id
string
sources
list
title
string
abstract
string
authors
list
categories
list
fields_of_study
list
published_date
timestamp[s]
url
string
pdf_url
string
arxiv_id
string
doi
string
citation_count
int64
influential_citation_count
int64
has_code
bool
code_url
string
venue
string
quality_score
float64
afd1624e4774072deb075cc400fdb6950ff9a25dc9fc83129f3197937f0c5677
[ "arxiv", "semantic_scholar" ]
Modeling Expert Interactions in Sparse Mixture of Experts via Graph Structures
Sparse Mixture of Experts (SMoE) has emerged as a promising solution to achieving unparalleled scalability in deep learning by decoupling model parameter count from computational cost. By activating only a small subset of parameters per sample, SMoE enables significant growth in model capacity while maintaining efficie...
[ "Minh-Khoi Nguyen-Nhat", "Rachel S. Y. Teo", "Laziz Abdullaev", "Maurice Mok", "Viet-Hoang Tran", "Tan Minh Nguyen" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-10-18T00:00:00
https://arxiv.org/abs/2510.16411
https://arxiv.org/pdf/2510.16411v1
2510.16411
10.48550/arXiv.2510.16411
1
0
false
null
arXiv.org
0.3392
cde562fbb4c913e83504097dee86bfe2f258961deefa653e83036aab4f1f1071
[ "arxiv", "semantic_scholar" ]
Expert Merging in Sparse Mixture of Experts with Nash Bargaining
Existing expert merging strategies for Sparse Mixture of Experts (SMoE) typically rely on input-dependent or input-independent averaging of expert parameters, but often lack a principled weighting mechanism. In this work, we reinterpret expert merging through the lens of game theory, revealing cooperative and competiti...
[ "Dung V. Nguyen", "Anh T. Nguyen", "Minh H. Nguyen", "Luc Q. Nguyen", "Shiqi Jiang", "Ethan Fetaya", "Linh Duy Tran", "Gal Chechik", "Tan M. Nguyen" ]
[ "cs.LG", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2025-10-17T00:00:00
https://arxiv.org/abs/2510.16138
https://arxiv.org/pdf/2510.16138v2
2510.16138
10.48550/arXiv.2510.16138
3
0
true
https://github.com/anh147/NAMEx
arXiv.org
0.5224
05660eff6de2c9d1022649de1aedea9fc777c909b82276c47308aadd230c987a
[ "arxiv", "semantic_scholar" ]
Mixture of Experts Approaches in Dense Retrieval Tasks
Dense Retrieval Models (DRMs) are a prominent development in Information Retrieval (IR). A key challenge with these neural Transformer-based models is that they often struggle to generalize beyond the specific tasks and domains they were trained on. To address this challenge, prior research in IR incorporated the Mixtu...
[ "Effrosyni Sokli", "Pranav Kasela", "Georgios Peikos", "Gabriella Pasi" ]
[ "cs.IR", "cs.AI" ]
[ "Computer Science" ]
2025-10-17T00:00:00
https://arxiv.org/abs/2510.15683
https://arxiv.org/pdf/2510.15683v1
2510.15683
10.48550/arXiv.2510.15683
0
0
true
https://github.com/FaySokli/SB-MoE
arXiv.org
0.5224
92703cef23ddf92818fca6d1d473a9ca865418cf3457290d9074bb3cf417349e
[ "arxiv", "semantic_scholar" ]
MergeMoE: Efficient Compression of MoE Models via Expert Output Merging
The Mixture-of-Experts (MoE) technique has proven to be a promising solution to efficiently scale the model size, which has been widely applied in recent LLM advancements. However, the substantial memory overhead of MoE models has made their compression an important research direction. In this work, we provide a theore...
[ "Ruijie Miao", "Yilun Yao", "Zihan Wang", "Zhiming Wang", "Bairen Yi", "LingJun Liu", "Yikai Zhao", "Tong Yang" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-10-16T00:00:00
https://arxiv.org/abs/2510.14436
https://arxiv.org/pdf/2510.14436v1
2510.14436
10.48550/arXiv.2510.14436
7
1
false
null
arXiv.org
0.3369
c72d3ddfe8a61de0449ab6bea395f7716355f88d72566c412238b836ab89b609
[ "arxiv", "semantic_scholar" ]
Who Speaks for the Trigger? Dynamic Expert Routing in Backdoored Mixture-of-Experts Transformers
Large language models (LLMs) with Mixture-of-Experts (MoE) architectures achieve impressive performance and efficiency by dynamically routing inputs to specialized subnetworks, known as experts. However, this sparse routing mechanism inherently exhibits task preferences due to expert specialization, introducing a new a...
[ "Xin Zhao", "Xiaojun Chen", "Bingshan Liu", "Haoyu Gao", "Zhendong Zhao", "Yilong Chen" ]
[ "cs.CR" ]
[ "Computer Science" ]
2025-10-15T00:00:00
https://arxiv.org/abs/2510.13462
https://arxiv.org/pdf/2510.13462v1
2510.13462
10.48550/arXiv.2510.13462
0
0
false
null
arXiv.org
0.3357
dc267a9d5e013280fe9a65b49b1d3d0b177002756c3a81c86298d6b9e5f775fb
[ "arxiv", "semantic_scholar" ]
MoBiLE: Efficient Mixture-of-Experts Inference on Consumer GPU with Mixture of Big Little Experts
Mixture-of-Experts (MoE) models have recently demonstrated exceptional performance across a diverse range of applications. The principle of sparse activation in MoE models facilitates an offloading strategy, wherein active experts are maintained in GPU HBM, while inactive experts are stored in CPU DRAM. The efficacy of...
[ "Yushu Zhao", "Yubin Qin", "Yang Wang", "Xiaolong Yang", "Huiming Han", "Shaojun Wei", "Yang Hu", "Shouyi Yin" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-10-14T00:00:00
https://arxiv.org/abs/2510.12357
https://arxiv.org/pdf/2510.12357v1
2510.12357
10.1109/ASP-DAC66049.2026.11420472
1
0
false
null
Asia and South Pacific Design Automation Conference
0.3346
914ec280ff9dbb3558c419c79e7768ff8d5f158b49e911353325643514a416ed
[ "arxiv", "semantic_scholar" ]
MC#: Mixture Compressor for Mixture-of-Experts Large Models
Mixture-of-Experts (MoE) effectively scales large language models (LLMs) and vision-language models (VLMs) by increasing capacity through sparse activation. However, preloading all experts into memory and activating multiple experts per input introduces significant computational and memory overhead, making the expert m...
[ "Wei Huang", "Yue Liao", "Yukang Chen", "Jianhui Liu", "Haoru Tan", "Si Liu", "Shiming Zhang", "Shuicheng Yan", "Xiaojuan Qi" ]
[ "cs.LG", "cs.AI" ]
[ "Medicine", "Computer Science" ]
2025-10-13T00:00:00
https://arxiv.org/abs/2510.10962
https://arxiv.org/pdf/2510.10962v1
2510.10962
10.1109/TPAMI.2026.3664873
1
0
false
null
IEEE Transactions on Pattern Analysis and Machine Intelligence
0.3334
2a530942eb5597836afd7e677be65dbd213603f7d792e0e4d9d0dc7dfba89836
[ "arxiv", "semantic_scholar" ]
Guided by the Experts: Provable Feature Learning Dynamic of Soft-Routed Mixture-of-Experts
Mixture-of-Experts (MoE) architectures have emerged as a cornerstone of modern AI systems. In particular, MoEs route inputs dynamically to specialized experts whose outputs are aggregated through weighted summation. Despite their widespread application, theoretical understanding of MoE training dynamics remains limited...
[ "Fangshuo Liao", "Anastasios Kyrillidis" ]
[ "cs.LG", "math.OC" ]
[ "Computer Science", "Mathematics" ]
2025-10-08T00:00:00
https://arxiv.org/abs/2510.07205
https://arxiv.org/pdf/2510.07205v1
2510.07205
10.48550/arXiv.2510.07205
1
0
false
null
arXiv.org
0.3277
e41b4339a83e0da574a8282f9f186fc0b30560298ecaac8f5cb7d353c5e6a163
[ "arxiv", "semantic_scholar" ]
Mixture of Neuron Experts
In this work, we first explore whether the parameters activated by the MoE layer remain highly sparse at inference. We perform a sparsification study on several representative MoE models. For each expert, we rank parameters by the magnitude of their activations from the gate projection and progressively prune the activ...
[ "Runxi Cheng", "Yuchen Guan", "Yucheng Ding", "Qingguo Hu", "Yongxian Wei", "Chun Yuan", "Yelong Shen", "Weizhu Chen", "Yeyun Gong" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-10-07T00:00:00
https://arxiv.org/abs/2510.05781
https://arxiv.org/pdf/2510.05781v1
2510.05781
10.48550/arXiv.2510.05781
4
1
false
null
arXiv.org
0.3266
a7d127629ac9ceaef5112ee129a0a425bf236aaafe5988f12d01945a8876f4ad
[ "arxiv", "semantic_scholar" ]
Multilingual Routing in Mixture-of-Experts
Mixture-of-Experts (MoE) architectures have become the key to scaling modern LLMs, yet little is understood about how their sparse routing dynamics respond to multilingual data. In this work, we analyze expert routing patterns using parallel multilingual datasets and present highly interpretable layer-wise phenomena. W...
[ "Lucas Bandarkar", "Chenyuan Yang", "Mohsen Fayyaz", "Junlin Hu", "Nanyun Peng" ]
[ "cs.CL", "cs.AI", "cs.LG" ]
[ "Computer Science" ]
2025-10-06T00:00:00
https://arxiv.org/abs/2510.04694
https://arxiv.org/pdf/2510.04694v2
2510.04694
10.48550/arXiv.2510.04694
18
2
false
null
arXiv.org
0.3254
6bd1c7b707409327165bcab6cfef094c3a72698f71b214e6ccc75f97db098d73
[ "arxiv", "semantic_scholar" ]
Dirichlet-Prior Shaping: Guiding Expert Specialization in Upcycled MoEs
Upcycling pre-trained dense models into sparse Mixture-of-Experts (MoEs) efficiently increases model capacity but often suffers from poor expert specialization due to naive weight replication. Our analysis reveals that upcycled MoEs, even with conventional regularization, exhibit low-confidence, weakly differentiated r...
[ "Leyla Mirvakhabova", "Babak Ehteshami Bejnordi", "Gaurav Kumar", "Hanxue Liang", "Wanru Zhao", "Paul Whatmough" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-10-01T00:00:00
https://arxiv.org/abs/2510.01185
https://arxiv.org/pdf/2510.01185v1
2510.01185
10.48550/arXiv.2510.01185
1
0
false
null
arXiv.org
0.3197
7306b0a046a460995257046c7e55f4a79e71d73011ed82a760636fd96379c4de
[ "arxiv", "semantic_scholar" ]
Adaptive Shared Experts with LoRA-Based Mixture of Experts for Multi-Task Learning
Mixture-of-Experts (MoE) has emerged as a powerful framework for multi-task learning (MTL). However, existing MoE-MTL methods often rely on single-task pretrained backbones and suffer from redundant adaptation and inefficient knowledge sharing during the transition from single-task to multi-task learning (STL to MTL). ...
