9 papers
GradientStabilizer:Fix the Norm, Not the Gradient
Tianjin Huang, Zhangyang Wang, Haotian Hu +10
Training instability in modern deep learning systems is frequently triggered by rare but extreme gradient-norm spikes, which can induce oversized parameter updates, corrupt optimiz…
STARS: Spike Tail-Aware Relational Synthesis for ANN-to-SNN Data-Free Knowledge Distillation
Shuhan Ye, Yi Yu, Qixin Zhang +5
SNNs promise energy-efficient and low-latency inference, but their performance still trails that of ANNs. ANN-to-SNN knowledge distillation helps narrow this gap, yet the original…
ELAS: Efficient Pre-Training of Low-Rank Large Language Models via 2:4 Activation Sparsity
Jiaxi Li, Lu Yin, Li Shen +5
Large Language Models (LLMs) have achieved remarkable capabilities, but their immense computational demands during training remain a critical bottleneck for widespread adoption. Lo…
Diffusion Language Models Know the Answer Before Decoding
Pengxiang Li, Yefan Zhou, Dilxat Muhtar +5
Diffusion language models (DLMs) have recently emerged as an alternative to autoregressive approaches, offering parallel sequence generation and flexible token orders. However, the…
Mask-Enhanced Autoregressive Prediction: Pay Less Attention to Learn More
Xialie Zhuang, Zhikai Jia, Jianjin Li +4
Large Language Models (LLMs) are discovered to suffer from accurately retrieving key information. To address this, we propose Mask-Enhanced Autoregressive Prediction (MEAP), a simp…
Rewiring Experts on the Fly:Continuous Rerouting for Better Online Adaptation in Mixture-of-Expert models
Guinan Su, Yanwu Yang, Li Shen +3
Mixture-of-Experts (MoE) models achieve efficient scaling through sparse expert activation, but often suffer from suboptimal routing decisions due to distribution shifts in deploym…