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
edbcea7aad07e43aea079497303aae1e574fdcb60eff57d226b6c53c3cb7c002
[ "arxiv", "semantic_scholar" ]
A Survey on Mixture of Experts in Large Language Models
Large language models (LLMs) have garnered unprecedented advancements across diverse fields, ranging from natural language processing to computer vision and beyond. The prowess of LLMs is underpinned by their substantial model size, extensive and diverse datasets, and the vast computational power harnessed during train...
[ "Weilin Cai", "Juyong Jiang", "Fan Wang", "Jing Tang", "Sunghun Kim", "Jiayi Huang" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2024-06-26T00:00:00
https://arxiv.org/abs/2407.06204
https://arxiv.org/pdf/2407.06204v3
2407.06204
10.1109/TKDE.2025.3554028
326
19
true
https://github.com/withinmiaov/A-Survey-on-Mixture-of-Experts-in-LLMs
IEEE Transactions on Knowledge and Data Engineering
0.6505
b9895974b1451933f4d17cc2e145e14ab89622df61d035eb6cdfb784aa31d12a
[ "arxiv", "semantic_scholar" ]
LLaMA-MoE: Building Mixture-of-Experts from LLaMA with Continual Pre-training
Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, training MoE from scratch in a large-scale setting still suffers from data-hungry and instability problems. Motivated by this limit, we investigate building MoE models from existing d...
[ "Tong Zhu", "Xiaoye Qu", "Daize Dong", "Jiacheng Ruan", "Jingqi Tong", "Conghui He", "Yu Cheng" ]
[ "cs.CL" ]
[ "Computer Science" ]
2024-06-24T00:00:00
https://arxiv.org/abs/2406.16554
https://arxiv.org/pdf/2406.16554v1
2406.16554
10.48550/arXiv.2406.16554
157
9
true
https://github.com/pjlab-sys4nlp/llama-moe
Conference on Empirical Methods in Natural Language Processing
0.5497
f29ec0e5eefa07eddb35d026cea6bfbee0b3f1e040819b8b2277cb73f116de18
[ "arxiv", "semantic_scholar" ]
AdaMoE: Token-Adaptive Routing with Null Experts for Mixture-of-Experts Language Models
Mixture of experts (MoE) has become the standard for constructing production-level large language models (LLMs) due to its promise to boost model capacity without causing significant overheads. Nevertheless, existing MoE methods usually enforce a constant top-k routing for all tokens, which is arguably restrictive beca...
[ "Zihao Zeng", "Yibo Miao", "Hongcheng Gao", "Hao Zhang", "Zhijie Deng" ]
[ "cs.AI" ]
[ "Computer Science" ]
2024-06-19T00:00:00
https://arxiv.org/abs/2406.13233
https://arxiv.org/pdf/2406.13233v2
2406.13233
10.48550/arXiv.2406.13233
41
6
false
null
Conference on Empirical Methods in Natural Language Processing
0.4225
eff8c2fc1e5bb369d6b3730b1f9f757c2522ca9ecc740253bf493be47add59a7
[ "arxiv", "semantic_scholar" ]
$\texttt{MoE-RBench}$: Towards Building Reliable Language Models with Sparse Mixture-of-Experts
Mixture-of-Experts (MoE) has gained increasing popularity as a promising framework for scaling up large language models (LLMs). However, the reliability assessment of MoE lags behind its surging applications. Moreover, when transferred to new domains such as in fine-tuning MoE models sometimes underperform their dense ...
[ "Guanjie Chen", "Xinyu Zhao", "Tianlong Chen", "Yu Cheng" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2024-06-17T00:00:00
https://arxiv.org/abs/2406.11353
https://arxiv.org/pdf/2406.11353v1
2406.11353
10.48550/arXiv.2406.11353
9
1
true
https://github.com/UNITES-Lab/MoE-RBench
International Conference on Machine Learning
0.25
91ac4ea42fe5beeee310922eb00669f342befdf84b8d5a5845d0773406a28bfe
[ "arxiv", "semantic_scholar" ]
Self-MoE: Towards Compositional Large Language Models with Self-Specialized Experts
We present Self-MoE, an approach that transforms a monolithic LLM into a compositional, modular system of self-specialized experts, named MiXSE (MiXture of Self-specialized Experts). Our approach leverages self-specialization, which constructs expert modules using self-generated synthetic data, each equipping a shared ...
[ "Junmo Kang", "Leonid Karlinsky", "Hongyin Luo", "Zhen Wang", "Jacob Hansen", "James Glass", "David Cox", "Rameswar Panda", "Rogerio Feris", "Alan Ritter" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2024-06-17T00:00:00
https://arxiv.org/abs/2406.12034
https://arxiv.org/pdf/2406.12034v2
2406.12034
10.48550/arXiv.2406.12034
23
3
false
null
International Conference on Learning Representations
0.3451
0b6ea7b16a21ead3c53583c915bedd30c36e6dbba760a0203b05fbfaf3d5a2a4
[ "arxiv", "semantic_scholar" ]
QuantMoE-Bench: Examining Post-Training Quantization for Mixture-of-Experts
Mixture-of-Experts (MoE) is a promising way to scale up the learning capacity of large language models. It increases the number of parameters while keeping FLOPs nearly constant during inference through sparse activation. Yet, it still suffers from significant memory overheads due to the vast parameter size, necessitat...
[ "Pingzhi Li", "Xiaolong Jin", "Zhen Tan", "Yu Cheng", "Tianlong Chen" ]
[ "cs.LG", "cs.AI", "cs.CL" ]
[ "Computer Science" ]
2024-06-12T00:00:00
https://arxiv.org/abs/2406.08155
https://arxiv.org/pdf/2406.08155v2
2406.08155
null
6
0
true
https://github.com/UNITES-Lab/moe-quantization
null
0.2113
5f2dde92f83da20fa72e549d911848d7f257cc09bf738e9c6341e0e0acab3b44
[ "arxiv", "semantic_scholar" ]
MoE Jetpack: From Dense Checkpoints to Adaptive Mixture of Experts for Vision Tasks
The sparsely activated mixture of experts (MoE) model presents a promising alternative to traditional densely activated (dense) models, enhancing both quality and computational efficiency. However, training MoE models from scratch demands extensive data and computational resources. Moreover, public repositories like ti...
[ "Xingkui Zhu", "Yiran Guan", "Dingkang Liang", "Yuchao Chen", "Yuliang Liu", "Xiang Bai" ]
[ "cs.CV" ]
[ "Computer Science" ]
2024-06-07T00:00:00
https://arxiv.org/abs/2406.04801
https://arxiv.org/pdf/2406.04801v1
2406.04801
10.48550/arXiv.2406.04801
13
0
true
https://github.com/Adlith/MoE-Jetpack
Neural Information Processing Systems
0.2865
297a438584c7419ab4c88b7d87ed762a3830dea24feabb5e63b2cd8b641dc324
[ "arxiv", "semantic_scholar" ]
Skywork-MoE: A Deep Dive into Training Techniques for Mixture-of-Experts Language Models
In this technical report, we introduce the training methodologies implemented in the development of Skywork-MoE, a high-performance mixture-of-experts (MoE) large language model (LLM) with 146 billion parameters and 16 experts. It is initialized from the pre-existing dense checkpoints of our Skywork-13B model. We explo...
[ "Tianwen Wei", "Bo Zhu", "Liang Zhao", "Cheng Cheng", "Biye Li", "Weiwei Lü", "Peng Cheng", "Jianhao Zhang", "Xiaoyu Zhang", "Liang Zeng", "Xiaokun Wang", "Yutuan Ma", "Rui Hu", "Shuicheng Yan", "Han Fang", "Yahui Zhou" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2024-06-03T00:00:00
https://arxiv.org/abs/2406.06563
https://arxiv.org/pdf/2406.06563v1
2406.06563
10.48550/arXiv.2406.06563
60
5
false
null
arXiv.org
0.4463
d2455165de97ad49098297f3306a4e462e3dc04e2b1bafa6c1172867e7f19df8
[ "arxiv", "semantic_scholar" ]
MoNDE: Mixture of Near-Data Experts for Large-Scale Sparse Models
Mixture-of-Experts (MoE) large language models (LLM) have memory requirements that often exceed the GPU memory capacity, requiring costly parameter movement from secondary memories to the GPU for expert computation. In this work, we present Mixture of Near-Data Experts (MoNDE), a near-data computing solution that effic...
[ "Taehyun Kim", "Kwanseok Choi", "Youngmock Cho", "Jaehoon Cho", "Hyuk-Jae Lee", "Jaewoong Sim" ]
[ "cs.LG", "cs.AI", "cs.AR" ]
[ "Computer Science" ]
2024-05-29T00:00:00
https://arxiv.org/abs/2405.18832
https://arxiv.org/pdf/2405.18832v1
2405.18832
10.1145/3649329.3655951
14
2
false
null
Design Automation Conference
0.294
1fde663289f2242784dc1fed3ea3ad7ec151c113005e71fdba2784b73562a300
[ "arxiv", "semantic_scholar" ]
MEMoE: Enhancing Model Editing with Mixture of Experts Adaptors
Model editing aims to efficiently alter the behavior of Large Language Models (LLMs) within a desired scope, while ensuring no adverse impact on other inputs. Recent years have witnessed various model editing methods been proposed. However, these methods either exhibit poor overall performance or struggle to strike a b...