[ "Minghao Yang", "Ren Togo", "Guang Li", "Takahiro Ogawa", "Miki Haseyama" ]
[ "cs.CV", "cs.AI", "cs.LG" ]
[ "Computer Science" ]
2025-10-01T00:00:00
https://arxiv.org/abs/2510.00570
https://arxiv.org/pdf/2510.00570v1
2510.00570
10.48550/arXiv.2510.00570
2
0
false
null
IEEE International Conference on Acoustics, Speech, and Signal Processing
0.3197
ef63e89fc84f1d521c062ef328712cd9007508578cb91933ffadc9b63da846aa
[ "arxiv", "semantic_scholar" ]
LD-MoLE: Learnable Dynamic Routing for Mixture of LoRA Experts
Recent studies have shown that combining parameter-efficient fine-tuning (PEFT) with mixture-of-experts (MoE) is an effective strategy for adapting large language models (LLMs) to the downstream tasks. However, most existing approaches rely on conventional TopK routing, which requires careful hyperparameter tuning and ...
[ "Yuan Zhuang", "Yi Shen", "Yuexin Bian", "Qing Su", "Shihao Ji", "Yuanyuan Shi", "Fei Miao" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2025-09-30T00:00:00
https://arxiv.org/abs/2509.25684
https://arxiv.org/pdf/2509.25684v2
2509.25684
10.48550/arXiv.2509.25684
5
0
false
null
arXiv.org
0.3185
9e476f63edd212a312f81d2e1e48c4a09e2bc452d2428f461f1ffb40454c9d51
[ "arxiv", "semantic_scholar" ]
Multimodal LLM With Hierarchical Mixture-of-Experts for VQA on 3D Brain MRI
Multiparametric 3D brain MRI (mpMRI) is central to neuroradiology, but producing tumor location, appearance, size, and involvement of critical structures for neurosurgical planning remains challenging. We introduce mpLLM, a multimodal LLM for visual question answering (VQA) on mpMRI that produces clinically interpretab...
[ "Arvind Murari Vepa", "Yannan Yu", "Jingru Gan", "Anthony Cuturrufo", "Michael F. Romano", "Weikai Li", "Fabien Scalzo", "Wei Wang", "Yizhou Sun" ]
[ "cs.CV", "cs.CL" ]
[ "Computer Science" ]
2025-09-30T00:00:00
https://arxiv.org/abs/2509.25889
https://arxiv.org/pdf/2509.25889v3
2509.25889
null
0
0
true
https://github.com/arvindmvepa/mpllm
null
0.3765
8deb5a6f0f24ff37abffa7cf8cb7f64eb7d938cfa5049b48065142d50bf85970
[ "arxiv", "semantic_scholar" ]
From Score Distributions to Balance: Plug-and-Play Mixture-of-Experts Routing
Mixture-of-Experts (MoE) models can scale parameter capacity by routing each token to a subset of experts through a learned gate function. While conditional routing reduces training costs, it shifts the burden on inference memory: expert parameters and activations consume memory, limiting the number of experts per devi...
[ "Rana Shahout", "Colin Cai", "Yilun Du", "Minlan Yu", "Michael Mitzenmacher" ]
[ "cs.LG", "cs.AI", "cs.DC" ]
[ "Computer Science" ]
2025-09-29T00:00:00
https://arxiv.org/abs/2510.03293
https://arxiv.org/pdf/2510.03293v1
2510.03293
10.48550/arXiv.2510.03293
4
1
false
null
arXiv.org
0.3174
3d3d87286eda6f5d0875c605528e9a0a95f2f8ed54d7759186674d83e6e93e8d
[ "arxiv", "semantic_scholar" ]
GRACE-MoE: Grouping and Replication with Locality-Aware Routing for Efficient Distributed MoE Inference
Sparse Mixture of Experts (SMoE) enables scalable parameter growth in large language models (LLMs) by selectively activating a subset of experts, and its large parameter count necessitates distributed deployment for inference. However, distributed inference faces a critical dilemma: although communication overhead cons...
[ "Yu Han", "Lehan Pan", "Jie Peng", "Ziyang Tao", "Hanqi Zhu", "Wuyang Zhang", "Yanyong Zhang" ]
[ "cs.DC" ]
[ "Computer Science" ]
2025-09-29T00:00:00
https://arxiv.org/abs/2509.25041
https://arxiv.org/pdf/2509.25041v4
2509.25041
10.48550/arXiv.2509.25041
9
0
false
null
arXiv.org
0.3174
190d4cbd0318ad77662df7b8bf792df9db6114899783f00da39c5b48da0002de
[ "arxiv", "semantic_scholar" ]
One-Prompt Strikes Back: Sparse Mixture of Experts for Prompt-based Continual Learning
Prompt-based methods have recently gained prominence in Continual Learning (CL) due to their strong performance and memory efficiency. A prevalent strategy in this paradigm assigns a dedicated subset of prompts to each task, which, while effective, incurs substantial computational overhead and causes memory requirement...
[ "Minh Le", "Bao-Ngoc Dao", "Huy Nguyen", "Quyen Tran", "Anh Nguyen", "Nhat Ho" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-09-29T00:00:00
https://arxiv.org/abs/2509.24483
https://arxiv.org/pdf/2509.24483v3
2509.24483
10.48550/arXiv.2509.24483
1
0
false
null
arXiv.org
0.3174
4c0d37c5c89339e581988e4c66a24f546eaa122ecbde9867d136e10298219561
[ "arxiv", "semantic_scholar" ]
LLaDA-MoE: A Sparse MoE Diffusion Language Model
We introduce LLaDA-MoE, a large language diffusion model with the Mixture-of-Experts (MoE) architecture, trained from scratch on approximately 20T tokens. LLaDA-MoE achieves competitive performance with significantly reduced computational overhead by maintaining a 7B-parameter capacity while activating only 1.4B parame...
[ "Fengqi Zhu", "Zebin You", "Yipeng Xing", "Zenan Huang", "Lin Liu", "Yihong Zhuang", "Guoshan Lu", "Kangyu Wang", "Xudong Wang", "Lanning Wei", "Hongrui Guo", "Jiaqi Hu", "Wentao Ye", "Tieyuan Chen", "Chenchen Li", "Chengfu Tang", "Haibo Feng", "Jun Hu", "Jun Zhou", "Xiaolu Zha...
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2025-09-29T00:00:00
https://arxiv.org/abs/2509.24389
https://arxiv.org/pdf/2509.24389v1
2509.24389
10.48550/arXiv.2509.24389
38
3
false
null
arXiv.org
0.3978
caebc1b6e088880ff69cb1442f4b80d837003118ee3e6f316be0f1482bc927c8
[ "arxiv", "semantic_scholar" ]
Bridging On-Device and Cloud LLMs for Collaborative Reasoning: A Unified Methodology for Local Routing and Post-Training
Device-cloud collaboration holds promise for deploying large language models (LLMs), leveraging lightweight on-device models for efficiency while relying on powerful cloud models for superior reasoning. A central challenge in this setting is determining, for each incoming query, whether it should be processed locally o...
[ "Wenzhi Fang", "Dong-Jun Han", "Liangqi Yuan", "Evan Chen", "Christopher Brinton" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-09-28T00:00:00
https://arxiv.org/abs/2509.24050
https://arxiv.org/pdf/2509.24050v4
2509.24050
null
2
0
false
null
null
0.2012
f1cb3de4b603c06a4a203cddf781ba656b19d43d0238be4d6a4419a5d4653f8b
[ "arxiv", "semantic_scholar" ]
Breaking the MoE LLM Trilemma: Dynamic Expert Clustering with Structured Compression
Mixture-of-Experts (MoE) Large Language Models (LLMs) face a trilemma of load imbalance, parameter redundancy, and communication overhead. We introduce a unified framework based on dynamic expert clustering and structured compression to address these issues cohesively. Our method employs an online clustering procedure ...
[ "Peijun Zhu", "Ning Yang", "Baoliang Tian", "Jiayu Wei", "Weihao Zhang", "Haijun Zhang", "Pin Lv" ]
[ "cs.CL", "cs.AI", "cs.DC", "cs.LG", "cs.NE" ]
[ "Computer Science" ]
2025-09-27T00:00:00
https://arxiv.org/abs/2510.02345
https://arxiv.org/pdf/2510.02345v3
2510.02345
10.48550/arXiv.2510.02345
2
0
true
https://github.com/szdtzpj/Breaking_the_moe_trilemma
arXiv.org
0.487
7bfe590415727f390e1af8c4c16da7663c8818002440d4b4e45cb407cc316506
[ "arxiv", "semantic_scholar" ]
Elastic MoE: Unlocking the Inference-Time Scalability of Mixture-of-Experts
Mixture-of-Experts (MoE) models typically fix the number of activated experts $k$ at both training and inference. However, real-world deployments often face heterogeneous hardware, fluctuating workloads, and diverse quality-latency requirements, while training separate models for each scenario is costly. Considering th...
[ "Naibin Gu", "Zhenyu Zhang", "Yuchen Feng", "Yilong Chen", "Peng Fu", "Zheng Lin", "Shuohuan Wang", "Yu Sun", "Hua Wu", "Weiping Wang", "Haifeng Wang" ]
[ "cs.CL", "cs.AI", "cs.LG" ]
[ "Computer Science" ]
2025-09-26T00:00:00
https://arxiv.org/abs/2509.21892
https://arxiv.org/pdf/2509.21892v2
2509.21892
10.48550/arXiv.2509.21892
8
0
false
null
arXiv.org
0.314
33f1eb9d9208b84f61e8fe6c06a4fba952e3d84783b21a4907570daa8abfd70d
[ "arxiv", "semantic_scholar" ]
Defending MoE LLMs against Harmful Fine-Tuning via Safety Routing Alignment
Recent large language models (LLMs) have increasingly adopted the Mixture-of-Experts (MoE) architecture for efficiency. MoE-based LLMs heavily depend on a superficial safety mechanism in which harmful inputs are routed safety-critical experts. However, our analysis reveals that routing decisions for harmful inputs drif...
[ "Jaehan Kim", "Minkyoo Song", "Seungwon Shin", "Sooel Son" ]
[ "cs.CR", "cs.AI" ]
[ "Computer Science" ]
2025-09-26T00:00:00
https://arxiv.org/abs/2509.22745
https://arxiv.org/pdf/2509.22745v2
2509.22745
10.48550/arXiv.2509.22745
8
1
true
null
arXiv.org
0.4852
bdd0d701020c0d7ead0f0118c47d1168829419e0a3f9c969cc6d32764c4c5ed4
[ "arxiv", "semantic_scholar" ]
Dynamic Experts Search: Enhancing Reasoning in Mixture-of-Experts LLMs at Test Time
Test-Time Scaling (TTS) enhances the reasoning ability of large language models (LLMs) by allocating additional computation during inference. However, existing approaches primarily rely on output-level sampling while overlooking the role of model architecture. In mainstream Mixture-of-Experts (MoE) LLMs, we observe tha...
[ "Yixuan Han", "Fan Ma", "Ruijie Quan", "Yi Yang" ]
[ "cs.AI", "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2025-09-26T00:00:00
https://arxiv.org/abs/2509.22572
https://arxiv.org/pdf/2509.22572v1
2509.22572
10.48550/arXiv.2509.22572
0
0
false
null
arXiv.org
0.314
9d2d43618fdaed60a29ed7c9bee403a7f777028b884d1275471f3443edab2099
[ "arxiv", "semantic_scholar" ]
Faster, Smaller, and Smarter: Task-Aware Expert Merging for Online MoE Inference
Sparse Mixture of Experts (SMoE) has become a preferred architecture for scaling Transformer capacity without increasing computational cost, as it activates only a small subset of experts for each input. However, deploying such an approach for \textit{online inference} remains challenging due to the large size of a ful...
