6 papers · 1 filter
Multi-Token Residual Prediction
Yufeng Xu, Zishuo Bao, Qian Wang +6
Diffusion Language Models (DLMs) generate text by iteratively denoising masked token sequences, offering a tradeoff between parallelism and quality compared to autoregressive model…
Capacity-Aware Inference: Mitigating the Straggler Effect in Mixture of Experts
Shwai He, Weilin Cai, Jiayi Huang +1
The Mixture of Experts (MoE) is an effective architecture for scaling large language models by leveraging sparse expert activation to balance performance and efficiency. However, u…
Beyond the Birkhoff Polytope: Spectral-Sphere-Constrained Hyper-Connections
Zhaoyi Liu, Haichuan Zhang, Ang Li
Hyper-Connections (HC) generalize residual connections into multiple streams, employing residual matrices for cross-stream feature mixing to enrich model expressivity. However, unc…
DualSparse-MoE: Coordinating Tensor/Neuron-Level Sparsity with Expert Partition and Reconstruction
Weilin Cai, Le Qin, Shwai He +3
Mixture of Experts (MoE) has become a mainstream architecture for building Large Language Models (LLMs) by reducing per-token computation while enabling model scaling. It can be vi…
Towards Efficient Mixture of Experts: A Holistic Study of Compression Techniques
Shwai He, Daize Dong, Liang Ding +1
Scaling large language models has driven remarkable advancements across various domains, yet the continual increase in model size presents significant challenges for real-world dep…
What Matters in Transformers? Not All Attention is Needed
Shwai He, Guoheng Sun, Zheyu Shen +1
While scaling Transformer-based large language models (LLMs) has demonstrated promising performance across various tasks, it also introduces redundant architectures, posing efficie…