[ "Renzhi Wang", "Piji Li" ]
[ "cs.CL" ]
[ "Computer Science" ]
2024-05-29T00:00:00
https://arxiv.org/abs/2405.19086
https://arxiv.org/pdf/2405.19086v2
2405.19086
10.48550/arXiv.2405.19086
8
1
false
null
arXiv.org
0.2386
7ca1fd470586f8873d86418af3b2ee803c4293914f583687167c103466f1fd00
[ "arxiv", "semantic_scholar" ]
A Provably Effective Method for Pruning Experts in Fine-tuned Sparse Mixture-of-Experts
The sparsely gated mixture of experts (MoE) architecture sends different inputs to different subnetworks, i.e., experts, through trainable routers. MoE reduces the training computation significantly for large models, but its deployment can be still memory or computation expensive for some downstream tasks. Model prunin...
[ "Mohammed Nowaz Rabbani Chowdhury", "Meng Wang", "Kaoutar El Maghraoui", "Naigang Wang", "Pin-Yu Chen", "Christopher Carothers" ]
[ "cs.LG" ]
[ "Computer Science" ]
2024-05-26T00:00:00
https://arxiv.org/abs/2405.16646
https://arxiv.org/pdf/2405.16646v3
2405.16646
10.48550/arXiv.2405.16646
21
0
false
null
International Conference on Machine Learning
0.3356
38bc669a9e5f2deaeeac81b5ac515ae6b1939473b1611f126788e766c035e8e1
[ "arxiv", "semantic_scholar" ]
Expert-Token Resonance MoE: Bidirectional Routing with Efficiency Affinity-Driven Active Selection
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 routing that causes excessive communication overhead, and (2) expert homogeniza...
[ "Jing Li", "Zhijie Sun", "Dachao Lin", "Xuan He", "Binfan Zheng", "Yi Lin", "Rongqian Zhao", "Xin Chen" ]
[ "cs.CL" ]
[ "Computer Science" ]
2024-05-24T00:00:00
https://arxiv.org/abs/2406.00023
https://arxiv.org/pdf/2406.00023v4
2406.00023
null
2
0
false
null
null
0.1193
c6212034bd43735942ea322bff53b9b655ae8496c5ae2ee4cad64ba091c11d5e
[ "arxiv", "semantic_scholar" ]
Unchosen Experts Can Contribute Too: Unleashing MoE Models' Power by Self-Contrast
Mixture-of-Experts (MoE) has emerged as a prominent architecture for scaling model size while maintaining computational efficiency. In MoE, each token in the input sequence activates a different subset of experts determined by a routing mechanism. However, the unchosen experts in MoE models do not contribute to the out...
[ "Chufan Shi", "Cheng Yang", "Xinyu Zhu", "Jiahao Wang", "Taiqiang Wu", "Siheng Li", "Deng Cai", "Yujiu Yang", "Yu Meng" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2024-05-23T00:00:00
https://arxiv.org/abs/2405.14507
https://arxiv.org/pdf/2405.14507v2
2405.14507
10.48550/arXiv.2405.14507
25
2
false
null
Neural Information Processing Systems
0.3537
50713241b14d7debcebd2efcd720f90f716ed8338d182f1ff57360668ab7384b
[ "arxiv", "semantic_scholar" ]
Revisiting MoE and Dense Speed-Accuracy Comparisons for LLM Training
Mixture-of-Experts (MoE) enjoys performance gain by increasing model capacity while keeping computation cost constant. When comparing MoE to dense models, prior work typically adopt the following setting: 1) use FLOPs or activated parameters as a measure of model complexity; 2) train all models to the same number of to...
[ "Xianzhi Du", "Tom Gunter", "Xiang Kong", "Mark Lee", "Zirui Wang", "Aonan Zhang", "Nan Du", "Ruoming Pang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2024-05-23T00:00:00
https://arxiv.org/abs/2405.15052
https://arxiv.org/pdf/2405.15052v2
2405.15052
10.48550/arXiv.2405.15052
8
0
true
https://github.com/apple/axlearn}
arXiv.org
0.2386
b428310811cfe6eb56efcb8a86ee2338fa30579a149640f548fc3c9d21d213bf
[ "arxiv", "semantic_scholar" ]
Uni-MoE: Scaling Unified Multimodal LLMs with Mixture of Experts
Recent advancements in Multimodal Large Language Models (MLLMs) underscore the significance of scalable models and data to boost performance, yet this often incurs substantial computational costs. Although the Mixture of Experts (MoE) architecture has been employed to efficiently scale large language and image-text mod...
[ "Yunxin Li", "Shenyuan Jiang", "Baotian Hu", "Longyue Wang", "Wanqi Zhong", "Wenhan Luo", "Lin Ma", "Min Zhang" ]
[ "cs.AI", "cs.CL", "cs.CV", "cs.MM" ]
[ "Computer Science", "Medicine" ]
2024-05-18T00:00:00
https://arxiv.org/abs/2405.11273
https://arxiv.org/pdf/2405.11273v1
2405.11273
10.1109/TPAMI.2025.3532688
144
4
true
https://github.com/HITsz-TMG/UMOE-Scaling-Unified-Multimodal-LLMs
IEEE Transactions on Pattern Analysis and Machine Intelligence
0.5403
5beb2545a623914f7e90d57fc4811bb56190ef6ea0ca9d9e2f55045a33a730c6
[ "arxiv", "semantic_scholar" ]
A Mixture of Experts (MoE) model to improve AI-based computational pathology prediction performance under variable levels of histopathology image blur
AI-based models for histopathology whole slide image (WSI) analysis are increasingly common, but unsharp or blurred areas within WSI can significantly reduce prediction performance. In this study, we investigated the effect of image blur on deep learning models and introduced a mixture of experts (MoE) strategy that co...
[ "Yujie Xiang", "Bojing Liu", "Mattias Rantalainen" ]
[ "eess.IV", "cs.CV" ]
[ "Computer Science", "Engineering" ]
2024-05-15T00:00:00
https://arxiv.org/abs/2405.09298
https://arxiv.org/pdf/2405.09298v5
2405.09298
10.48550/arXiv.2405.09298
0
0
false
null
arXiv.org
0
249a5a4822bf1c83f6e3d35822b9296d0f5d6f5f82ca9eda481e940af348e43c
[ "arxiv", "semantic_scholar" ]
Mixture of insighTful Experts (MoTE): The Synergy of Thought Chains and Expert Mixtures in Self-Alignment
As the capabilities of large language models (LLMs) continue to expand, aligning these models with human values remains a significant challenge. Recent studies show that reasoning abilities contribute significantly to model safety, while integrating Mixture-of-Experts (MoE) architectures can further enhance alignment. ...
[ "Zhili Liu", "Yunhao Gou", "Kai Chen", "Lanqing Hong", "Jiahui Gao", "Fei Mi", "Yu Zhang", "Zhenguo Li", "Xin Jiang", "Qun Liu", "James T. Kwok" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2024-05-01T00:00:00
https://arxiv.org/abs/2405.00557
https://arxiv.org/pdf/2405.00557v5
2405.00557
10.18653/v1/2025.acl-long.151
13
0
false
null
Annual Meeting of the Association for Computational Linguistics
0.2865
9d2a1b65e55b45250e85efbfedce65ff9270ab3f71a8e3b6d9f041987912b2c5
[ "arxiv", "semantic_scholar" ]
Med-MoE: Mixture of Domain-Specific Experts for Lightweight Medical Vision-Language Models
Recent advancements in general-purpose or domain-specific multimodal large language models (LLMs) have witnessed remarkable progress for medical decision-making. However, they are designated for specific classification or generative tasks, and require model training or finetuning on large-scale datasets with sizeable p...
[ "Songtao Jiang", "Tuo Zheng", "Yan Zhang", "Yeying Jin", "Li Yuan", "Zuozhu Liu" ]
[ "cs.CV", "cs.CL" ]
[ "Computer Science" ]
2024-04-16T00:00:00
https://arxiv.org/abs/2404.10237
https://arxiv.org/pdf/2404.10237v3
2404.10237
10.18653/v1/2024.findings-emnlp.221
86
2
false
null
Conference on Empirical Methods in Natural Language Processing
0.4849
14751e718a4121bb245eca5ade0621839471a21a6006ea8485a7efbc8e34589c
[ "arxiv", "semantic_scholar" ]
MING-MOE: Enhancing Medical Multi-Task Learning in Large Language Models with Sparse Mixture of Low-Rank Adapter Experts
Large language models like ChatGPT have shown substantial progress in natural language understanding and generation, proving valuable across various disciplines, including the medical field. Despite advancements, challenges persist due to the complexity and diversity inherent in medical tasks which often require multi-...
[ "Yusheng Liao", "Shuyang Jiang", "Yu Wang", "Yanfeng Wang" ]
[ "cs.CL" ]
[ "Computer Science" ]
2024-04-13T00:00:00
https://arxiv.org/abs/2404.09027
https://arxiv.org/pdf/2404.09027v1
2404.09027
10.48550/arXiv.2404.09027
14
0
false
null
arXiv.org
0.294
c2836682c1cea45bb1d3fa93d741ccfe22a30e0855eeef674fafddfb0cf040bd
[ "arxiv", "semantic_scholar" ]
MoE-FFD: Mixture of Experts for Generalized and Parameter-Efficient Face Forgery Detection
Deepfakes have recently raised significant trust issues and security concerns among the public. Compared to CNN face forgery detectors, ViT-based methods take advantage of the expressivity of transformers, achieving superior detection performance. However, these approaches still exhibit the following limitations: (1) F...