[ "Ziyi Han", "Xutong Liu", "Ruiting Zhou", "Xiangxiang Dai", "John C. S. Lui" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-09-24T00:00:00
https://arxiv.org/abs/2509.19781
https://arxiv.org/pdf/2509.19781v2
2509.19781
10.48550/arXiv.2509.19781
1
0
false
null
arXiv.org
0.3117
f33389032cb9bbdf15e5d037727dcb410c140647fafa44f8cd6199e2908d58f5
[ "arxiv", "semantic_scholar" ]
Symphony-MoE: Harmonizing Disparate Pre-trained Models into a Coherent Mixture-of-Experts
Mixture-of-Experts (MoE) models enable scalable performance by activating large parameter sets sparsely, minimizing computational overhead. To mitigate the prohibitive cost of training MoEs from scratch, recent work employs upcycling, reusing a single pre-trained dense model by replicating its feed-forward network (FFN...
[ "Qi Wang", "Hanyang Peng", "Yue Yu" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-09-23T00:00:00
https://arxiv.org/abs/2509.18542
https://arxiv.org/pdf/2509.18542v2
2509.18542
10.48550/arXiv.2509.18542
1
0
false
null
AAAI Conference on Artificial Intelligence
0.3105
aa66404c7953d86b31a00188d01c5434d1d1433cfcc7e3a275c9bb8131e934cd
[ "arxiv", "semantic_scholar" ]
TrueMoE: Dual-Routing Mixture of Discriminative Experts for Synthetic Image Detection
The rapid progress of generative models has made synthetic image detection an increasingly critical task. Most existing approaches attempt to construct a single, universal discriminative space to separate real from fake content. However, such unified spaces tend to be complex and brittle, often struggling to generalize...
[ "Laixin Zhang", "Shuaibo Li", "Wei Ma", "Hongbin Zha" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-09-19T00:00:00
https://arxiv.org/abs/2509.15741
https://arxiv.org/pdf/2509.15741v1
2509.15741
10.48550/arXiv.2509.15741
1
0
false
null
arXiv.org
0.3059
405ea56e6594c687fc8c59a3c79b15f625b487fc33c13090161ac5650c25efb5
[ "arxiv", "semantic_scholar" ]
Semi-MoE: Mixture-of-Experts meets Semi-Supervised Histopathology Segmentation
Semi-supervised learning has been employed to alleviate the need for extensive labeled data for histopathology image segmentation, but existing methods struggle with noisy pseudo-labels due to ambiguous gland boundaries and morphological misclassification. This paper introduces Semi-MOE, to the best of our knowledge, t...
[ "Nguyen Lan Vi Vu", "Thanh-Huy Nguyen", "Thien Nguyen", "Daisuke Kihara", "Tianyang Wang", "Xingjian Li", "Min Xu" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-09-17T00:00:00
https://arxiv.org/abs/2509.13834
https://arxiv.org/pdf/2509.13834v1
2509.13834
10.48550/arXiv.2509.13834
5
0
true
https://github.com/vnlvi2k3/Semi-MoE
arXiv.org
0.4693
b50471c90aea8c3e103f00ae4afdb2e692c6d590239d269431ea6dbad65d5a1e
[ "arxiv", "semantic_scholar" ]
Exploring Expert Specialization through Unsupervised Training in Sparse Mixture of Experts
Understanding the internal organization of neural networks remains a fundamental challenge in deep learning interpretability. We address this challenge by exploring a novel Sparse Mixture of Experts Variational Autoencoder (SMoE-VAE) architecture. We test our model on the QuickDraw dataset, comparing unsupervised exper...
[ "Strahinja Nikolic", "Ilker Oguz", "Demetri Psaltis" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-09-12T00:00:00
https://arxiv.org/abs/2509.10025
https://arxiv.org/pdf/2509.10025v1
2509.10025
10.48550/arXiv.2509.10025
4
1
false
null
arXiv.org
0.2979
3b764df325e774fe747b07ec546f62eee147f760105e1bb7ff5757d71487770c
[ "arxiv", "semantic_scholar" ]
Cosine-Similarity Routing with Semantic Anchors for Interpretable Mixture-of-Experts Language Models
Mixture-of-Experts (MoE) models improve efficiency through sparse activation, but their learned gating functions provide limited insight into routing decisions. This work introduces the Semantic Resonance Architecture (SRA), which routes tokens to experts via cosine similarity between token representations and learnabl...
[ "Ivan Ternovtsii", "Yurii Bilak" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2025-09-12T00:00:00
https://arxiv.org/abs/2509.14255
https://arxiv.org/pdf/2509.14255v2
2509.14255
null
0
0
true
https://github.com/ITernovtsii/semantic-resonance
null
0.3521
2b29dd056f7ecbd8ccd59426a901abfabb5ba9d056c9c5c1f7daed1d3894cc90
[ "arxiv", "semantic_scholar" ]
Dropping Experts, Recombining Neurons: Retraining-Free Pruning for Sparse Mixture-of-Experts LLMs
Sparse Mixture-of-Experts (SMoE) architectures are widely used in large language models (LLMs) due to their computational efficiency. However, though only a few experts are activated for each token, SMoE still requires loading all expert parameters, leading to high memory usage and challenges in deployment. Previous wo...
[ "Yixiao Zhou", "Ziyu Zhao", "Dongzhou Cheng", "zhiliang wu", "Jie Gui", "Yi Yang", "Fei Wu", "Yu Cheng", "Hehe Fan" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-09-12T00:00:00
https://arxiv.org/abs/2509.10377
https://arxiv.org/pdf/2509.10377v1
2509.10377
10.48550/arXiv.2509.10377
27
1
false
null
Conference on Empirical Methods in Natural Language Processing
0.3618
fcbdd91a6677dcd2f2f6ea96d7da3fbcb39869ffa16b5770bf06f426e552bf39
[ "arxiv", "semantic_scholar" ]
Steering MoE LLMs via Expert (De)Activation
Mixture-of-Experts (MoE) in Large Language Models (LLMs) routes each token through a subset of specialized Feed-Forward Networks (FFN), known as experts. We present SteerMoE, a framework to steer MoE models by detecting and controlling behavior-associated experts. We detect key experts by comparing how often they activ...
[ "Mohsen Fayyaz", "Ali Modarressi", "Hanieh Deilamsalehy", "Franck Dernoncourt", "Ryan Rossi", "Trung Bui", "Hinrich Schütze", "Nanyun Peng" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2025-09-11T00:00:00
https://arxiv.org/abs/2509.09660
https://arxiv.org/pdf/2509.09660v2
2509.09660
10.48550/arXiv.2509.09660
21
4
true
https://github.com/adobe-research/SteerMoE
arXiv.org
0.4586
55ee14951b682ee0f028cf50fbef7a5bf025bf4498d457901de300ab6cf8432f
[ "arxiv", "semantic_scholar" ]
HD-MoE: Hybrid and Dynamic Parallelism for Mixture-of-Expert LLMs with 3D Near-Memory Processing
Large Language Models (LLMs) with Mixture-of-Expert (MoE) architectures achieve superior model performance with reduced computation costs, but at the cost of high memory capacity and bandwidth requirements. Near-Memory Processing (NMP) accelerators that stack memory directly on the compute through hybrid bonding have d...
[ "Haochen Huang", "Shuzhang Zhong", "Zhe Zhang", "Shuangchen Li", "Dimin Niu", "Hongzhong Zheng", "Runsheng Wang", "Meng Li" ]
[ "cs.PF" ]
[ "Computer Science" ]
2025-09-11T00:00:00
https://arxiv.org/abs/2509.09420
https://arxiv.org/pdf/2509.09420v1
2509.09420
10.1109/ICCAD66269.2025.11240984
4
0
false
null
null
0.1889
e5bc229e021cf5da0bdd1d8f2c5056b05532bb035d3720a98d871a178fd5caf4
[ "arxiv", "semantic_scholar" ]
MoE-Compression: How the Compression Error of Experts Affects the Inference Accuracy of MoE Model?
With the widespread application of Mixture of Experts (MoE) reasoning models in the field of LLM learning, efficiently serving MoE models under limited GPU memory constraints has emerged as a significant challenge. Offloading the non-activated experts to main memory has been identified as an efficient approach to addre...
[ "Songkai Ma", "Zhaorui Zhang", "Sheng Di", "Benben Liu", "Xiaodong Yu", "Xiaoyi Lu", "Dan Wang" ]
[ "cs.LG", "cs.DC" ]
[ "Computer Science" ]
2025-09-09T00:00:00
https://arxiv.org/abs/2509.07727
https://arxiv.org/pdf/2509.07727v1
2509.07727
10.48550/arXiv.2509.07727
3
0
false
null
arXiv.org
0.2945
1b6ca26112f56919089aa8763b66fd1adfa0f8ba8d469d9728945312c54565fc
[ "arxiv", "semantic_scholar" ]
DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance
Large Language Models (LLMs) are increasingly deployed as Internet/Web services (LLM-as-a-Service) with strict latency Service-Level Objectives (SLOs) under tight GPU memory budgets. Mixture-of-Experts (MoE) models improve quality and throughput via sparse expert activation, but serving them efficiently is challenging ...
[ "Yuning Zhang", "Grant Pinkert", "Nan Yang", "Yanli Li", "Dong Yuan" ]
[ "cs.DC" ]
[ "Computer Science" ]
2025-09-09T00:00:00
https://arxiv.org/abs/2509.07379
https://arxiv.org/pdf/2509.07379v2
2509.07379
null
1
0
false
null
null
0.1874
bfe5ce1d1eaf5cf2442350686951e952c86952bff6dce48650c67b1cf94f9204
[ "arxiv", "semantic_scholar" ]
Ban&Pick: Ehancing Performance and Efficiency of MoE-LLMs via Smarter Routing
Sparse Mixture-of-Experts (MoE) has become a key architecture for scaling large language models (LLMs) efficiently. Recent fine-grained MoE designs introduce hundreds of experts per layer, with multiple experts activated per token, enabling stronger specialization. However, during pre-training, routers are optimized ma...
[ "Yuanteng Chen", "Peisong Wang", "Yuantian Shao", "Nanxin Zeng", "Chang Xu", "Jian Cheng" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-09-08T00:00:00
https://arxiv.org/abs/2509.06346
https://arxiv.org/pdf/2509.06346v2
2509.06346
10.48550/arXiv.2509.06346
0
0
false
null
arXiv.org
0.2933
8cb1af491c54d2df9e237d053d443c229858b8f6ffa135cc268226ecd6eb6a3c
[ "arxiv", "semantic_scholar" ]
Robust Experts: the Effect of Adversarial Training on CNNs with Sparse Mixture-of-Experts Layers
Robustifying convolutional neural networks (CNNs) against adversarial attacks remains challenging and often requires resource-intensive countermeasures. We explore the use of sparse mixture-of-experts (MoE) layers to improve robustness by replacing selected residual blocks or convolutional layers, thereby increasing mo...
[ "Svetlana Pavlitska", "Haixi Fan", "Konstantin Ditschuneit", "J. Marius Zöllner" ]
[ "cs.CV", "cs.LG" ]
[ "Computer Science" ]
2025-09-05T00:00:00
https://arxiv.org/abs/2509.05086
https://arxiv.org/pdf/2509.05086v1
2509.05086
10.1109/ICCVW69036.2025.00032
3
0
true
https://github.com/KASTEL-MobilityLab/robust-sparse-moes
null
0.3426
32827409f2ffae9010d3b25e615f7bf888691801016463500e48b5d5575916e7
[ "arxiv", "semantic_scholar" ]
Extracting Uncertainty Estimates from Mixtures of Experts for Semantic Segmentation
Estimating accurate and well-calibrated predictive uncertainty is important for enhancing the reliability of computer vision models, especially in safety-critical applications like traffic scene perception. While ensemble methods are commonly used to quantify uncertainty by combining multiple models, a mixture of exper...
