangyeol Yoo +10
This work presents the first large-scale investigation into constructing a fully open bilingual large language model (LLM) for a non-English language, specifically Korean, trained…
MENTOR: Reinforcement Learning via Flexible Teacher-Optimized Rewards for Tool-Use Distillation
ChangSu Choi, Hoyun Song, Dongyeon Kim +6
Distilling the tool-use capabilities of large language models (LLMs) into small language models (SLMs) is essential for their practical application. The predominant approach, super…