[ "Chenqi Kong", "Anwei Luo", "Peijun Bao", "Yi Yu", "Haoliang Li", "Zengwei Zheng", "Shiqi Wang", "Alex C. Kot" ]
[ "cs.CV" ]
[ "Computer Science" ]
2024-04-12T00:00:00
https://arxiv.org/abs/2404.08452
https://arxiv.org/pdf/2404.08452v3
2404.08452
10.1109/TDSC.2025.3604443
54
6
true
https://github.com/LoveSiameseCat/MoE-FFD
IEEE Transactions on Dependable and Secure Computing
0.4351
8b0c2b6e0d92e6cf013940776f88e110ecb5fbf7252f2c44a3abebb551e8ef6b
[ "arxiv", "semantic_scholar" ]
Dense Training, Sparse Inference: Rethinking Training of Mixture-of-Experts Language Models
Mixture-of-Experts (MoE) language models can reduce computational costs by 2-4$\times$ compared to dense models without sacrificing performance, making them more efficient in computation-bounded scenarios. However, MoE models generally require 2-4$\times$ times more parameters to achieve comparable performance to a den...
[ "Bowen Pan", "Yikang Shen", "Haokun Liu", "Mayank Mishra", "Gaoyuan Zhang", "Aude Oliva", "Colin Raffel", "Rameswar Panda" ]
[ "cs.LG", "cs.AI", "cs.CL" ]
[ "Computer Science" ]
2024-04-08T00:00:00
https://arxiv.org/abs/2404.05567
https://arxiv.org/pdf/2404.05567v1
2404.05567
10.48550/arXiv.2404.05567
41
2
false
null
arXiv.org
0.4058
07ed2a685b37ac9d04adb203ec78277150b1e46b57c91e0f5cc585c8fac02004
[ "arxiv", "semantic_scholar" ]
SEER-MoE: Sparse Expert Efficiency through Regularization for Mixture-of-Experts
The advancement of deep learning has led to the emergence of Mixture-of-Experts (MoEs) models, known for their dynamic allocation of computational resources based on input. Despite their promise, MoEs face challenges, particularly in terms of memory requirements. To address this, our work introduces SEER-MoE, a novel t...
[ "Alexandre Muzio", "Alex Sun", "Churan He" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2024-04-07T00:00:00
https://arxiv.org/abs/2404.05089
https://arxiv.org/pdf/2404.05089v1
2404.05089
10.48550/arXiv.2404.05089
37
6
false
null
arXiv.org
0.4225
e8281e850740ea260ad1494faff1a9c7d76749c1d20bb89eaa39d5d25df0153a
[ "arxiv", "semantic_scholar" ]
Generalization Error Analysis for Sparse Mixture-of-Experts: A Preliminary Study
Mixture-of-Experts (MoE) represents an ensemble methodology that amalgamates predictions from several specialized sub-models (referred to as experts). This fusion is accomplished through a router mechanism, dynamically assigning weights to each expert's contribution based on the input data. Conventional MoE mechanisms ...
[ "Jinze Zhao", "Peihao Wang", "Zhangyang Wang" ]
[ "cs.LG" ]
[ "Computer Science" ]
2024-03-26T00:00:00
https://arxiv.org/abs/2403.17404
https://arxiv.org/pdf/2403.17404v1
2403.17404
10.48550/arXiv.2403.17404
6
0
false
null
arXiv.org
0.2113
9e0b1d2adcb48c613c434baea6dc308380adb82370d4e00c13540abab36d975d
[ "arxiv", "semantic_scholar" ]
An Expert is Worth One Token: Synergizing Multiple Expert LLMs as Generalist via Expert Token Routing
We present Expert-Token-Routing, a unified generalist framework that facilitates seamless integration of multiple expert LLMs. Our framework represents expert LLMs as special expert tokens within the vocabulary of a meta LLM. The meta LLM can route to an expert LLM like generating new tokens. Expert-Token-Routing not o...
[ "Ziwei Chai", "Guoyin Wang", "Jing Su", "Tianjie Zhang", "Xuanwen Huang", "Xuwu Wang", "Jingjing Xu", "Jianbo Yuan", "Hongxia Yang", "Fei Wu", "Yang Yang" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2024-03-25T00:00:00
https://arxiv.org/abs/2403.16854
https://arxiv.org/pdf/2403.16854v3
2403.16854
10.48550/arXiv.2403.16854
11
2
false
null
Annual Meeting of the Association for Computational Linguistics
0.2698
0f214d43fb2fc34d9edb056d4c5322fed4f7d21ed66e5fdb8702bb37bb1a38f3
[ "arxiv", "semantic_scholar" ]
Harder Tasks Need More Experts: Dynamic Routing in MoE Models
In this paper, we introduce a novel dynamic expert selection framework for Mixture of Experts (MoE) models, aiming to enhance computational efficiency and model performance by adjusting the number of activated experts based on input difficulty. Unlike traditional MoE approaches that rely on fixed Top-K routing, which a...
[ "Quzhe Huang", "Zhenwei An", "Nan Zhuang", "Mingxu Tao", "Chen Zhang", "Yang Jin", "Kun Xu", "Kun Xu", "Liwei Chen", "Songfang Huang", "Yansong Feng" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2024-03-12T00:00:00
https://arxiv.org/abs/2403.07652
https://arxiv.org/pdf/2403.07652v1
2403.07652
10.48550/arXiv.2403.07652
80
7
true
https://github.com/ZhenweiAn/Dynamic_MoE
arXiv.org
0.4771
e08f34e22377259558561a7d1968c7e2540af6f7b9655277e6daf2ba0b4f19c1
[ "arxiv", "semantic_scholar" ]
Branch-Train-MiX: Mixing Expert LLMs into a Mixture-of-Experts LLM
We investigate efficient methods for training Large Language Models (LLMs) to possess capabilities in multiple specialized domains, such as coding, math reasoning and world knowledge. Our method, named Branch-Train-MiX (BTX), starts from a seed model, which is branched to train experts in embarrassingly parallel fashio...
[ "Sainbayar Sukhbaatar", "Olga Golovneva", "Vasu Sharma", "Hu Xu", "Xi Victoria Lin", "Baptiste Rozière", "Jacob Kahn", "Daniel Li", "Wen-tau Yih", "Jason Weston", "Xian Li" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2024-03-12T00:00:00
https://arxiv.org/abs/2403.07816
https://arxiv.org/pdf/2403.07816v1
2403.07816
10.48550/arXiv.2403.07816
109
14
false
null
arXiv.org
0.588
d18f8774c918ca823c13e83395854647d6123fc17c62a0c093317c9d7160a07b
[ "arxiv", "semantic_scholar" ]
Video Relationship Detection Using Mixture of Experts
Machine comprehension of visual information from images and videos by neural networks faces two primary challenges. Firstly, there exists a computational and inference gap in connecting vision and language, making it difficult to accurately determine which object a given agent acts on and represent it through language....
[ "Ala Shaabana", "Zahra Gharaee", "Paul Fieguth" ]
[ "cs.CV", "cs.LG" ]
[ "Computer Science" ]
2024-03-06T00:00:00
https://arxiv.org/abs/2403.03994
https://arxiv.org/pdf/2403.03994v1
2403.03994
10.1109/ACCESS.2023.3257280
6
0
false
null
IEEE Access
0.2113
4db19b656ad2afffb9b54b9fecf50e80ad7426ead3a98a899e9ecd22abe2ffc4
[ "arxiv", "semantic_scholar" ]
Not All Experts are Equal: Efficient Expert Pruning and Skipping for Mixture-of-Experts Large Language Models
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 to deploy them due to their immense parameter sizes. Different from previous weig...
[ "Xudong Lu", "Qi Liu", "Yuhui Xu", "Aojun Zhou", "Siyuan Huang", "Bo Zhang", "Junchi Yan", "Hongsheng Li" ]
[ "cs.CL", "cs.AI", "cs.LG" ]
[ "Computer Science" ]
2024-02-22T00:00:00
https://arxiv.org/abs/2402.14800
https://arxiv.org/pdf/2402.14800v2
2402.14800
10.48550/arXiv.2402.14800
122
24
true
https://github.com/Lucky-Lance/Expert_Sparsity
Annual Meeting of the Association for Computational Linguistics
0.699
b95e1e4f0764d04c328d09924217651b2df92b3dcdb5c309d0cbc3a8cbb110c6
[ "arxiv", "semantic_scholar" ]
HyperMoE: Towards Better Mixture of Experts via Transferring Among Experts
The Mixture of Experts (MoE) for language models has been proven effective in augmenting the capacity of models by dynamically routing each input token to a specific subset of experts for processing. Despite the success, most existing methods face a challenge for balance between sparsity and the availability of expert ...
[ "Hao Zhao", "Zihan Qiu", "Huijia Wu", "Zili Wang", "Zhaofeng He", "Jie Fu" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2024-02-20T00:00:00
https://arxiv.org/abs/2402.12656
https://arxiv.org/pdf/2402.12656v4
2402.12656
10.18653/v1/2024.acl-long.571
28
0
false
null
Annual Meeting of the Association for Computational Linguistics
0.3656
8bf3cfeebdf26f6f1a0cd2563ab131292a6bd4889d9f964de107cc2539ad3c62
[ "arxiv", "semantic_scholar" ]
Multilinear Mixture of Experts: Scalable Expert Specialization through Factorization
The Mixture of Experts (MoE) paradigm provides a powerful way to decompose dense layers into smaller, modular computations often more amenable to human interpretation, debugging, and editability. However, a major challenge lies in the computational cost of scaling the number of experts high enough to achieve fine-grain...