[ "Svetlana Pavlitska", "Beyza Keskin", "Alwin Faßbender", "Christian Hubschneider", "J. Marius Zöllner" ]
[ "cs.CV", "cs.LG" ]
[ "Computer Science" ]
2025-09-05T00:00:00
https://arxiv.org/abs/2509.04816
https://arxiv.org/pdf/2509.04816v1
2509.04816
10.1109/ICCVW69036.2025.00038
5
0
true
https://github.com/KASTEL-MobilityLab/mixtures-of-experts/
null
0.3426
2746d45bd43354c84d81501fa0fe35c7237b7fc8d78aa30221d02490e2509950
[ "arxiv", "semantic_scholar" ]
Mixture-of-Clustered-Experts: Advancing Expert Specialization and Generalization in Instruction Tuning
A sparse Mixture-of-Experts (MoE) architecture has emerged as a highly scalable solution by conditionally activating sub-modules without a proportional increase in computational costs. However, improving expert specialization to enhance performance and generalization remains a challenge for MoE, especially in instructi...
[ "Sugyeong Eo", "Jungjun Lee", "Chanjun Park", "Heuiseok Lim" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-09-03T00:00:00
https://arxiv.org/abs/2509.10513
https://arxiv.org/pdf/2509.10513v1
2509.10513
10.48550/arXiv.2509.10513
3
0
false
null
Conference on Empirical Methods in Natural Language Processing
0.2876
5c9179b3d5330344223584866b2e8371e7e3d01d00dd9164c3dfec7a00c66c72
[ "arxiv", "semantic_scholar" ]
LExI: Layer-Adaptive Active Experts for Efficient MoE Model Inference
Mixture-of-Experts (MoE) models scale efficiently by activating only a subset of experts per token, offering a computationally sparse alternative to dense architectures. While prior post-training optimizations, such as inter- and intra-expert pruning, reduce memory usage they provide limited gains in inference-time com...
[ "Krishna Teja Chitty-Venkata", "Sandeep Madireddy", "Murali Emani", "Venkatram Vishwanath" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-09-02T00:00:00
https://arxiv.org/abs/2509.02753
https://arxiv.org/pdf/2509.02753v1
2509.02753
10.48550/arXiv.2509.02753
2
0
false
null
arXiv.org
0.2865
15d6a2f927d4c2ebd145bd467769306cfdc6d0ccd385a856fcf02f73aae45288
[ "arxiv", "semantic_scholar" ]
Accelerating Mixture-of-Experts Inference by Hiding Offloading Latency with Speculative Decoding
Recent advancements in Mixture of Experts (MoE) models have significantly increased their parameter scale as well as model performance. Extensive offloading techniques have been proposed to address the GPU memory limitations of MoE inference. However, due to the I/O bottleneck and sparse computation of MoE models, exis...
[ "Zhibin Wang", "Zhonghui Zhang", "Yuhang Zhou", "Zibo Wang", "Mo Zhou", "Peng Jiang", "Weilin Cai", "Chengying Huan", "Rong Gu", "Sheng Zhong", "Chen Tian" ]
[ "cs.DC" ]
[ "Computer Science" ]
2025-08-29T00:00:00
https://arxiv.org/abs/2508.21706
https://arxiv.org/pdf/2508.21706v2
2508.21706
10.48550/arXiv.2508.21706
2
0
false
null
arXiv.org
0.2819
2bf68e3f3728a591967d3b99de57a2198e82171e4af2a530d8516d56ff915306
[ "arxiv", "semantic_scholar" ]
MoE-Health: A Mixture of Experts Framework for Robust Multimodal Healthcare Prediction
Healthcare systems generate diverse multimodal data, including Electronic Health Records (EHR), clinical notes, and medical images. Effectively leveraging this data for clinical prediction is challenging, particularly as real-world samples often present with varied or incomplete modalities. Existing approaches typicall...
[ "Xiaoyang Wang", "Christopher C. Yang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-08-29T00:00:00
https://arxiv.org/abs/2508.21793
https://arxiv.org/pdf/2508.21793v1
2508.21793
10.1145/3765612.3767207
5
1
false
null
ACM International Conference on Bioinformatics, Computational Biology and Biomedicine
0.2819
ccbc1f55ab26e4d8de4a8f8b67fa862992c38766901ed104e0f0eb1a6b986cd5
[ "arxiv", "semantic_scholar" ]
FFT-MoE: Efficient Federated Fine-Tuning for Foundation Models via Large-scale Sparse MoE under Heterogeneous Edge
As FMs drive progress toward Artificial General Intelligence (AGI), fine-tuning them under privacy and resource constraints has become increasingly critical particularly when highquality training data resides on distributed edge devices. Federated Learning (FL) offers a compelling solution through Federated Fine-Tuning...
[ "Gang Hu", "Yinglei Teng", "Pengfei Wu", "Nan Wang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-08-26T00:00:00
https://arxiv.org/abs/2508.18663
https://arxiv.org/pdf/2508.18663v1
2508.18663
10.48550/arXiv.2508.18663
6
1
false
null
arXiv.org
0.2784
f8a0763ba0ad62d6a83070c9411524d4d968714a521db849fd586b8ef8a51fff
[ "arxiv", "semantic_scholar" ]
DualSparse-MoE: Coordinating Tensor/Neuron-Level Sparsity with Expert Partition and Reconstruction
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 viewed as partitioning a large Feed-Forward Network (FFN) at the tensor level into fine-grained sub-FFNs, or experts, and activating only a sp...
[ "Weilin Cai", "Le Qin", "Shwai He", "Junwei Cui", "Ang Li", "Jiayi Huang" ]
[ "cs.LG", "cs.DC" ]
[ "Computer Science" ]
2025-08-25T00:00:00
https://arxiv.org/abs/2508.18376
https://arxiv.org/pdf/2508.18376v1
2508.18376
10.48550/arXiv.2508.18376
0
0
false
null
arXiv.org
0.2773
a7be5241cfe765939d030a842075d3d40770121f1f808883fc77bcb8bfe4664a
[ "arxiv", "semantic_scholar" ]
MoE-Inference-Bench: Performance Evaluation of Mixture of Expert Large Language and Vision Models
Mixture of Experts (MoE) models have enabled the scaling of Large Language Models (LLMs) and Vision Language Models (VLMs) by achieving massive parameter counts while maintaining computational efficiency. However, MoEs introduce several inference-time challenges, including load imbalance across experts and the addition...
[ "Krishna Teja Chitty-Venkata", "Sylvia Howland", "Golara Azar", "Daria Soboleva", "Natalia Vassilieva", "Siddhisanket Raskar", "Murali Emani", "Venkatram Vishwanath" ]
[ "cs.LG", "cs.PF" ]
[ "Computer Science" ]
2025-08-24T00:00:00
https://arxiv.org/abs/2508.17467
https://arxiv.org/pdf/2508.17467v1
2508.17467
10.1145/3731599.3767706
7
0
false
null
null
0.2258
db724565b6206a474fe9b243d00aa03111359e97c715b0f3df37e7d427c1c35a
[ "arxiv", "semantic_scholar" ]
Routing Distilled Knowledge via Mixture of LoRA Experts for Large Language Model based Bundle Generation
Large Language Models (LLMs) have shown potential in automatic bundle generation but suffer from prohibitive computational costs. Although knowledge distillation offers a pathway to more efficient student models, our preliminary study reveals that naively integrating diverse types of distilled knowledge from teacher LL...
[ "Kaidong Feng", "Zhu Sun", "Hui Fang", "Jie Yang", "Wenyuan Liu", "Yew-Soon Ong" ]
[ "cs.CL", "cs.IR" ]
[ "Computer Science" ]
2025-08-24T00:00:00
https://arxiv.org/abs/2508.17250
https://arxiv.org/pdf/2508.17250v1
2508.17250
10.48550/arXiv.2508.17250
2
0
false
null
arXiv.org
0.2761
7d7cdb270a4f403a8610cb3912726e5cb4665b507d870b24a3fa9befcb807af1
[ "arxiv", "semantic_scholar" ]
MoE-Beyond: Learning-Based Expert Activation Prediction on Edge Devices
The deployment of large-scale Mixture-of-Experts (MoE) models on edge devices presents significant challenges due to memory constraints. While MoE architectures enable efficient utilization of computational resources by activating only a subset of experts per inference, they require careful memory management to operate...
[ "Nishant Gavhane", "Arush Mehrotra", "Rohit Chawla", "Peter Proenca" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-08-23T00:00:00
https://arxiv.org/abs/2508.17137
https://arxiv.org/pdf/2508.17137v1
2508.17137
10.48550/arXiv.2508.17137
1
0
false
null
arXiv.org
0.275
d9e6788376eaa6cdfc39762bc126ed164763c0ffd85c2551e94a8ac845489f41
[ "arxiv", "semantic_scholar" ]
MultiPL-MoE: Multi-Programming-Lingual Extension of Large Language Models through Hybrid Mixture-of-Experts
Despite LLMs' excellent code creation capabilities, multilingual code generation remains extremely challenging. To address this, we intent to improve the multi-programming-lingual (MultiPL) performance of the base LLMs while retaining the most popular ones using restricted computational resources. We consider MultiPL t...
[ "Qing Wang", "Xue Han", "Jiahui Wang", "Lehao Xing", "Qian Hu", "Lianlian Zhang", "Chao Deng", "Junlan Feng" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2025-08-22T00:00:00
https://arxiv.org/abs/2508.19268
https://arxiv.org/pdf/2508.19268v2
2508.19268
10.48550/arXiv.2508.19268
3
0
false
null
Conference on Empirical Methods in Natural Language Processing
0.2739
6c0dec220b933f81c418bba3429e7cf9402f6f3aed989bb9804c32af63be7416
[ "arxiv", "semantic_scholar" ]
Anchor-MoE: A Mean-Anchored Mixture of Experts For Probabilistic Regression
Regression under uncertainty is fundamental across science and engineering. We present an Anchored Mixture of Experts (Anchor-MoE), a model that handles both probabilistic and point regression. For simplicity, we use a tuned gradient-boosting model to furnish the anchor mean; however, any off-the-shelf point regressor ...
[ "Baozhuo Su", "Zhengxian Qu" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-08-22T00:00:00
https://arxiv.org/abs/2508.16802
https://arxiv.org/pdf/2508.16802v1
2508.16802
10.48550/arXiv.2508.16802
0
0
true
https://github.com/BaozhuoSU/Probabilistic_Regression
arXiv.org
0.4232
6fd5e8a8a126c4205de32dc3a43644fb28a1bd7d0d3451f465a4d80ec96ed567
[ "arxiv", "semantic_scholar" ]
Maximum Score Routing For Mixture-of-Experts
Routing networks in sparsely activated mixture-of-experts (MoE) dynamically allocate input tokens to top-k experts through differentiable sparse transformations, enabling scalable model capacity while preserving computational efficiency. Traditional MoE networks impose an expert capacity constraint to ensure GPU-friend...
[ "Bowen Dong", "Yilong Fan", "Yutao Sun", "Zhenyu Li", "Tengyu Pan", "Xun Zhou", "Jianyong Wang" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2025-08-18T00:00:00
https://arxiv.org/abs/2508.12801
https://arxiv.org/pdf/2508.12801v1
2508.12801
10.18653/v1/2025.findings-acl.653
5
0
true
https://github.com/dongbw18/MaxScore.git}{MaxScore}$
Annual Meeting of the Association for Computational Linguistics
0.4161
e23d49ca00075e75a929aa309a74cbc840aaf918bc961c2dc006bdfe72bf6d22
[ "arxiv", "semantic_scholar" ]
X-MoE: Enabling Scalable Training for Emerging Mixture-of-Experts Architectures on HPC Platforms
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 activation memory overhead and costly all-to-all communication. Furthermore, curre...
