English
Related papers

Related papers: GRIN: GRadient-INformed MoE

200 papers

Parameter Efficient Tuning has been an prominent approach to adapt the Large Language Model to downstream tasks. Most previous works considers adding the dense trainable parameters, where all parameters are used to adapt certain task. We…

Computation and Language · Computer Science 2023-11-16 Yun Zhu , Nevan Wichers , Chu-Cheng Lin , Xinyi Wang , Tianlong Chen , Lei Shu , Han Lu , Canoee Liu , Liangchen Luo , Jindong Chen , Lei Meng

The Mixture-of-Experts (MoE) architecture is a powerful technique for scaling language models, yet it often suffers from expert homogenization, where experts learn redundant functionalities, thereby limiting MoE's full potential. To address…

Machine Learning · Computer Science 2026-03-03 Jiaang Li , Haibin Chen , Langming Liu , Yujin Yuan , Yadao Wang , Yizhen Zhang , Chengting Yu , Xin Tong , Weidong Zhang , Shilei Liu , Wenbo Su , Bo Zheng

The Mixture of Experts (MoE) architecture reduces the training and inference cost significantly compared to a dense model of equivalent capacity. Upcycling is an approach that initializes and trains an MoE model using a pre-trained dense…

Computation and Language · Computer Science 2025-03-18 Taishi Nakamura , Takuya Akiba , Kazuki Fujii , Yusuke Oda , Rio Yokota , Jun Suzuki

Mixture-of-Experts (MoE) activates only a subset of experts during inference, allowing the model to maintain low inference FLOPs and latency even as the parameter count scales up. However, since MoE dynamically selects the experts, all the…

Machine Learning · Computer Science 2025-05-27 Shibo Jie , Yehui Tang , Kai Han , Yitong Li , Duyu Tang , Zhi-Hong Deng , Yunhe Wang

Mixture-of-experts (MoE) models that employ sparse activation have demonstrated effectiveness in significantly increasing the number of parameters while maintaining low computational requirements per token. However, recent studies have…

Computation and Language · Computer Science 2023-10-24 Haoran Xu , Maha Elbayad , Kenton Murray , Jean Maillard , Vedanuj Goswami

The feedforward (FFW) layers in standard transformer architectures incur a linear increase in computational costs and activation memory as the hidden layer width grows. Sparse mixture-of-experts (MoE) architectures have emerged as a viable…

Machine Learning · Computer Science 2024-07-08 Xu Owen He

Emerging expert-specialized Mixture-of-Experts (MoE) architectures, such as DeepSeek-MoE, deliver strong model quality through fine-grained expert segmentation and large top-k routing. However, their scalability is limited by substantial…

Machine Learning · Computer Science 2025-08-20 Yueming Yuan , Ahan Gupta , Jianping Li , Sajal Dash , Feiyi Wang , Minjia Zhang

A pivotal advancement in the progress of large language models (LLMs) is the emergence of the Mixture-of-Experts (MoE) LLMs. Compared to traditional LLMs, MoE LLMs can achieve higher performance with fewer parameters, but it is still hard…

Computation and Language · Computer Science 2024-05-31 Xudong Lu , Qi Liu , Yuhui Xu , Aojun Zhou , Siyuan Huang , Bo Zhang , Junchi Yan , Hongsheng Li

Mixture-of-Experts (MoE) models have become the consensus approach for enabling parameter-efficient scaling and cost-effective deployment in large language models. However, existing scaling laws for dense models are inapplicable to MoE…

Machine learning models based on the aggregated outputs of submodels, either at the activation or prediction levels, often exhibit strong performance compared to individual models. We study the interplay of two popular classes of such…

Class-incremental learning (CIL) requires deep learning models to continuously acquire new knowledge from streaming data while preserving previously learned information. Recently, CIL based on pre-trained models (PTMs) has achieved…

Machine Learning · Computer Science 2025-06-16 Linjie Li , Zhenyu Wu , Yang Ji

We study dense and mixture-of-experts (MoE) transformers in a tiny-scale pretraining regime under a shared LLaMA-style decoder training recipe. The sparse model replaces dense feed-forward blocks with Mixtral-style routed experts. Dense…

Computation and Language · Computer Science 2026-05-14 Abdalrahman Wael

Mixture-of-Experts (MoE) architectures enable efficient scaling of large language models by activating only a subset of parameters per input. However, existing MoE models suffer from two critical limitations: (1) inefficient token-to-expert…

Computation and Language · Computer Science 2025-10-10 Jing Li , Zhijie Sun , Dachao Lin , Xuan He , Binfan Zheng , Yi Lin , Rongqian Zhao , Xin Chen

Mixture-of-Experts (MoE) is a neural network architecture that adds sparsely activated expert blocks to a base model, increasing the number of parameters without impacting computational costs. However, current distributed deep learning…

Machine Learning · Computer Science 2023-05-16 Siddharth Singh , Olatunji Ruwase , Ammar Ahmad Awan , Samyam Rajbhandari , Yuxiong He , Abhinav Bhatele

Mixture-of-Experts (MoE) language models dramatically expand model capacity and achieve remarkable performance without increasing per-token compute. However, can MoEs surpass dense architectures under strictly equal resource constraints --…

Computation and Language · Computer Science 2026-05-19 Houyi Li , Ka Man Lo , Shijie Xuyang , Ziqi Wang , Wenzhen Zheng , Haocheng Zhang , Zhao Li , Shuigeng Zhou , Xiangyu Zhang , Daxin Jiang

Transformer-based text embedding models have improved their performance on benchmarks like MIRACL and BEIR by increasing their parameter counts. However, this scaling approach introduces significant deployment challenges, including…

Computation and Language · Computer Science 2025-03-11 Zach Nussbaum , Brandon Duderstadt

Sparse Mixture-of-Experts (MoE) models can outperform dense large language models at similar computation by activating only a small set of experts per token. However, stacking many expert modules introduces substantial parameter memory,…

Artificial Intelligence · Computer Science 2026-04-03 Xin He , Shunkang Zhang , Kaijie Tang , Shaohuai Shi , Yuxin Wang , Zihao Zeng , Zhenheng Tang , Xiaowen Chu , Haiyan Yin , Ivor W. Tsang , Yew Soon Ong

Mixture-of-Experts (MoE) has emerged as a promising paradigm for efficiently scaling large language models without a proportional increase in computational cost. However, the standard training strategy of Top-K router prevents MoE models…

Computation and Language · Computer Science 2025-10-01 Yaoxiang Wang , Qingguo Hu , Yucheng Ding , Ruizhe Wang , Yeyun Gong , Jian Jiao , Yelong Shen , Peng Cheng , Jinsong Su

Sparsely-activated Mixture-of-Expert (MoE) layers have found practical applications in enlarging the model size of large-scale foundation models, with only a sub-linear increase in computation demands. Despite the wide adoption of hybrid…

Distributed, Parallel, and Cluster Computing · Computer Science 2024-07-04 Xinglin Pan , Wenxiang Lin , Shaohuai Shi , Xiaowen Chu , Weinong Sun , Bo Li

Traditional Mixture-of-Experts (MoE) training typically proceeds without any structural priors, effectively requiring the model to simultaneously train expert weights while searching for an optimal routing policy within a vast combinatorial…

Machine Learning · Computer Science 2026-05-26 Yuqi Xu , Rizhen Hu , Zihan Liu , Mou Sun , Kun Yuan