[ "James Oldfield", "Markos Georgopoulos", "Grigorios G. Chrysos", "Christos Tzelepis", "Yannis Panagakis", "Mihalis A. Nicolaou", "Jiankang Deng", "Ioannis Patras" ]
[ "cs.CV", "cs.LG" ]
[ "Computer Science" ]
2024-02-19T00:00:00
https://arxiv.org/abs/2402.12550
https://arxiv.org/pdf/2402.12550v4
2402.12550
10.48550/arXiv.2402.12550
24
1
true
https://github.com/james-oldfield/muMoE
Neural Information Processing Systems
0.3495
5af02393c8be4a6b9ce9fa23e226443fbfb43b18d8702ae5f3c3d35a21df25c0
[ "arxiv", "semantic_scholar" ]
Buffer Overflow in Mixture of Experts
Mixture of Experts (MoE) has become a key ingredient for scaling large foundation models while keeping inference costs steady. We show that expert routing strategies that have cross-batch dependencies are vulnerable to attacks. Malicious queries can be sent to a model and can affect a model's output on other benign que...
[ "Jamie Hayes", "Ilia Shumailov", "Itay Yona" ]
[ "cs.CR", "cs.LG" ]
[ "Computer Science" ]
2024-02-08T00:00:00
https://arxiv.org/abs/2402.05526
https://arxiv.org/pdf/2402.05526v1
2402.05526
10.48550/arXiv.2402.05526
14
1
false
null
arXiv.org
0.294
50325cdf0492ef38b2fbe7f3113cc479d7bd565598de7645ff51b6bd1544193b
[ "arxiv", "semantic_scholar" ]
Approximation Rates and VC-Dimension Bounds for (P)ReLU MLP Mixture of Experts
Mixture-of-Experts (MoEs) can scale up beyond traditional deep learning models by employing a routing strategy in which each input is processed by a single "expert" deep learning model. This strategy allows us to scale up the number of parameters defining the MoE while maintaining sparse activation, i.e., MoEs only loa...
[ "Anastasis Kratsios", "Haitz Sáez de Ocáriz Borde", "Takashi Furuya", "Marc T. Law" ]
[ "stat.ML", "cs.LG", "cs.NE", "math.CO", "math.NA" ]
[ "Mathematics", "Computer Science" ]
2024-02-05T00:00:00
https://arxiv.org/abs/2402.03460
https://arxiv.org/pdf/2402.03460v2
2402.03460
null
1
0
false
null
null
0.0753
5743f4db6d90ea691ed873ad6999b7d2f6a2e34ae57b65103b6341aaf52670a2
[ "arxiv", "semantic_scholar" ]
CompeteSMoE -- Effective Training of Sparse Mixture of Experts via Competition
Sparse mixture of experts (SMoE) offers an appealing solution to scale up the model complexity beyond the mean of increasing the network's depth or width. However, effective training of SMoE has proven to be challenging due to the representation collapse issue, which causes parameter redundancy and limited representati...
[ "Quang Pham", "Giang Do", "Huy Nguyen", "TrungTin Nguyen", "Chenghao Liu", "Mina Sartipi", "Binh T. Nguyen", "Savitha Ramasamy", "Xiaoli Li", "Steven Hoi", "Nhat Ho" ]
[ "cs.LG" ]
[ "Computer Science" ]
2024-02-04T00:00:00
https://arxiv.org/abs/2402.02526
https://arxiv.org/pdf/2402.02526v1
2402.02526
10.48550/arXiv.2402.02526
24
2
false
null
arXiv.org
0.3495
a99b5a50e13d8a4db9632cbb45e7b5ff79f3afd643ec6cfa4a99530087d7d4a1
[ "arxiv", "semantic_scholar" ]
MoDE: A Mixture-of-Experts Model with Mutual Distillation among the Experts
The application of mixture-of-experts (MoE) is gaining popularity due to its ability to improve model's performance. In an MoE structure, the gate layer plays a significant role in distinguishing and routing input features to different experts. This enables each expert to specialize in processing their corresponding su...
[ "Zhitian Xie", "Yinger Zhang", "Chenyi Zhuang", "Qitao Shi", "Zhining Liu", "Jinjie Gu", "Guannan Zhang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2024-01-31T00:00:00
https://arxiv.org/abs/2402.00893
https://arxiv.org/pdf/2402.00893v1
2402.00893
10.48550/arXiv.2402.00893
23
1
false
null
AAAI Conference on Artificial Intelligence
0.3451
bb0e4b0f8cd968c7cc2526c5df3301bef928e173bd78344a47bd59f25d9fa23b
[ "arxiv", "semantic_scholar" ]
LLaVA-MoLE: Sparse Mixture of LoRA Experts for Mitigating Data Conflicts in Instruction Finetuning MLLMs
Instruction finetuning on a variety of image-text instruction data is the key to obtaining a versatile Multimodal Large Language Model (MLLM), and different configurations of the instruction data can lead to finetuned models with different capabilities. However, we have discovered that data conflicts are inevitable whe...
[ "Shaoxiang Chen", "Zequn Jie", "Lin Ma" ]
[ "cs.CV" ]
[ "Computer Science" ]
2024-01-29T00:00:00
https://arxiv.org/abs/2401.16160
https://arxiv.org/pdf/2401.16160v2
2401.16160
10.48550/arXiv.2401.16160
101
7
false
null
arXiv.org
0.5022
acc03edc62e4e3ab3d6201aae2d831a0fcf0a46b8b9cc96db8f35527091982ec
[ "arxiv", "semantic_scholar" ]
Routers in Vision Mixture of Experts: An Empirical Study
Mixture-of-Experts (MoE) models are a promising way to scale up model capacity without significantly increasing computational cost. A key component of MoEs is the router, which decides which subset of parameters (experts) process which feature embeddings (tokens). In this paper, we present a comprehensive study of rout...
[ "Tianlin Liu", "Mathieu Blondel", "Carlos Riquelme", "Joan Puigcerver" ]
[ "cs.CV", "cs.AI", "cs.LG" ]
[ "Computer Science" ]
2024-01-29T00:00:00
https://arxiv.org/abs/2401.15969
https://arxiv.org/pdf/2401.15969v2
2401.15969
10.48550/arXiv.2401.15969
21
0
false
null
null
0.3356
ff734668a04e79abf0cfb182d91fe1a7d94ec87c8718aa80717badfb0ce8f09f
[ "arxiv", "semantic_scholar" ]
MoE-LLaVA: Mixture of Experts for Large Vision-Language Models
Recent advances demonstrate that scaling Large Vision-Language Models (LVLMs) effectively improves downstream task performances. However, existing scaling methods enable all model parameters to be active for each token in the calculation, which brings massive training and inferring costs. In this work, we propose a sim...
[ "Bin Lin", "Zhenyu Tang", "Yang Ye", "Jinfa Huang", "Junwu Zhang", "Yatian Pang", "Peng Jin", "Munan Ning", "Jiebo Luo", "Li Yuan" ]
[ "cs.CV" ]
[ "Computer Science" ]
2024-01-29T00:00:00
https://arxiv.org/abs/2401.15947
https://arxiv.org/pdf/2401.15947v5
2401.15947
10.1109/TMM.2026.3654458
337
35
true
https://github.com/PKU-YuanGroup/MoE-LLaVA
IEEE transactions on multimedia
0.7782
0dceb431468b5b3d9d41021398602156e3e63f86cc3355adb0f6ee8097442ad6
[ "arxiv", "semantic_scholar" ]
MoE-Infinity: Efficient MoE Inference on Personal Machines with Sparsity-Aware Expert Cache
This paper presents MoE-Infinity, an efficient MoE inference system designed for personal machines with limited GPU memory capacity. The key idea for MoE-Infinity is that on personal machines, which are often single-user environments, MoE-based LLMs typically operate with a batch size of one. In this setting, MoE model...
[ "Leyang Xue", "Yao Fu", "Zhan Lu", "Luo Mai", "Mahesh Marina" ]
[ "cs.LG", "cs.PF" ]
[ "Computer Science" ]
2024-01-25T00:00:00
https://arxiv.org/abs/2401.14361
https://arxiv.org/pdf/2401.14361v3
2401.14361
null
16
4
true
https://github.com/EfficientMoE/MoE-Infinity
null
0.3495
e29183fbae54a78080d5e81f37bbab99a6d33b4eb279c1d4081e723f2ab10c81
[ "arxiv", "semantic_scholar" ]
Is Temperature Sample Efficient for Softmax Gaussian Mixture of Experts?
Dense-to-sparse gating mixture of experts (MoE) has recently become an effective alternative to a well-known sparse MoE. Rather than fixing the number of activated experts as in the latter model, which could limit the investigation of potential experts, the former model utilizes the temperature to control the softmax w...
[ "Huy Nguyen", "Pedram Akbarian", "Nhat Ho" ]
[ "stat.ML", "cs.LG" ]
[ "Computer Science", "Mathematics" ]
2024-01-25T00:00:00
https://arxiv.org/abs/2401.13875
https://arxiv.org/pdf/2401.13875v2
2401.13875
10.48550/arXiv.2401.13875
20
0
false
null
International Conference on Machine Learning
0.3306
9cf88d2e677dc5703a2b3e4ae68cd63cd81c016e73d4abf896f4244132e1989d
[ "arxiv", "semantic_scholar" ]
Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference
In large language models like the Generative Pre-trained Transformer, the Mixture of Experts paradigm has emerged as a powerful technique for enhancing model expressiveness and accuracy. However, deploying GPT MoE models for parallel inference on distributed systems presents significant challenges, primarily due to the...