[ "Yueming Yuan", "Ahan Gupta", "Jianping Li", "Sajal Dash", "Feiyi Wang", "Minjia Zhang" ]
[ "cs.LG", "cs.CL", "cs.DC" ]
[ "Computer Science" ]
2025-08-18T00:00:00
https://arxiv.org/abs/2508.13337
https://arxiv.org/pdf/2508.13337v1
2508.13337
10.1145/3712285.3759886
2
0
true
https://github.com/Supercomputing-System-AI-Lab/X-MoE
International Conference on Software Composition
0.4161
43f1f6f9624816450eaeef4ea26ba228ad76ba7a8776fb2ccb74b7b42a69ef0b
[ "arxiv", "semantic_scholar" ]
HierMoE: Accelerating MoE Training with Hierarchical Token Deduplication and Expert Swap
The sparsely activated mixture-of-experts (MoE) transformer has become a common architecture for large language models (LLMs) due to its sparsity, which requires fewer computational demands while easily scaling the model size. In MoE models, each MoE layer requires to dynamically choose tokens to activate particular ex...
[ "Wenxiang Lin", "Xinglin Pan", "Lin Zhang", "Shaohuai Shi", "Xuan Wang", "Xiaowen Chu" ]
[ "cs.DC", "cs.LG" ]
[ "Computer Science" ]
2025-08-13T00:00:00
https://arxiv.org/abs/2508.09591
https://arxiv.org/pdf/2508.09591v1
2508.09591
10.48550/arXiv.2508.09591
2
0
false
null
arXiv.org
0.2635
a53393b02c06a95e03f886eb7649d570e276105dfbd5647150c72f824bb5811c
[ "arxiv", "semantic_scholar" ]
MoIIE: Mixture of Intra- and Inter-Modality Experts for Large Vision Language Models
Large Vision-Language Models (LVLMs) have demonstrated remarkable performance across multi-modal tasks by scaling model size and training data. However, these dense LVLMs incur significant computational costs and motivate the exploration of sparse Mixture of Experts (MoE) architectures. While MoE improve parameter effi...
[ "Dianyi Wang", "Siyuan Wang", "Zejun Li", "Yikun Wang", "Yitong Li", "Duyu Tang", "Xiaoyu Shen", "Xuanjing Huang", "Zhongyu Wei" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-08-13T00:00:00
https://arxiv.org/abs/2508.09779
https://arxiv.org/pdf/2508.09779v2
2508.09779
10.48550/arXiv.2508.09779
5
0
true
https://github.com/AlenjandroWang/MoIIE
arXiv.org
0.4073
cacd80aeb54359b740d084aefddf7814c9ea9b30466b73fb9f593822c2c46a26
[ "arxiv", "semantic_scholar" ]
Grove MoE: Towards Efficient and Superior MoE LLMs with Adjugate Experts
The Mixture of Experts (MoE) architecture is a cornerstone of modern state-of-the-art (SOTA) large language models (LLMs). MoE models facilitate scalability by enabling sparse parameter activation. However, traditional MoE architecture uses homogeneous experts of a uniform size, activating a fixed number of parameters ...
[ "Haoyuan Wu", "Haoxing Chen", "Xiaodong Chen", "Zhanchao Zhou", "Tieyuan Chen", "Yihong Zhuang", "Guoshan Lu", "Zenan Huang", "Junbo Zhao", "Lin Liu", "Zhenzhong Lan", "Bei Yu", "Jianguo Li" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-08-11T00:00:00
https://arxiv.org/abs/2508.07785
https://arxiv.org/pdf/2508.07785v1
2508.07785
10.48550/arXiv.2508.07785
8
0
true
null
arXiv.org
0.4037
576b60e01c00309afb1372004dffa02485557b2cee45e851a0c125b3e81f843b
[ "arxiv", "semantic_scholar" ]
CBDES MoE: Hierarchically Decoupled Mixture-of-Experts for Functional Modules in Autonomous Driving
Bird's Eye View (BEV) perception systems based on multi-sensor feature fusion have become a fundamental cornerstone for end-to-end autonomous driving. However, existing multi-modal BEV methods commonly suffer from limited input adaptability, constrained modeling capacity, and suboptimal generalization. To address these...
[ "Qi Xiang", "Kunsong Shi", "Zhigui Lin", "Lei He" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-08-11T00:00:00
https://arxiv.org/abs/2508.07838
https://arxiv.org/pdf/2508.07838v1
2508.07838
10.48550/arXiv.2508.07838
4
0
false
null
arXiv.org
0.2612
d3cd8fd2425ac79b41ce6ed439368c4552e2d6d4a87c7c64556d472a0efeeba3
[ "arxiv", "semantic_scholar" ]
Efficient Multimodal Streaming Recommendation via Expandable Side Mixture-of-Experts
Streaming recommender systems (SRSs) are widely deployed in real-world applications, where user interests shift and new items arrive over time. As a result, effectively capturing users' latest preferences is challenging, as interactions reflecting recent interests are limited and new items often lack sufficient feedbac...
[ "Yunke Qu", "Liang Qu", "Tong Chen", "Quoc Viet Hung Nguyen", "Hongzhi Yin" ]
[ "cs.IR" ]
[ "Computer Science" ]
2025-08-08T00:00:00
https://arxiv.org/abs/2508.05993
https://arxiv.org/pdf/2508.05993v3
2508.05993
10.1145/3746252.3761390
6
0
true
https://github.com/qykcq/Efficient-Multimodal-Streaming-Recommendation-via-Expandable-Side-Mixture-of-Experts
International Conference on Information and Knowledge Management
0.3984
81818ae372c0059e5e2e99e6a4ca37ad0ff3c0f9ca9c49d8b33da88017b2de2d
[ "arxiv", "semantic_scholar" ]
MoBE: Mixture-of-Basis-Experts for Compressing MoE-based LLMs
The Mixture-of-Experts (MoE) architecture has become a predominant paradigm for scaling large language models (LLMs). Despite offering strong performance and computational efficiency, large MoE-based LLMs like DeepSeek-V3-0324 and Kimi-K2-Instruct present serious challenges due to substantial memory requirements in dep...
[ "Xiaodong Chen", "Mingming Ha", "Zhenzhong Lan", "Jing Zhang", "Jianguo Li" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-08-07T00:00:00
https://arxiv.org/abs/2508.05257
https://arxiv.org/pdf/2508.05257v1
2508.05257
10.48550/arXiv.2508.05257
5
2
false
null
arXiv.org
0.2567
6f68b5a2916a75fea3a2a24d7c564ab45af8baf77a2eec80300c0fa893ef9ce0
[ "arxiv", "semantic_scholar" ]
Parameter-Efficient Routed Fine-Tuning: Mixture-of-Experts Demands Mixture of Adaptation Modules
Mixture-of-Experts (MoE) benefits from a dynamic routing mechanism among their specialized experts, which existing Parameter- Efficient Fine-Tuning (PEFT) strategies fail to leverage. This motivates us to investigate whether adaptation modules themselves should incorporate routing mechanisms to align with MoE's multi-e...
[ "Yilun Liu", "Yunpu Ma", "Yuetian Lu", "Shuo Chen", "Zifeng Ding", "Volker Tresp" ]
[ "cs.LG", "cs.AI", "cs.CL" ]
[ "Computer Science" ]
2025-08-04T00:00:00
https://arxiv.org/abs/2508.02587
https://arxiv.org/pdf/2508.02587v1
2508.02587
10.48550/arXiv.2508.02587
0
0
false
null
Conference of the European Chapter of the Association for Computational Linguistics
0.2532
c34691f02cbdff0ef09dc7c09d0c5ba8af2de845da0feaaf0d66443ea5ac90b7
[ "arxiv", "semantic_scholar" ]
EAC-MoE: Expert-Selection Aware Compressor for Mixture-of-Experts Large Language Models
Mixture-of-Experts (MoE) has demonstrated promising potential in scaling LLMs. However, it is hindered by two critical challenges: (1) substantial GPU memory consumption to load all experts; (2) low activated parameters cannot be equivalently translated into inference acceleration effects. In this work, we propose EAC-...
[ "Yuanteng Chen", "Yuantian Shao", "Peisong Wang", "Jian Cheng" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-08-03T00:00:00
https://arxiv.org/abs/2508.01625
https://arxiv.org/pdf/2508.01625v1
2508.01625
10.18653/v1/2025.acl-long.633
14
1
false
null
Annual Meeting of the Association for Computational Linguistics
0.294
0b21d163f7dbdfd1b80a778cf4fda4124d774b8a5076ae7fad5a23fa9ff39607
[ "arxiv", "semantic_scholar" ]
RouteMark: A Fingerprint for Intellectual Property Attribution in Routing-based Model Merging
Model merging via Mixture-of-Experts (MoE) has emerged as a scalable solution for consolidating multiple task-specific models into a unified sparse architecture, where each expert is derived from a model fine-tuned on a distinct task. While effective for multi-task integration, this paradigm introduces a critical yet u...
[ "Xin He", "Junxi Shen", "Zhenheng Tang", "Xiaowen Chu", "Bo Li", "Ivor W. Tsang", "Yew-Soon Ong" ]
[ "cs.CR", "cs.AI", "cs.ET", "cs.LG" ]
[ "Computer Science" ]
2025-08-03T00:00:00
https://arxiv.org/abs/2508.01784
https://arxiv.org/pdf/2508.01784v1
2508.01784
10.48550/arXiv.2508.01784
4
1
false
null
arXiv.org
0.2521
4d451a988ba5947b17dfb01408edda247f8adbe3a6da490f998af5a137368d36
[ "arxiv", "semantic_scholar" ]
A3D-MoE: Acceleration of Large Language Models with Mixture of Experts via 3D Heterogeneous Integration
Conventional large language models (LLMs) are equipped with dozens of GB to TB of model parameters, making inference highly energy-intensive and costly as all the weights need to be loaded to onboard processing elements during computation. Recently, the Mixture-of-Experts (MoE) architecture has emerged as an efficient ...
[ "Wei-Hsing Huang", "Janak Sharda", "Cheng-Jhih Shih", "Yuyao Kong", "Faaiq Waqar", "Pin-Jun Chen", " Yingyan", " Lin", "Shimeng Yu" ]
[ "cs.AR" ]
[ "Computer Science" ]
2025-07-25T00:00:00
https://arxiv.org/abs/2507.19142
https://arxiv.org/pdf/2507.19142v1
2507.19142
10.48550/arXiv.2507.19142
1
0
false
null
arXiv.org
0.2418
8f4a1c3f547790d836d0f9a0573c10755bc11f55bf596054b7ad1a800f3f7ee4
[ "arxiv", "semantic_scholar" ]
Astro-MoE: Mixture of Experts for Multiband Astronomical Time Series
Multiband astronomical time series exhibit heterogeneous variability patterns, sampling cadences, and signal characteristics across bands. Standard transformers apply shared parameters to all bands, potentially limiting their ability to model this rich structure. In this work, we introduce Astro-MoE, a foundational tra...
[ "Martina Cádiz-Leyton", "Guillermo Cabrera-Vives", "Pavlos Protopapas", "Daniel Moreno-Cartagena", "Ignacio Becker" ]
[ "astro-ph.IM" ]
[ "Physics" ]
2025-07-16T00:00:00
https://arxiv.org/abs/2507.12611
https://arxiv.org/pdf/2507.12611v1
2507.12611
null
0
0
false
null
null
0.1473
561763d7db457c6b87ccfb2ec92db8ffae43c6f1bbd9e59e660364652c7ebfeb
[ "arxiv", "semantic_scholar" ]
Mixture of Raytraced Experts
We introduce a Mixture of Raytraced Experts, a stacked Mixture of Experts (MoE) architecture which can dynamically select sequences of experts, producing computational graphs of variable width and depth. Existing MoE architectures generally require a fixed amount of computation for a given sample. Our approach, in cont...
