5 papers
Demystifying OPD: Length Inflation and Stabilization Strategies for Large Language Models
Feng Luo, Yu-Neng Chuang, Guanchu Wang +4
On-policy distillation (OPD) trains student models under their own induced distribution while leveraging supervision from stronger teachers. We identify a failure mode of OPD: as t…
When Truthful Representations Flip Under Deceptive Instructions?
Xianxuan Long, Yao Fu, Runchao Li +4
Large language models (LLMs) tend to follow maliciously crafted instructions to generate deceptive responses, posing safety challenges. How deceptive instructions alter the interna…
Pruning Weights but Not Truth: Safeguarding Truthfulness While Pruning LLMs
Yao Fu, Runchao Li, Xianxuan Long +4
Neural network pruning has emerged as a promising approach for deploying LLMs in low-resource scenarios while preserving downstream task performance. However, for the first time, w…
Quantized but Deceptive? A Multi-Dimensional Truthfulness Evaluation of Quantized LLMs
Yao Fu, Xianxuan Long, Runchao Li +5
Quantization enables efficient deployment of large language models (LLMs) in resource-constrained environments by significantly reducing memory and computation costs. While quantiz…
Dynamic Self-Distillation via Previous Mini-batches for Fine-tuning Small Language Models
Yao Fu, Yin Yu, Xiaotian Han +4
Knowledge distillation (KD) has become a widely adopted approach for compressing large language models (LLMs) to reduce computational costs and memory footprints. However, the avai…