[ "Jinghan Yao", "Quentin Anthony", "Aamir Shafi", "Hari Subramoni", "Dhabaleswar K.", " Panda" ]
[ "cs.LG", "cs.AI", "cs.DC" ]
[ "Computer Science" ]
2024-01-16T00:00:00
https://arxiv.org/abs/2401.08383
https://arxiv.org/pdf/2401.08383v2
2401.08383
10.1109/IPDPS57955.2024.00086
40
7
false
null
IEEE International Parallel and Distributed Processing Symposium
0.4515
1f701b3e1bae9ceb6e22d474cecdc14f9db106aeef4a020a446439ab73050647
[ "arxiv", "semantic_scholar" ]
MoE-Mamba: Efficient Selective State Space Models with Mixture of Experts
State Space Models (SSMs) have become serious contenders in the field of sequential modeling, challenging the dominance of Transformers. At the same time, Mixture of Experts (MoE) has significantly improved Transformer-based Large Language Models, including recent state-of-the-art open models. We propose that to unlock...
[ "Maciej Pióro", "Kamil Ciebiera", "Krystian Król", "Jan Ludziejewski", "Michał Krutul", "Jakub Krajewski", "Szymon Antoniak", "Piotr Miłoś", "Marek Cygan", "Sebastian Jaszczur" ]
[ "cs.LG", "cs.AI", "cs.CL" ]
[ "Computer Science" ]
2024-01-08T00:00:00
https://arxiv.org/abs/2401.04081
https://arxiv.org/pdf/2401.04081v2
2401.04081
10.48550/arXiv.2401.04081
97
5
false
null
arXiv.org
0.4978
10590fa10041f5b9dfcd5170653fdfa75fb8d9849d644369b769da63bb7e58d2
[ "arxiv", "semantic_scholar" ]
Mixtral of Experts
We introduce Mixtral 8x7B, a Sparse Mixture of Experts (SMoE) language model. Mixtral has the same architecture as Mistral 7B, with the difference that each layer is composed of 8 feedforward blocks (i.e. experts). For every token, at each layer, a router network selects two experts to process the current state and com...
[ "Albert Q. Jiang", "Alexandre Sablayrolles", "Antoine Roux", "Arthur Mensch", "Blanche Savary", "Chris Bamford", "Devendra Singh Chaplot", "Diego de las Casas", "Emma Bou Hanna", "Florian Bressand", "Gianna Lengyel", "Guillaume Bour", "Guillaume Lample", "Lélio Renard Lavaud", "Lucile Sa...
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2024-01-08T00:00:00
https://arxiv.org/abs/2401.04088
https://arxiv.org/pdf/2401.04088v1
2401.04088
10.48550/arXiv.2401.04088
1,950
189
false
null
arXiv.org
1
7f0bcbde324fa4f59cc67328e56cee937f1be0c45c253fbab0ae5652a743c429
[ "arxiv", "semantic_scholar" ]
HyperRouter: Towards Efficient Training and Inference of Sparse Mixture of Experts
By routing input tokens to only a few split experts, Sparse Mixture-of-Experts has enabled efficient training of large language models. Recent findings suggest that fixing the routers can achieve competitive performance by alleviating the collapsing problem, where all experts eventually learn similar representations. H...
[ "Giang Do", "Khiem Le", "Quang Pham", "TrungTin Nguyen", "Thanh-Nam Doan", "Bint T. Nguyen", "Chenghao Liu", "Savitha Ramasamy", "Xiaoli Li", "Steven Hoi" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2023-12-12T00:00:00
https://arxiv.org/abs/2312.07035
https://arxiv.org/pdf/2312.07035v1
2312.07035
10.18653/v1/2023.emnlp-main.351
26
6
true
https://github.com/giangdip2410/HyperRouter}}
Conference on Empirical Methods in Natural Language Processing
0.4225
daa8a223eaef66cc9a9ceb2c06f422c4d722122e137e9e284f4b52ee0c9278e4
[ "arxiv", "semantic_scholar" ]
GraphMETRO: Mitigating Complex Graph Distribution Shifts via Mixture of Aligned Experts
Graph data are inherently complex and heterogeneous, leading to a high natural diversity of distributional shifts. However, it remains unclear how to build machine learning architectures that generalize to the complex distributional shifts naturally occurring in the real world. Here, we develop GraphMETRO, a Graph Neur...
[ "Shirley Wu", "Kaidi Cao", "Bruno Ribeiro", "James Zou", "Jure Leskovec" ]
[ "cs.LG" ]
[ "Computer Science" ]
2023-12-07T00:00:00
https://arxiv.org/abs/2312.04693
https://arxiv.org/pdf/2312.04693v3
2312.04693
10.52202/079017-0297
23
2
true
https://github.com/Wuyxin/GraphMETRO
Neural Information Processing Systems
0.3451
14e3febb9c49f09a6d7c20f2a9af88e6424d757a3c49a4937d5e84ab58f56e4b
[ "arxiv", "semantic_scholar" ]
Memory Augmented Language Models through Mixture of Word Experts
Scaling up the number of parameters of language models has proven to be an effective approach to improve performance. For dense models, increasing model size proportionally increases the model's computation footprint. In this work, we seek to aggressively decouple learning capacity and FLOPs through Mixture-of-Experts ...
[ "Cicero Nogueira dos Santos", "James Lee-Thorp", "Isaac Noble", "Chung-Ching Chang", "David Uthus" ]
[ "cs.CL" ]
[ "Computer Science" ]
2023-11-15T00:00:00
https://arxiv.org/abs/2311.10768
https://arxiv.org/pdf/2311.10768v1
2311.10768
10.48550/arXiv.2311.10768
12
0
false
null
North American Chapter of the Association for Computational Linguistics
0.2785
4248e563c80a18759d78f686f64bec6a6419855036d6595eda3c44934e1eb298
[ "arxiv", "semantic_scholar" ]
SiDA-MoE: Sparsity-Inspired Data-Aware Serving for Efficient and Scalable Large Mixture-of-Experts Models
Mixture-of-Experts (MoE) has emerged as a favorable architecture in the era of large models due to its inherent advantage, i.e., enlarging model capacity without incurring notable computational overhead. Yet, the realization of such benefits often results in ineffective GPU memory utilization, as large portions of the ...
[ "Zhixu Du", "Shiyu Li", "Yuhao Wu", "Xiangyu Jiang", "Jingwei Sun", "Qilin Zheng", "Yongkai Wu", "Ang Li", "Hai \"Helen\" Li", "Yiran Chen" ]
[ "cs.LG", "cs.DC" ]
[ "Computer Science" ]
2023-10-29T00:00:00
https://arxiv.org/abs/2310.18859
https://arxiv.org/pdf/2310.18859v2
2310.18859
10.48550/arXiv.2310.18859
43
5
true
https://github.com/timlee0212/SiDA-MoE
Conference on Machine Learning and Systems
0.4109
c0f206350db0125de66bc11612ccd70946aed5a03699b884fa44432aa4b4683f
[ "arxiv", "semantic_scholar" ]
Mixture of Tokens: Continuous MoE through Cross-Example Aggregation
Mixture of Experts (MoE) models based on Transformer architecture are pushing the boundaries of language and vision tasks. The allure of these models lies in their ability to substantially increase the parameter count without a corresponding increase in FLOPs. Most widely adopted MoE models are discontinuous with respe...
[ "Szymon Antoniak", "Michał Krutul", "Maciej Pióro", "Jakub Krajewski", "Jan Ludziejewski", "Kamil Ciebiera", "Krystian Król", "Tomasz Odrzygóźdź", "Marek Cygan", "Sebastian Jaszczur" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2023-10-24T00:00:00
https://arxiv.org/abs/2310.15961
https://arxiv.org/pdf/2310.15961v2
2310.15961
10.52202/079017-3300
4
0
false
null
Neural Information Processing Systems
0.1747
19997a0c2f918ccbf60b93b2835e75572a886b3f8cd6910f19ccf7ceab774fcd
[ "arxiv", "semantic_scholar" ]
Merging Experts into One: Improving Computational Efficiency of Mixture of Experts
Scaling the size of language models usually leads to remarkable advancements in NLP tasks. But it often comes with a price of growing computational cost. Although a sparse Mixture of Experts (MoE) can reduce the cost by activating a small subset of parameters (e.g., one expert) for each input, its computation escalates...
[ "Shwai He", "Run-Ze Fan", "Liang Ding", "Li Shen", "Tianyi Zhou", "Dacheng Tao" ]
[ "cs.CL" ]
[ "Computer Science" ]
2023-10-15T00:00:00
https://arxiv.org/abs/2310.09832
https://arxiv.org/pdf/2310.09832v3
2310.09832
10.48550/arXiv.2310.09832
44
5
true
https://github.com/Shwai-He/MEO}
Conference on Empirical Methods in Natural Language Processing
0.4133
d5232ec70247029361361f639f48a9f30700001e551e5283c6448a4bbc0fa37a
[ "arxiv", "semantic_scholar" ]
Sparse Backpropagation for MoE Training
One defining characteristic of Mixture-of-Expert (MoE) models is their capacity for conducting sparse computation via expert routing, leading to remarkable scalability. However, backpropagation, the cornerstone of deep learning, requires dense computation, thereby posting challenges in MoE gradient computations. Here, ...