[ "Andrea Perin", "Giacomo Lagomarsini", "Claudio Gallicchio", "Giuseppe Nuti" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-07-16T00:00:00
https://arxiv.org/abs/2507.12419
https://arxiv.org/pdf/2507.12419v1
2507.12419
10.48550/arXiv.2507.12419
0
0
true
https://github.com/nutig/RayTracing
arXiv.org
0.3577
9e452af04149c17fb2edf43217870e77e4f1b29193cdfe55d4a9ad65d5b36461
[ "arxiv", "semantic_scholar" ]
Mixture of Experts in Large Language Models
This paper presents a comprehensive review of the Mixture-of-Experts (MoE) architecture in large language models, highlighting its ability to significantly enhance model performance while maintaining minimal computational overhead. Through a systematic analysis spanning theoretical foundations, core architectural desig...
[ "Danyang Zhang", "Junhao Song", "Ziqian Bi", "Xinyuan Song", "Yingfang Yuan", "Tianyang Wang", "Joe Yeong", "Junfeng Hao" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-07-15T00:00:00
https://arxiv.org/abs/2507.11181
https://arxiv.org/pdf/2507.11181v2
2507.11181
10.48550/arXiv.2507.11181
17
0
false
null
arXiv.org
0.3138
90931df041fe069ee89a106bcba0bbc51e51c6d2465381f2246a9a5c0518b80c
[ "arxiv", "semantic_scholar" ]
Omni-Router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition
Mixture-of-experts (MoE) architectures have expanded from language modeling to automatic speech recognition (ASR). Traditional MoE methods, such as the Switch Transformer, route experts independently within each layer. Our analysis reveals that routers in most layers make expert choices that are not strongly correlated...
[ "Zijin Gu", "Tatiana Likhomanenko", "Navdeep Jaitly" ]
[ "cs.CL", "cs.AI", "cs.LG", "cs.SD", "eess.AS" ]
[ "Computer Science", "Engineering" ]
2025-07-08T00:00:00
https://arxiv.org/abs/2507.05724
https://arxiv.org/pdf/2507.05724v3
2507.05724
10.1109/ASRU65441.2025.11434645
5
0
false
null
Automatic Speech Recognition & Understanding
0.2223
7b9dc73071cc766c5efb917c0483d7f3bf791822016a02b89b1a7755227dc26f
[ "arxiv", "semantic_scholar" ]
Neural Inhibition Improves Dynamic Routing and Mixture of Experts
To be effective, efficient, and diverse, deep learning models need to dynamically choose its architecture based on signals from a population of neurons. We hypothesize dynamic routing models can be improved with neural inhibition in those neural populations. This means signals commonly shared among the various modes of...
[ "Will Y. Zou", "Jennifer Y. Zhang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-07-03T00:00:00
https://arxiv.org/abs/2507.03221
https://arxiv.org/pdf/2507.03221v1
2507.03221
10.48550/arXiv.2507.03221
0
0
false
null
arXiv.org
0.2166
7f759f14df81d99c9f15c9070ed72eafc402ff157161665a867af981c4ff0572
[ "arxiv", "semantic_scholar" ]
Sub-MoE: Efficient Mixture-of-Expert LLMs Compression via Subspace Expert Merging
Mixture of Experts (MoE) LLMs face significant obstacles due to their massive parameter scale, which imposes memory, storage, and deployment challenges. Although recent expert merging methods promise greater efficiency by consolidating multiple experts, they are fundamentally hindered by parameter conflicts arising fro...
[ "Lujun Li", "Zhu Qiyuan", "Jiacheng Wang", "Wei Li", "Hao Gu", "Sirui Han", "Yike Guo" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-06-29T00:00:00
https://arxiv.org/abs/2506.23266
https://arxiv.org/pdf/2506.23266v1
2506.23266
10.48550/arXiv.2506.23266
20
3
true
https://github.com/lliai/MoERazor
arXiv.org
0.3306
b06deff662ae80a82faf5acbc179dbc02078ea76f74e4a8ad11be5a9d3a76b3c
[ "arxiv", "semantic_scholar" ]
Latent Prototype Routing: Achieving Near-Perfect Load Balancing in Mixture-of-Experts
Mixture-of-Experts (MoE) architectures have emerged as a key strategy for scaling large language models (LLMs) efficiently. However, current MoE systems suffer from severe load imbalance, where only a small subset of experts is consistently activated during training and inference, leading to significant underutilizatio...
[ "Jiajie Yang" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2025-06-26T00:00:00
https://arxiv.org/abs/2506.21328
https://arxiv.org/pdf/2506.21328v1
2506.21328
10.48550/arXiv.2506.21328
3
1
true
null
arXiv.org
0.3223
3358b6341012f9f51d0327dc3c65043c732884cc77987e744e268ef0ee5fb680
[ "arxiv", "semantic_scholar" ]
Little by Little: Continual Learning via Incremental Mixture of Rank-1 Associative Memory Experts
Continual learning (CL) with large pre-trained models aims to incrementally acquire knowledge without catastrophic forgetting. Existing LoRA-based Mixture-of-Experts (MoE) methods expand capacity by adding isolated new experts while freezing old ones, but still suffer from redundancy, interference, routing ambiguity, a...
[ "Haodong Lu", "Chongyang Zhao", "Minhui Xue", "Lina Yao", "Kristen Moore", "Dong Gong" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-06-26T00:00:00
https://arxiv.org/abs/2506.21035
https://arxiv.org/pdf/2506.21035v6
2506.21035
null
1
0
false
null
null
0.1327
6bef6539037940b89a3ed683fc5d05f74085caa76e2b384fd833ceff4fe26e17
[ "arxiv" ]
SlimMoE: Structured Compression of Large MoE Models via Expert Slimming and Distillation
The Mixture of Experts (MoE) architecture has emerged as a powerful paradigm for scaling large language models (LLMs) while maintaining inference efficiency. However, their enormous memory requirements make them prohibitively expensive to fine-tune or deploy in resource-constrained environments. To address this challen...
[ "Zichong Li", "Chen Liang", "Zixuan Zhang", "Ilgee Hong", "Young Jin Kim", "Weizhu Chen", "Tuo Zhao" ]
[ "cs.LG", "cs.CL" ]
[]
2025-06-23T00:00:00
https://arxiv.org/abs/2506.18349
https://arxiv.org/pdf/2506.18349v1
2506.18349
null
0
0
false
null
null
0.1305
70975b8c57bd95b226981727407b06668d67bc171c6a48aca965f3802035deb4
[ "arxiv", "semantic_scholar" ]
Routing Mamba: Scaling State Space Models with Mixture-of-Experts Projection
Linear State Space Models (SSMs) offer remarkable performance gains in efficient sequence modeling, with constant inference-time computation and memory complexity. Recent advances, such as Mamba, further enhance SSMs with input-dependent gating and hardware-aware implementations, positioning them as strong alternatives...
[ "Zheng Zhan", "Liliang Ren", "Shuohang Wang", "Liyuan Liu", "Yang Liu", "Yeyun Gong", "Yanzhi Wang", "Yelong Shen" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-06-22T00:00:00
https://arxiv.org/abs/2506.18145
https://arxiv.org/pdf/2506.18145v1
2506.18145
10.48550/arXiv.2506.18145
2
0
false
null
arXiv.org
0.204
bb5bb3d3619fa535c65634c6e918f6a2ea2ea5fd0428a383f25cbc5d1e76dd21
[ "arxiv", "semantic_scholar" ]
SAFEx: Analyzing Vulnerabilities of MoE-Based LLMs via Stable Safety-critical Expert Identification
Large language models with Mixture-of-Experts (MoE) architectures achieve efficiency and scalability, yet their routing mechanisms introduce safety alignment challenges insufficiently addressed by techniques developed for dense models. In this work, the MoE-specific safety risk of positional vulnerability-that safety-a...
[ "Zhenglin Lai", "Mengyao Liao", "Bingzhe Wu", "Dong Xu", "Zebin Zhao", "Zhihang Yuan", "Chao Fan", "Jianqiang Li" ]
[ "cs.LG", "cs.AI", "cs.CR" ]
[ "Computer Science" ]
2025-06-20T00:00:00
https://arxiv.org/abs/2506.17368
https://arxiv.org/pdf/2506.17368v2
2506.17368
10.48550/arXiv.2506.17368
15
8
false
null
arXiv.org
0.4771
0d907cc216d8f42f4201f8761a51c6a63d67b81474867f20cc48449a8cb75844
[ "arxiv", "semantic_scholar" ]
GuiLoMo: Allocating Expert Number and Rank for LoRA-MoE via Bilevel Optimization with GuidedSelection Vectors
Parameter-efficient fine-tuning (PEFT) methods, particularly Low-Rank Adaptation (LoRA), offer an efficient way to adapt large language models with reduced computational costs. However, their performance is limited by the small number of trainable parameters. Recent work combines LoRA with the Mixture-of-Experts (MoE),...
[ "Hengyuan Zhang", "Xinrong Chen", "Yingmin Qiu", "Xiao Liang", "Ziyue Li", "Guanyu Wang", "Weiping Li", "Tong Mo", "Hayden Kwok-Hay So", "Ngai Wong" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-06-17T00:00:00
https://arxiv.org/abs/2506.14646
https://arxiv.org/pdf/2506.14646v2
2506.14646
10.48550/arXiv.2506.14646
5
0
true
https://github.com/Liar406/Gui-LoMo.git
arXiv.org
0.3064
ecc4958fb51fcf1e958702d726b0cc8d15004f602491b0fe8c5244cd696a82fc
[ "arxiv", "semantic_scholar" ]
EAQuant: Enhancing Post-Training Quantization for MoE Models via Expert-Aware Optimization
Mixture-of-Experts (MoE) models enable scalable computation and performance in large-scale deep learning but face quantization challenges due to sparse expert activation and dynamic routing. Existing post-training quantization (PTQ) methods fail to address activation outliers, routing instability, and sparse expert cal...
[ "Zhongqian Fu", "Tianyi Zhao", "Ning Ding", "Xianzhi Yu", "Xiaosong Li", "Yehui Tang", "Yunhe Wang" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-06-16T00:00:00
https://arxiv.org/abs/2506.13329
https://arxiv.org/pdf/2506.13329v3
2506.13329
10.48550/arXiv.2506.13329
5
1
true
https://github.com/darren-fzq1/EAQuant
arXiv.org
0.3046
d5c66741fb8caad467dbdbd21b80574faf1b2918595ecbb5db4567b1ddaea712
[ "arxiv", "semantic_scholar" ]
Load Balancing Mixture of Experts with Similarity Preserving Routers
Sparse Mixture of Experts (MoE) models offer a scalable and efficient architecture for training large neural networks by activating only a subset of parameters ("experts") for each input. A learned router computes a distribution over these experts, and assigns input tokens to a small subset. However, without auxiliary ...
[ "Nabil Omi", "Siddhartha Sen", "Ali Farhadi" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-06-16T00:00:00
https://arxiv.org/abs/2506.14038
https://arxiv.org/pdf/2506.14038v2
2506.14038
10.48550/arXiv.2506.14038
20
1
false
null
arXiv.org
0.3306
28ff1d0cc25efc98fd2f31075c3b19b391cfac6a310c2d1cd5c538dfb3b44dff
[ "arxiv", "semantic_scholar" ]
Automatic Expert Discovery in LLM Upcycling via Sparse Interpolated Mixture-of-Experts
We present Sparse Interpolated Mixture-of-Experts (SIMoE) instruction-tuning, an end-to-end algorithm designed to fine-tune a dense pre-trained Large Language Model (LLM) into a MoE-style model that possesses capabilities in multiple specialized domains. During instruction-tuning, SIMoE automatically identifies multipl...
[ "Shengzhuang Chen", "Ying Wei", "Jonathan Richard Schwarz" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-06-14T00:00:00
https://arxiv.org/abs/2506.12597
https://arxiv.org/pdf/2506.12597v1
2506.12597
10.48550/arXiv.2506.12597
2
0
false
null
Annual Meeting of the Association for Computational Linguistics
0.1948
16742791c22dc07462091a4c64de8d0cf91d9849504166466f70a626693f2598
[ "arxiv", "semantic_scholar" ]
DIVE into MoE: Diversity-Enhanced Reconstruction of Large Language Models from Dense into Mixture-of-Experts
Large language models (LLMs) with the Mixture-of-Experts (MoE) architecture achieve high cost-efficiency by selectively activating a subset of the parameters. Despite the inference efficiency of MoE LLMs, the training of extensive experts from scratch incurs substantial overhead, whereas reconstructing a dense LLM into...
[ "Yuchen Feng", "Bowen Shen", "Naibin Gu", "Jiaxuan Zhao", "Peng Fu", "Zheng Lin", "Weiping Wang" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-06-11T00:00:00
https://arxiv.org/abs/2506.09351
https://arxiv.org/pdf/2506.09351v1
2506.09351
10.48550/arXiv.2506.09351
7
0
true
null
Annual Meeting of the Association for Computational Linguistics
0.2957
34994366161d794a934641ae79e71d97d44a71b86871674dbd71095e2ec4199e
[ "arxiv", "semantic_scholar" ]
MoE-GPS: Guidlines for Prediction Strategy for Dynamic Expert Duplication in MoE Load Balancing
In multi-GPU Mixture-of-Experts (MoE) network, experts are distributed across different GPUs, which creates load imbalance as each expert processes different number of tokens. Recent works improve MoE inference load balance by dynamically duplicating popular experts to more GPUs to process excessive tokens, which requi...
[ "Haiyue Ma", "Zhixu Du", "Yiran Chen" ]
[ "cs.LG", "cs.AR" ]
[ "Computer Science" ]
2025-06-09T00:00:00
https://arxiv.org/abs/2506.07366
https://arxiv.org/pdf/2506.07366v1
2506.07366
10.48550/arXiv.2506.07366
3
1
false
null
arXiv.org
0.1891
3d5be2f18e2ba2a66f0b85097901fd0fd031325484b458c188038183bee1ff68
[ "arxiv", "semantic_scholar" ]
SMAR: Soft Modality-Aware Routing Strategy for MoE-based Multimodal Large Language Models Preserving Language Capabilities
Mixture of Experts (MoE) architectures have become a key approach for scaling large language models, with growing interest in extending them to multimodal tasks. Existing methods to build multimodal MoE models either incur high training costs or suffer from degraded language capabilities when adapting pretrained models...
[ "Guoyang Xia", "Yifeng Ding", "Fengfa Li", "Lei Ren", "Wei Chen", "Fangxiang Feng", "Xiaojie Wang" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2025-06-06T00:00:00
https://arxiv.org/abs/2506.06406
https://arxiv.org/pdf/2506.06406v2
2506.06406
10.48550/arXiv.2506.06406
3
0
false
null
arXiv.org
0.1856
254cda689695d6c290a2db92e579a416a30d299d659c60a8bc046326c978e5f6
[ "arxiv", "semantic_scholar" ]
PC-MoE: Memory-Efficient and Privacy-Preserving Collaborative Training for Mixture-of-Experts LLMs
Mixture-of-Experts (MoE) has been gaining popularity due to its successful adaptation to large language models (LLMs). In this work, we introduce Privacy-preserving Collaborative Mixture-of-Experts (PC-MoE), which leverages the sparsity of the MoE architecture for memory-efficient decentralized collaborative LLM traini...
[ "Ze Yu Zhang", "Bolin Ding", "Bryan Kian Hsiang Low" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-06-03T00:00:00
https://arxiv.org/abs/2506.02965
https://arxiv.org/pdf/2506.02965v2
2506.02965
10.1007/s10994-025-06901-2
0
0
false
null
Machine-mediated learning
0.1822
133c8b83d21acdfc50d881dcddf2a16956e0e6c16ab080da4c4164de2e878836
[ "arxiv", "semantic_scholar" ]
EvoMoE: Expert Evolution in Mixture of Experts for Multimodal Large Language Models
Recent advancements have shown that the Mixture of Experts (MoE) approach significantly enhances the capacity of large language models (LLMs) and improves performance on downstream tasks. Building on these promising results, multi-modal large language models (MLLMs) have increasingly adopted MoE techniques. However, ex...
[ "Linglin Jing", "Yuting Gao", "Zhigang Wang", "Wang Lan", "Yiwen Tang", "Wenhai Wang", "Kaipeng Zhang", "Qingpei Guo" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-28T00:00:00
https://arxiv.org/abs/2505.23830
https://arxiv.org/pdf/2505.23830v1
2505.23830
10.48550/arXiv.2505.23830
4
0
false
null
AAAI Conference on Artificial Intelligence
0.1753
efe0fba3bf46dbe23b276390f98d3ab08490eb826cb09d11305c979648104bdd
[ "arxiv", "semantic_scholar" ]
Advancing Expert Specialization for Better MoE
Mixture-of-Experts (MoE) models enable efficient scaling of large language models (LLMs) by activating only a subset of experts per input. However, we observe that the commonly used auxiliary load balancing loss often leads to expert overlap and overly uniform routing, which hinders expert specialization and degrades o...
[ "Hongcan Guo", "Haolang Lu", "Guoshun Nan", "Bolun Chu", "Jialin Zhuang", "Yuan Yang", "Wenhao Che", "Xinye Cao", "Sicong Leng", "Qimei Cui", "Xudong Jiang" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-28T00:00:00
https://arxiv.org/abs/2505.22323
https://arxiv.org/pdf/2505.22323v5
2505.22323
10.48550/arXiv.2505.22323
33
4
false
null
arXiv.org
0.3829
52f4e94628ca52cdc18e6fd6231248b42f4edddfd6c29eed9c278c1a6e185080
[ "arxiv", "semantic_scholar" ]
Less, but Better: Efficient Multilingual Expansion for LLMs via Layer-wise Mixture-of-Experts
Continually expanding new languages for existing large language models (LLMs) is a promising yet challenging approach to building powerful multilingual LLMs. The biggest challenge is to make the model continuously learn new languages while preserving the proficient ability of old languages. To achieve this, recent work...
[ "Xue Zhang", "Yunlong Liang", "Fandong Meng", "Songming Zhang", "Yufeng Chen", "Jinan Xu", "Jie Zhou" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-28T00:00:00
https://arxiv.org/abs/2505.22582
https://arxiv.org/pdf/2505.22582v1
2505.22582
10.48550/arXiv.2505.22582
13
1
false
null
Annual Meeting of the Association for Computational Linguistics
0.2865
ef75723a797c024eb7f3baf12df51ad90bff17ebd314a62e9e7eeb5e316586bf
[ "arxiv", "semantic_scholar" ]
Pangu Pro MoE: Mixture of Grouped Experts for Efficient Sparsity
The surgence of Mixture of Experts (MoE) in Large Language Models promises a small price of execution cost for a much larger model parameter count and learning capacity, because only a small fraction of parameters are activated for each input token. However, it is commonly observed that some experts are activated far m...
[ "Yehui Tang", "Xiaosong Li", "Fangcheng Liu", "Wei Guo", "Hang Zhou", "Yaoyuan Wang", "Kai Han", "Xianzhi Yu", "Jinpeng Li", "Hui Zang", "Fei Mi", "Xiaojun Meng", "Zhicheng Liu", "Hanting Chen", "Binfan Zheng", "Can Chen", "Youliang Yan", "Ruiming Tang", "Peifeng Qin", "Xinghao...
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-27T00:00:00
https://arxiv.org/abs/2505.21411
https://arxiv.org/pdf/2505.21411v2
2505.21411
10.48550/arXiv.2505.21411
16
3
true
null
arXiv.org
0.3076
2f5ad05c41b713f4bf8dcc6b3520f43f12cf016e8e89ba5a12e3ff5db0932376
[ "arxiv", "semantic_scholar" ]
Uni3D-MoE: Scalable Multimodal 3D Scene Understanding via Mixture of Experts
Recent advancements in multimodal large language models (MLLMs) have demonstrated considerable potential for comprehensive 3D scene understanding. However, existing approaches typically utilize only one or a limited subset of 3D modalities, resulting in incomplete representations of 3D scenes and reduced interpretive a...
[ "Yue Zhang", "Yingzhao Jian", "Hehe Fan", "Yi Yang", "Roger Zimmermann" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-05-27T00:00:00
https://arxiv.org/abs/2505.21079
https://arxiv.org/pdf/2505.21079v1
2505.21079
10.48550/arXiv.2505.21079
3
0
false
null
arXiv.org
0.1742
95d8394f44a9518a2b18dc2b861c93a623aae02a9ce233f011e11b1983e482a8
[ "arxiv", "semantic_scholar" ]
FLAME-MoE: A Transparent End-to-End Research Platform for Mixture-of-Experts Language Models
Recent large language models such as Gemini-1.5, DeepSeek-V3, and Llama-4 increasingly adopt Mixture-of-Experts (MoE) architectures, which offer strong efficiency-performance trade-offs by activating only a fraction of the model per token. Yet academic researchers still lack a fully open, end-to-end MoE platform for in...
[ "Hao Kang", "Zichun Yu", "Chenyan Xiong" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2025-05-26T00:00:00
https://arxiv.org/abs/2505.20225
https://arxiv.org/pdf/2505.20225v1
2505.20225
10.48550/arXiv.2505.20225
3
0
true
https://github.com/cmu-flame/FLAME-MoE
arXiv.org
0.2674
afc6cdfb6273e683d3d25a94edcba4c5a9032e9423a79a92272725f4589d4a90
[ "arxiv", "semantic_scholar" ]
Guiding the Experts: Semantic Priors for Efficient and Focused MoE Routing
Mixture-of-Experts (MoE) models have emerged as a promising direction for scaling vision architectures efficiently. Among them, Soft MoE improves training stability by assigning each token to all experts via continuous dispatch weights. However, current designs overlook the semantic structure which is implicitly encode...
[ "Chengxi Min", "Wei Wang", "Yahui Liu", "Weixin Ye", "Enver Sangineto", "Qi Wang", "Yao Zhao" ]
[ "cs.CV" ]
[ "Computer Science" ]
2025-05-24T00:00:00
https://arxiv.org/abs/2505.18586
https://arxiv.org/pdf/2505.18586v1
2505.18586
10.48550/arXiv.2505.18586
1
0
false
null
arXiv.org
0.1707
1ca5f0f0d3da940a24065d0c53de9d53316538767e4d9ef9289eeb860ea3db14
[ "arxiv", "semantic_scholar" ]
$μ$-MoE: Test-Time Pruning as Micro-Grained Mixture-of-Experts
To tackle the huge computational demand of large foundation models, activation-aware compression techniques without retraining have been introduced. However, since these rely on calibration data, domain shift may arise for unknown downstream tasks. With a computationally efficient calibration, activation-aware pruning ...
[ "Toshiaki Koike-Akino", "Jing Liu", "Ye Wang" ]
[ "cs.LG", "cs.AI", "cs.CL" ]
[ "Computer Science" ]
2025-05-24T00:00:00
https://arxiv.org/abs/2505.18451
https://arxiv.org/pdf/2505.18451v1
2505.18451
10.48550/arXiv.2505.18451
3
0
false
null
arXiv.org
0.1707
49eda568f4355086ac1c9df76e3fed369b34ef50e87ea506eea52e69fa3a25af
[ "arxiv", "semantic_scholar" ]
CoMoE: Contrastive Representation for Mixture-of-Experts in Parameter-Efficient Fine-tuning
In parameter-efficient fine-tuning, mixture-of-experts (MoE), which involves specializing functionalities into different experts and sparsely activating them appropriately, has been widely adopted as a promising approach to trade-off between model capacity and computation overhead. However, current MoE variants fall sh...