[ "Liyuan Liu", "Jianfeng Gao", "Weizhu Chen" ]
[ "cs.LG", "cs.AI", "cs.CL", "cs.CV" ]
[ "Computer Science" ]
2023-10-01T00:00:00
https://arxiv.org/abs/2310.00811
https://arxiv.org/pdf/2310.00811v1
2310.00811
10.48550/arXiv.2310.00811
18
0
false
null
arXiv.org
0.3197
2a194205da75cc3bdb3065d80bfc364a0aa5f175987d714b989be473f7dfb97d
[ "arxiv", "semantic_scholar" ]
Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts
Top-K sparse softmax gating mixture of experts has been widely used for scaling up massive deep-learning architectures without increasing the computational cost. Despite its popularity in real-world applications, the theoretical understanding of that gating function has remained an open problem. The main challenge come...
[ "Huy Nguyen", "Pedram Akbarian", "Fanqi Yan", "Nhat Ho" ]
[ "stat.ML", "cs.LG" ]
[ "Computer Science", "Mathematics" ]
2023-09-25T00:00:00
https://arxiv.org/abs/2309.13850
https://arxiv.org/pdf/2309.13850v2
2309.13850
10.48550/arXiv.2309.13850
31
0
false
null
International Conference on Learning Representations
0.3763
6a460675cb8309e557ef631d6fbdc4e5fb14de7b7f297cb38229d853e5e72748
[ "arxiv", "semantic_scholar" ]
Pushing Mixture of Experts to the Limit: Extremely Parameter Efficient MoE for Instruction Tuning
The Mixture of Experts (MoE) is a widely known neural architecture where an ensemble of specialized sub-models optimizes overall performance with a constant computational cost. However, conventional MoEs pose challenges at scale due to the need to store all experts in memory. In this paper, we push MoE to the limit. We...
[ "Ted Zadouri", "Ahmet Üstün", "Arash Ahmadian", "Beyza Ermiş", "Acyr Locatelli", "Sara Hooker" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2023-09-11T00:00:00
https://arxiv.org/abs/2309.05444
https://arxiv.org/pdf/2309.05444v1
2309.05444
10.48550/arXiv.2309.05444
176
19
true
https://github.com/for-ai/parameter-efficient-moe
International Conference on Learning Representations
0.6505
529a4852a009032b0d81dafd4007176efe81e3ec98696c7220f2f630c8a7a25f
[ "arxiv", "semantic_scholar" ]
Mobile V-MoEs: Scaling Down Vision Transformers via Sparse Mixture-of-Experts
Sparse Mixture-of-Experts models (MoEs) have recently gained popularity due to their ability to decouple model size from inference efficiency by only activating a small subset of the model parameters for any given input token. As such, sparse MoEs have enabled unprecedented scalability, resulting in tremendous successe...
[ "Erik Daxberger", "Floris Weers", "Bowen Zhang", "Tom Gunter", "Ruoming Pang", "Marcin Eichner", "Michael Emmersberger", "Yinfei Yang", "Alexander Toshev", "Xianzhi Du" ]
[ "cs.CV", "cs.LG", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2023-09-08T00:00:00
https://arxiv.org/abs/2309.04354
https://arxiv.org/pdf/2309.04354v1
2309.04354
10.48550/arXiv.2309.04354
19
0
false
null
arXiv.org
0.3253
f4d85c8780a3774cf1d4e2907100bfca389be06ed799ec9912bac09be891b9b0
[ "arxiv", "semantic_scholar" ]
Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference
Large language models (LLMs) based on transformers have made significant strides in recent years, the success of which is driven by scaling up their model size. Despite their high algorithmic performance, the computational and memory requirements of LLMs present unprecedented challenges. To tackle the high compute requ...
[ "Ranggi Hwang", "Jianyu Wei", "Shijie Cao", "Changho Hwang", "Xiaohu Tang", "Ting Cao", "Mao Yang" ]
[ "cs.LG", "cs.AI", "cs.AR" ]
[ "Computer Science" ]
2023-08-23T00:00:00
https://arxiv.org/abs/2308.12066
https://arxiv.org/pdf/2308.12066v3
2308.12066
10.1109/ISCA59077.2024.00078
126
13
false
null
International Symposium on Computer Architecture
0.5731
0988fa127abfcefd919f5baf9f3941aa8abde4913dafa9d7498e5f2c3b3c42bb
[ "arxiv", "semantic_scholar" ]
From Sparse to Soft Mixtures of Experts
Sparse mixture of expert architectures (MoEs) scale model capacity without significant increases in training or inference costs. Despite their success, MoEs suffer from a number of issues: training instability, token dropping, inability to scale the number of experts, or ineffective finetuning. In this work, we propose...
[ "Joan Puigcerver", "Carlos Riquelme", "Basil Mustafa", "Neil Houlsby" ]
[ "cs.LG", "cs.AI", "cs.CV" ]
[ "Computer Science" ]
2023-08-02T00:00:00
https://arxiv.org/abs/2308.00951
https://arxiv.org/pdf/2308.00951v2
2308.00951
10.48550/arXiv.2308.00951
282
22
false
null
International Conference on Learning Representations
0.6809
b6528947169a189fa5c1dbb31b229655d8e3d5153f608d2d4e6dd03debf1be35
[ "arxiv", "semantic_scholar" ]
Language-Routing Mixture of Experts for Multilingual and Code-Switching Speech Recognition
Multilingual speech recognition for both monolingual and code-switching speech is a challenging task. Recently, based on the Mixture of Experts (MoE), many works have made good progress in multilingual and code-switching ASR, but present huge computational complexity with the increase of supported languages. In this wo...
[ "Wenxuan Wang", "Guodong Ma", "Yuke Li", "Binbin Du" ]
[ "cs.SD", "eess.AS" ]
[ "Computer Science", "Engineering" ]
2023-07-12T00:00:00
https://arxiv.org/abs/2307.05956
https://arxiv.org/pdf/2307.05956v2
2307.05956
10.48550/arXiv.2307.05956
46
5
false
null
Interspeech
0.418
f9e702d107f50de36a9af4dbbab53596e249b084e5b68c6463e9455797bea33b
[ "arxiv", "semantic_scholar" ]
Patch-level Routing in Mixture-of-Experts is Provably Sample-efficient for Convolutional Neural Networks
In deep learning, mixture-of-experts (MoE) activates one or few experts (sub-networks) on a per-sample or per-token basis, resulting in significant computation reduction. The recently proposed \underline{p}atch-level routing in \underline{MoE} (pMoE) divides each input into $n$ patches (or tokens) and sends $l$ patches...
[ "Mohammed Nowaz Rabbani Chowdhury", "Shuai Zhang", "Meng Wang", "Sijia Liu", "Pin-Yu Chen" ]
[ "cs.LG" ]
[ "Computer Science" ]
2023-06-07T00:00:00
https://arxiv.org/abs/2306.04073
https://arxiv.org/pdf/2306.04073v1
2306.04073
10.48550/arXiv.2306.04073
47
3
false
null
International Conference on Machine Learning
0.4203
17fece3a8ff526c85ca5c987482d88006b12b4a9b4a6449e04a0611c31206dfd
[ "arxiv", "semantic_scholar" ]
Soft Merging of Experts with Adaptive Routing
Sparsely activated neural networks with conditional computation learn to route their inputs through different "expert" subnetworks, providing a form of modularity that densely activated models lack. Despite their possible benefits, models with learned routing often underperform their parameter-matched densely activated...
[ "Mohammed Muqeeth", "Haokun Liu", "Colin Raffel" ]
[ "cs.LG" ]
[ "Computer Science" ]
2023-06-06T00:00:00
https://arxiv.org/abs/2306.03745
https://arxiv.org/pdf/2306.03745v2
2306.03745
10.48550/arXiv.2306.03745
95
8
false
null
null
0.4956
17f55ed1609682d4c8d2f52d4136145573dc721a552053c6f73931e7a227a991
[ "arxiv", "semantic_scholar" ]
LightCurve MoE: A Dynamic Sparse Routing Mixture-of-Experts Architecture for Efficient Stellar Light Curve Classification
The classification of stellar light curves has become a key task in modern time-domain astronomy, fueled by the rapid growth of data from large-scale surveys such as Kepler and TESS. Although deep learning models have achieved high accuracy in this area, their computational costs can limit scalability. To tackle this i...
[ "Cunshi Wang", "Yu Bai", "Xinrui Song", "Jiacheng Xu", "Henggeng Han", "Yuyang Li", "Xinjie Hu", "Huiqin Yang", "Jifeng Liu" ]
[ "astro-ph.IM", "astro-ph.SR" ]
[ "Physics" ]
2023-05-23T00:00:00
https://arxiv.org/abs/2305.13745
https://arxiv.org/pdf/2305.13745v3
2305.13745
10.1088/1674-4527/adfa73
4
0
false
null
null
0.1747
0e298a946d639759e91058921e334dbf9e6e09b73c1e29f9ddc7260202973f39
[ "arxiv", "semantic_scholar" ]
Pipeline MoE: A Flexible MoE Implementation with Pipeline Parallelism
The Mixture of Experts (MoE) model becomes an important choice of large language models nowadays because of its scalability with sublinear computational complexity for training and inference. However, existing MoE models suffer from two critical drawbacks, 1) tremendous inner-node and inter-node communication overhead ...