[ "Jinyuan Feng", "Chaopeng Wei", "Tenghai Qiu", "Tianyi Hu", "Zhiqiang Pu" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2025-05-23T00:00:00
https://arxiv.org/abs/2505.17553
https://arxiv.org/pdf/2505.17553v2
2505.17553
10.48550/arXiv.2505.17553
2
0
false
null
Conference on Empirical Methods in Natural Language Processing
0.1696
cc6e8d4212cd1dd0c60401e14027458d29acf384350ea7b2a9370bbe9701495f
[ "arxiv", "semantic_scholar" ]
Not All Models Suit Expert Offloading: On Local Routing Consistency of Mixture-of-Expert Models
Mixture-of-Experts (MoE) enables efficient scaling of large language models (LLMs) with sparsely activated experts during inference. To effectively deploy large MoE models on memory-constrained devices, many systems introduce *expert offloading* that caches a subset of experts in fast memory, leaving others on slow mem...
[ "Jingcong Liang", "Siyuan Wang", "Miren Tian", "Yitong Li", "Duyu Tang", "Zhongyu Wei" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-05-21T00:00:00
https://arxiv.org/abs/2505.16056
https://arxiv.org/pdf/2505.16056v4
2505.16056
10.48550/arXiv.2505.16056
4
0
true
https://github.com/ljcleo/moe-lrc
arXiv.org
0.2585
3b4f9c0f6faef23099891e07a0634baf7a676d48e2535afec63a54f35d08a8a4
[ "arxiv", "semantic_scholar" ]
MoTE: Mixture of Task-specific Experts for Pre-Trained ModelBased Class-incremental Learning
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 remarkable success. However, prompt-based approaches suffer from prompt overwriting, whi...
[ "Linjie Li", "Zhenyu Wu", "Yang Ji" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-05-21T00:00:00
https://arxiv.org/abs/2506.11038
https://arxiv.org/pdf/2506.11038v1
2506.11038
10.1016/j.knosys.2025.113795
3
0
true
https://github.com/Franklilinjie/MoTE
Knowledge-Based Systems
0.2585
f02597040f3f6a8d34a96ee4385a8e3f316a41cdf0d985023200b80fd60027c1
[ "arxiv", "semantic_scholar" ]
THOR-MoE: Hierarchical Task-Guided and Context-Responsive Routing for Neural Machine Translation
The sparse Mixture-of-Experts (MoE) has achieved significant progress for neural machine translation (NMT). However, there exist two limitations in current MoE solutions which may lead to sub-optimal performance: 1) they directly use the task knowledge of NMT into MoE (\emph{e.g.}, domain/linguistics-specific knowledge...
[ "Yunlong Liang", "Fandong Meng", "Jie Zhou" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-20T00:00:00
https://arxiv.org/abs/2505.14173
https://arxiv.org/pdf/2505.14173v1
2505.14173
10.48550/arXiv.2505.14173
1
0
false
null
Annual Meeting of the Association for Computational Linguistics
0.1661
7f6262269245f0d07b1a685080557ac95b1353418925233d6aec3c72d11b3ee0
[ "arxiv", "semantic_scholar" ]
MoE-CAP: Benchmarking Cost, Accuracy and Performance of Sparse Mixture-of-Experts Systems
The sparse Mixture-of-Experts (MoE) architecture is increasingly favored for scaling Large Language Models (LLMs) efficiently, but it depends on heterogeneous compute and memory resources. These factors jointly affect system Cost, Accuracy, and Performance (CAP), making trade-offs inevitable. Existing benchmarks often ...
[ "Yinsicheng Jiang", "Yao Fu", "Yeqi Huang", "Ping Nie", "Zhan Lu", "Leyang Xue", "Congjie He", "Man-Kit Sit", "Jilong Xue", "Li Dong", "Ziming Miao", "Dayou Du", "Tairan Xu", "Kai Zou", "Edoardo Ponti", "Luo Mai" ]
[ "cs.LG", "cs.DC" ]
[ "Computer Science" ]
2025-05-16T00:00:00
https://arxiv.org/abs/2505.11415
https://arxiv.org/pdf/2505.11415v2
2505.11415
10.48550/arXiv.2505.11415
5
0
false
null
arXiv.org
0.1945
357454536d4c796eee8bb7f6bc1ea619ec57ff993dfef8eb47aeabf77adcf209
[ "arxiv", "semantic_scholar" ]
MegaScale-MoE: Large-Scale Communication-Efficient Training of Mixture-of-Experts Models in Production
We present MegaScale-MoE, a production system tailored for the efficient training of large-scale mixture-of-experts (MoE) models. MoE emerges as a promising architecture to scale large language models (LLMs) to unprecedented sizes, thereby enhancing model performance. However, existing MoE training systems experience a...
[ "Chao Jin", "Ziheng Jiang", "Zhihao Bai", "Zheng Zhong", "Juncai Liu", "Xiang Li", "Ningxin Zheng", "Xi Wang", "Cong Xie", "Qi Huang", "Wen Heng", "Yiyuan Ma", "Wenlei Bao", "Size Zheng", "Yanghua Peng", "Haibin Lin", "Xuanzhe Liu", "Xin Jin", "Xin Liu" ]
[ "cs.LG", "cs.DC" ]
[ "Computer Science" ]
2025-05-16T00:00:00
https://arxiv.org/abs/2505.11432
https://arxiv.org/pdf/2505.11432v3
2505.11432
10.48550/arXiv.2505.11432
25
1
false
null
European Conference on Computer Systems
0.3537
481906f3c7f5be018872996a1626f0dc3202145618ae021f1a184a1ce3b3506f
[ "arxiv", "semantic_scholar" ]
PT-MoE: An Efficient Finetuning Framework for Integrating Mixture-of-Experts into Prompt Tuning
Parameter-efficient fine-tuning (PEFT) methods have shown promise in adapting large language models, yet existing approaches exhibit counter-intuitive phenomena: integrating router into prompt tuning (PT) increases training efficiency yet does not improve performance universally; parameter reduction through matrix deco...
[ "Zongqian Li", "Yixuan Su", "Nigel Collier" ]
[ "cs.CL" ]
[ "Computer Science" ]
2025-05-14T00:00:00
https://arxiv.org/abs/2505.09519
https://arxiv.org/pdf/2505.09519v1
2505.09519
10.48550/arXiv.2505.09519
5
0
false
null
arXiv.org
0.1945
de12de7aeb43f6e33fdb6e1f5ebc017a0bf8872e6c6d73d27af8b84574ec116f
[ "arxiv", "semantic_scholar" ]
PWC-MoE: Privacy-Aware Wireless Collaborative Mixture of Experts
Large language models (LLMs) hosted on cloud servers alleviate the computational and storage burdens on local devices but raise privacy concerns due to sensitive data transmission and require substantial communication bandwidth, which is challenging in constrained environments. In contrast, small language models (SLMs)...
[ "Yang Su", "Na Yan", "Yansha Deng", "Robert Schober" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-05-13T00:00:00
https://arxiv.org/abs/2505.08719
https://arxiv.org/pdf/2505.08719v1
2505.08719
10.48550/arXiv.2505.08719
2
0
false
null
arXiv.org
0.1581
ebdf9e6b4e572b0c2a2488ec9b0a794e7da546ddf1e791858fde484d02b4e668
[ "arxiv", "semantic_scholar" ]
Faster MoE LLM Inference for Extremely Large Models
Sparse Mixture of Experts (MoE) large language models (LLMs) are gradually becoming the mainstream approach for ultra-large-scale models. Existing optimization efforts for MoE models have focused primarily on coarse-grained MoE architectures. With the emergence of DeepSeek Models, fine-grained MoE models are gaining po...
[ "Haoqi Yang", "Luohe Shi", "Qiwei Li", "Zuchao Li", "Ping Wang", "Bo Du", "Mengjia Shen", "Hai Zhao" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2025-05-06T00:00:00
https://arxiv.org/abs/2505.03531
https://arxiv.org/pdf/2505.03531v1
2505.03531
10.48550/arXiv.2505.03531
6
0
false
null
arXiv.org
0.2113
865fc294938d2a135ccef74d6a83d905234c1b0a8b636627e104cf8a0bf6c1b8
[ "arxiv", "semantic_scholar" ]
BadPatches: Routing-aware Backdoor Attacks on Vision Mixture of Experts
Mixture of Experts (MoE) architectures have gained popularity for reducing computational costs in deep neural networks by activating only a subset of parameters during inference. While this efficiency makes MoE attractive for vision tasks, the patch-based processing in vision models introduces new methods for adversari...
[ "Cedric Chan", "Jona te Lintelo", "Stjepan Picek" ]
[ "cs.CR" ]
[ "Computer Science" ]
2025-05-03T00:00:00
https://arxiv.org/abs/2505.01811
https://arxiv.org/pdf/2505.01811v3
2505.01811
null
0
0
false
null
null
0.0933
8b8a43d8b3be39541d857a30645bde8fbe8f19fd2bca5c29339ea5f94777de23
[ "arxiv", "semantic_scholar" ]
MoEQuant: Enhancing Quantization for Mixture-of-Experts Large Language Models via Expert-Balanced Sampling and Affinity Guidance
Mixture-of-Experts (MoE) large language models (LLMs), which leverage dynamic routing and sparse activation to enhance efficiency and scalability, have achieved higher performance while reducing computational costs. However, these models face significant memory overheads, limiting their practical deployment and broader...
[ "Xing Hu", "Zhixuan Chen", "Dawei Yang", "Zukang Xu", "Chen Xu", "Zhihang Yuan", "Sifan Zhou", "Jiangyong Yu" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2025-05-02T00:00:00
https://arxiv.org/abs/2505.03804
https://arxiv.org/pdf/2505.03804v1
2505.03804
10.48550/arXiv.2505.03804
27
4
false
null
International Conference on Machine Learning
0.3618
49d84ea516a58a2027c12fd553db197cd551e197155be319c39e951a8572ef28
[ "arxiv", "semantic_scholar" ]
Improving Routing in Sparse Mixture of Experts with Graph of Tokens
Sparse Mixture of Experts (SMoE) has emerged as a key to achieving unprecedented scalability in deep learning. By activating only a small subset of parameters per sample, SMoE achieves an exponential increase in parameter counts while maintaining a constant computational overhead. However, SMoE models are susceptible t...
[ "Tam Nguyen", "Ngoc N. Tran", "Khai Nguyen", "Richard G. Baraniuk" ]
[ "cs.LG" ]
[ "Computer Science" ]
2025-05-01T00:00:00
https://arxiv.org/abs/2505.00792
https://arxiv.org/pdf/2505.00792v1
2505.00792
10.48550/arXiv.2505.00792
5
1
false
null
arXiv.org
0.1945
5e30efb942712f326734ad86b70b49da828429274c808faaa835bf7957ce2a95
[ "arxiv", "semantic_scholar" ]
Mixture of Sparse Attention: Content-Based Learnable Sparse Attention via Expert-Choice Routing
Recent advances in large language models highlighted the excessive quadratic cost of self-attention. Despite the significant research efforts, subquadratic attention methods still suffer from inferior performance in practice. We hypothesize that dynamic, learned content-based sparsity can lead to more efficient attenti...
[ "Piotr Piękos", "Róbert Csordás", "Jürgen Schmidhuber" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2025-05-01T00:00:00
https://arxiv.org/abs/2505.00315
https://arxiv.org/pdf/2505.00315v1
2505.00315
10.48550/arXiv.2505.00315
10
0
false
null
arXiv.org
0.2603