[ "Xin Chen", "Hengheng Zhang", "Xiaotao Gu", "Kaifeng Bi", "Lingxi Xie", "Qi Tian" ]
[ "cs.DC", "cs.LG" ]
[ "Computer Science" ]
2023-04-22T00:00:00
https://arxiv.org/abs/2304.11414
https://arxiv.org/pdf/2304.11414v1
2304.11414
10.48550/arXiv.2304.11414
6
0
false
null
arXiv.org
0.2113
18f4a8a8dcf1f82f9fc257fc967fa23deac54e9c63c8757e1ee21ba5852cca20
[ "arxiv", "semantic_scholar" ]
WM-MoE: Weather-aware Multi-scale Mixture-of-Experts for Blind Adverse Weather Removal
Adverse weather removal tasks like deraining, desnowing, and dehazing are usually treated as separate tasks. However, in practical autonomous driving scenarios, the type, intensity,and mixing degree of weather are unknown, so handling each task separately cannot deal with the complex practical scenarios. In this paper,...
[ "Yulin Luo", "Rui Zhao", "Xiaobao Wei", "Jinwei Chen", "Yijie Lu", "Shenghao Xie", "Tianyu Wang", "Ruiqin Xiong", "Ming Lu", "Shanghang Zhang" ]
[ "cs.CV" ]
[ "Computer Science" ]
2023-03-24T00:00:00
https://arxiv.org/abs/2303.13739
https://arxiv.org/pdf/2303.13739v2
2303.13739
null
10
0
false
null
null
0.2603
d9e5a800e30b28a8996607942ef10e2df89b0c0e3031365210b6b04d9bfd2d9e
[ "arxiv", "semantic_scholar" ]
Scaling Vision-Language Models with Sparse Mixture of Experts
The field of natural language processing (NLP) has made significant strides in recent years, particularly in the development of large-scale vision-language models (VLMs). These models aim to bridge the gap between text and visual information, enabling a more comprehensive understanding of multimedia data. However, as t...
[ "Sheng Shen", "Zhewei Yao", "Chunyuan Li", "Trevor Darrell", "Kurt Keutzer", "Yuxiong He" ]
[ "cs.CV", "cs.CL" ]
[ "Computer Science" ]
2023-03-13T00:00:00
https://arxiv.org/abs/2303.07226
https://arxiv.org/pdf/2303.07226v1
2303.07226
10.48550/arXiv.2303.07226
116
3
false
null
Conference on Empirical Methods in Natural Language Processing
0.517
812cc422f79d2390ebb082e97378bd2f6cced4fb64910b98506c740e72c5f127
[ "arxiv", "semantic_scholar" ]
Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference
Mixture-of-Experts (MoE) models have gained popularity in achieving state-of-the-art performance in a wide range of tasks in computer vision and natural language processing. They effectively expand the model capacity while incurring a minimal increase in computation cost during training. However, deploying such models ...
[ "Haiyang Huang", "Newsha Ardalani", "Anna Sun", "Liu Ke", "Hsien-Hsin S. Lee", "Anjali Sridhar", "Shruti Bhosale", "Carole-Jean Wu", "Benjamin Lee" ]
[ "cs.DC", "cs.AR", "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2023-03-10T00:00:00
https://arxiv.org/abs/2303.06182
https://arxiv.org/pdf/2303.06182v2
2303.06182
10.48550/arXiv.2303.06182
47
8
false
null
arXiv.org
0.4771
88664b6ec4aee1f109552fba913030581bd317e4f2e6a7f64894d34bd8ec2552
[ "arxiv", "semantic_scholar" ]
Improving Expert Specialization in Mixture of Experts
Mixture of experts (MoE), introduced over 20 years ago, is the simplest gated modular neural network architecture. There is renewed interest in MoE because the conditional computation allows only parts of the network to be used during each inference, as was recently demonstrated in large scale natural language processi...
[ "Yamuna Krishnamurthy", "Chris Watkins", "Thomas Gaertner" ]
[ "cs.LG", "cs.AI", "cs.NE" ]
[ "Computer Science" ]
2023-02-28T00:00:00
https://arxiv.org/abs/2302.14703
https://arxiv.org/pdf/2302.14703v1
2302.14703
10.48550/arXiv.2302.14703
22
0
false
null
arXiv.org
0.3404
d202c846ad5eee7d54bba0e9692bba3c7e8535cecf1c66eedde76a0bb221973b
[ "arxiv", "semantic_scholar" ]
TA-MoE: Topology-Aware Large Scale Mixture-of-Expert Training
Sparsely gated Mixture-of-Expert (MoE) has demonstrated its effectiveness in scaling up deep neural networks to an extreme scale. Despite that numerous efforts have been made to improve the performance of MoE from the model design or system optimization perspective, existing MoE dispatch patterns are still not able to ...
[ "Chang Chen", "Min Li", "Zhihua Wu", "Dianhai Yu", "Chao Yang" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2023-02-20T00:00:00
https://arxiv.org/abs/2302.09915
https://arxiv.org/pdf/2302.09915v1
2302.09915
10.48550/arXiv.2302.09915
27
2
false
null
Neural Information Processing Systems
0.3618
e93f2f5d9ab1192e16b6a67eef786dbffbb477b46073163026e0e0a1eac6308f
[ "arxiv", "semantic_scholar" ]
SMILE: Scaling Mixture-of-Experts with Efficient Bi-level Routing
The mixture of Expert (MoE) parallelism is a recent advancement that scales up the model size with constant computational cost. MoE selects different sets of parameters (i.e., experts) for each incoming token, resulting in a sparsely-activated model. Despite several successful applications of MoE, its training efficien...
[ "Chaoyang He", "Shuai Zheng", "Aston Zhang", "George Karypis", "Trishul Chilimbi", "Mahdi Soltanolkotabi", "Salman Avestimehr" ]
[ "cs.LG" ]
[ "Computer Science" ]
2022-12-10T00:00:00
https://arxiv.org/abs/2212.05191
https://arxiv.org/pdf/2212.05191v1
2212.05191
10.48550/arXiv.2212.05191
5
0
false
null
arXiv.org
0.1945
fb74f3d655b4e988d1c7f345783ce60352c39ad886d609ac01d7c93c3d261b0f
[ "arxiv", "semantic_scholar" ]
MegaBlocks: Efficient Sparse Training with Mixture-of-Experts
We present MegaBlocks, a system for efficient Mixture-of-Experts (MoE) training on GPUs. Our system is motivated by the limitations of current frameworks, which restrict the dynamic routing in MoE layers to satisfy the constraints of existing software and hardware. These formulations force a tradeoff between model qual...
[ "Trevor Gale", "Deepak Narayanan", "Cliff Young", "Matei Zaharia" ]
[ "cs.LG", "cs.AI", "cs.DC" ]
[ "Computer Science" ]
2022-11-29T00:00:00
https://arxiv.org/abs/2211.15841
https://arxiv.org/pdf/2211.15841v1
2211.15841
10.48550/arXiv.2211.15841
221
21
false
null
Conference on Machine Learning and Systems
0.6712
3ee3889efa32b44d470cad3406ccb16213a6e19caeb1a4c4eb238d76714d3206
[ "arxiv", "semantic_scholar" ]
Handling Trade-Offs in Speech Separation with Sparsely-Gated Mixture of Experts
Employing a monaural speech separation (SS) model as a front-end for automatic speech recognition (ASR) involves balancing two kinds of trade-offs. First, while a larger model improves the SS performance, it also requires a higher computational cost. Second, an SS model that is more optimized for handling overlapped sp...
[ "Xiaofei Wang", "Zhuo Chen", "Yu Shi", "Jian Wu", "Naoyuki Kanda", "Takuya Yoshioka" ]
[ "eess.AS", "cs.SD", "eess.SP" ]
[ "Engineering", "Computer Science" ]
2022-11-11T00:00:00
https://arxiv.org/abs/2211.06493
https://arxiv.org/pdf/2211.06493v2
2211.06493
null
2
0
false
null
null
0.1193
023f0bb174237258d52133d7487f5e278cfb03546ae857b44938a4c88312b9b7
[ "arxiv", "semantic_scholar" ]
On the Adversarial Robustness of Mixture of Experts
Adversarial robustness is a key desirable property of neural networks. It has been empirically shown to be affected by their sizes, with larger networks being typically more robust. Recently, Bubeck and Sellke proved a lower bound on the Lipschitz constant of functions that fit the training data in terms of their numbe...
[ "Joan Puigcerver", "Rodolphe Jenatton", "Carlos Riquelme", "Pranjal Awasthi", "Srinadh Bhojanapalli" ]
[ "cs.LG", "cs.AI", "cs.CR", "cs.CV" ]
[ "Computer Science" ]
2022-10-19T00:00:00
https://arxiv.org/abs/2210.10253
https://arxiv.org/pdf/2210.10253v1
2210.10253
10.48550/arXiv.2210.10253
29
1
false
null
Neural Information Processing Systems
0.3693
a4b1d1639c6a61345d5b2f8157bcab5bbe27516fef3175e1a193795f65204cce
[ "arxiv", "semantic_scholar" ]
Towards Understanding Mixture of Experts in Deep Learning
The Mixture-of-Experts (MoE) layer, a sparsely-activated model controlled by a router, has achieved great success in deep learning. However, the understanding of such architecture remains elusive. In this paper, we formally study how the MoE layer improves the performance of neural network learning and why the mixture ...
[ "Zixiang Chen", "Yihe Deng", "Yue Wu", "Quanquan Gu", "Yuanzhi Li" ]
[ "cs.LG", "cs.AI", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2022-08-04T00:00:00
https://arxiv.org/abs/2208.02813
https://arxiv.org/pdf/2208.02813v1
2208.02813
10.48550/arXiv.2208.02813
90
3
false
null
arXiv.org
0.4898
f975165a5db65000bdaa39fe2dea19c55931357e72abdf06a2c90d21f68fe909
[ "arxiv", "semantic_scholar" ]
MoEC: Mixture of Expert Clusters
Sparsely Mixture of Experts (MoE) has received great interest due to its promising scaling capability with affordable computational overhead. MoE converts dense layers into sparse experts, and utilizes a gated routing network to make experts conditionally activated. However, as the number of experts grows, MoE with out...
[ "Yuan Xie", "Shaohan Huang", "Tianyu Chen", "Furu Wei" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2022-07-19T00:00:00
https://arxiv.org/abs/2207.09094
https://arxiv.org/pdf/2207.09094v1
2207.09094
10.48550/arXiv.2207.09094
27
0
false
null
AAAI Conference on Artificial Intelligence
0.3618
c785f0d799b0db5b0286520d8d0b502db5a6d404adf607914853dc2446490cde
[ "arxiv", "semantic_scholar" ]
Task-Specific Expert Pruning for Sparse Mixture-of-Experts
The sparse Mixture-of-Experts (MoE) model is powerful for large-scale pre-training and has achieved promising results due to its model capacity. However, with trillions of parameters, MoE is hard to be deployed on cloud or mobile environment. The inference of MoE requires expert parallelism, which is not hardware-frien...
[ "Tianyu Chen", "Shaohan Huang", "Yuan Xie", "Binxing Jiao", "Daxin Jiang", "Haoyi Zhou", "Jianxin Li", "Furu Wei" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2022-06-01T00:00:00
https://arxiv.org/abs/2206.00277
https://arxiv.org/pdf/2206.00277v2
2206.00277
10.48550/arXiv.2206.00277
81
5
false
null
arXiv.org
0.4785
b108a3b5051a5417e6fe5c7898d2eab91572f3c49cf0daa98683a282e875cf60
[ "arxiv", "semantic_scholar" ]
Sparse Mixers: Combining MoE and Mixing to build a more efficient BERT
We combine the capacity of sparsely gated Mixture-of-Experts (MoE) with the speed and stability of linear, mixing transformations to design the Sparse Mixer encoder model. Sparse Mixer slightly outperforms (<1%) BERT on GLUE and SuperGLUE, but more importantly trains 65% faster and runs inference 61% faster. We also pr...
[ "James Lee-Thorp", "Joshua Ainslie" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2022-05-24T00:00:00
https://arxiv.org/abs/2205.12399
https://arxiv.org/pdf/2205.12399v2
2205.12399
10.48550/arXiv.2205.12399
15
0
false
null
Conference on Empirical Methods in Natural Language Processing
0.301
c914af26731e1c154a1801eccc32dfceee33706f6c12f65361c6babde60f7377
[ "arxiv", "semantic_scholar" ]
Sparsely-gated Mixture-of-Expert Layers for CNN Interpretability
Sparsely-gated Mixture of Expert (MoE) layers have been recently successfully applied for scaling large transformers, especially for language modeling tasks. An intriguing side effect of sparse MoE layers is that they convey inherent interpretability to a model via natural expert specialization. In this work, we apply ...
[ "Svetlana Pavlitska", "Christian Hubschneider", "Lukas Struppek", "J. Marius Zöllner" ]
[ "cs.CV", "cs.LG" ]
[ "Computer Science" ]
2022-04-22T00:00:00
https://arxiv.org/abs/2204.10598
https://arxiv.org/pdf/2204.10598v3
2204.10598
10.1109/IJCNN54540.2023.10191904
21
2
false
null
IEEE International Joint Conference on Neural Network
0.3356
121e40f91caf82937fb29f7494d54eb4789939a4cc40d349bf445d10481225fd
[ "arxiv", "semantic_scholar" ]
On the Representation Collapse of Sparse Mixture of Experts
Sparse mixture of experts provides larger model capacity while requiring a constant computational overhead. It employs the routing mechanism to distribute input tokens to the best-matched experts according to their hidden representations. However, learning such a routing mechanism encourages token clustering around exp...
[ "Zewen Chi", "Li Dong", "Shaohan Huang", "Damai Dai", "Shuming Ma", "Barun Patra", "Saksham Singhal", "Payal Bajaj", "Xia Song", "Xian-Ling Mao", "Heyan Huang", "Furu Wei" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2022-04-20T00:00:00
https://arxiv.org/abs/2204.09179
https://arxiv.org/pdf/2204.09179v3
2204.09179
10.48550/arXiv.2204.09179
188
23
false
null
Neural Information Processing Systems
0.6901
963ffb520cf2c7d94f0deab834e4b2e64ca38ae1001fb253f7bd1e06ff2e5326
[ "arxiv", "semantic_scholar" ]
Residual Mixture of Experts
Mixture of Experts (MoE) is able to scale up vision transformers effectively. However, it requires prohibiting computation resources to train a large MoE transformer. In this paper, we propose Residual Mixture of Experts (RMoE), an efficient training pipeline for MoE vision transformers on downstream tasks, such as seg...
[ "Lemeng Wu", "Mengchen Liu", "Yinpeng Chen", "Dongdong Chen", "Xiyang Dai", "Lu Yuan" ]
[ "cs.CV" ]
[ "Computer Science" ]
2022-04-20T00:00:00
https://arxiv.org/abs/2204.09636
https://arxiv.org/pdf/2204.09636v3
2204.09636
10.48550/arXiv.2204.09636
54
4
false
null
arXiv.org
0.4351
9bcf711e10fdafbeb24c8e0df8ae51ed860dbcc11a31a92b5c705e597e18ae9f
[ "arxiv", "semantic_scholar" ]
StableMoE: Stable Routing Strategy for Mixture of Experts
The Mixture-of-Experts (MoE) technique can scale up the model size of Transformers with an affordable computational overhead. We point out that existing learning-to-route MoE methods suffer from the routing fluctuation issue, i.e., the target expert of the same input may change along with training, but only one expert ...
[ "Damai Dai", "Li Dong", "Shuming Ma", "Bo Zheng", "Zhifang Sui", "Baobao Chang", "Furu Wei" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2022-04-18T00:00:00
https://arxiv.org/abs/2204.08396
https://arxiv.org/pdf/2204.08396v1
2204.08396
10.48550/arXiv.2204.08396
123
8
false
null
Annual Meeting of the Association for Computational Linguistics
0.5234
717325a1b7a7d890848626432b01bd84d564f2bdc273db4a9359329cb271ed39
[ "arxiv", "semantic_scholar" ]
Sparsely Activated Mixture-of-Experts are Robust Multi-Task Learners
Traditional multi-task learning (MTL) methods use dense networks that use the same set of shared weights across several different tasks. This often creates interference where two or more tasks compete to pull model parameters in different directions. In this work, we study whether sparsely activated Mixture-of-Experts ...
[ "Shashank Gupta", "Subhabrata Mukherjee", "Krishan Subudhi", "Eduardo Gonzalez", "Damien Jose", "Ahmed H. Awadallah", "Jianfeng Gao" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2022-04-16T00:00:00
https://arxiv.org/abs/2204.07689
https://arxiv.org/pdf/2204.07689v1
2204.07689
10.48550/arXiv.2204.07689
64
1
false
null
arXiv.org
0.4532
c12259d573f08635e2e84281eb5503d30f9baf10c4b5989acd4c155dc6404731
[ "arxiv", "semantic_scholar" ]
Mixture of Experts for Biomedical Question Answering
Biomedical Question Answering (BQA) has attracted increasing attention in recent years due to its promising application prospect. It is a challenging task because the biomedical questions are professional and usually vary widely. Existing question answering methods answer all questions with a homogeneous model, leading...
[ "Damai Dai", "Wenbin Jiang", "Jiyuan Zhang", "Weihua Peng", "Yajuan Lyu", "Zhifang Sui", "Baobao Chang", "Yong Zhu" ]
[ "cs.CL" ]
[ "Computer Science" ]
2022-04-15T00:00:00
https://arxiv.org/abs/2204.07469
https://arxiv.org/pdf/2204.07469v1
2204.07469
10.48550/arXiv.2204.07469
13
0
false
null
Natural Language Processing and Chinese Computing
0.2865
a0898c4a33d8375f6654e6d7924008211d10197295e70848fa2d92f8f3fdbe30
[ "arxiv", "semantic_scholar" ]
Build a Robust QA System with Transformer-based Mixture of Experts
In this paper, we aim to build a robust question answering system that can adapt to out-of-domain datasets. A single network may overfit to the superficial correlation in the training distribution, but with a meaningful number of expert sub-networks, a gating network that selects a sparse combination of experts for eac...
[ "Yu Qing Zhou", "Xixuan Julie Liu", "Yuanzhe Dong" ]
[ "cs.CL", "cs.AI" ]
[ "Computer Science" ]
2022-03-20T00:00:00
https://arxiv.org/abs/2204.09598
https://arxiv.org/pdf/2204.09598v1
2204.09598
10.48550/arXiv.2204.09598
1
0
false
null
arXiv.org
0.0753
c33bfa34eb3a42efdddcdcee04ccdc1c58dd7b9f095e4b90cbde9f500c5c308e
[ "arxiv", "semantic_scholar" ]
Support Recovery in Mixture Models with Sparse Parameters
Mixture models are widely used to fit complex and multimodal datasets. In this paper we study mixtures with high dimensional sparse latent parameter vectors and consider the problem of support recovery of those vectors. While parameter learning in mixture models is well-studied, the sparsity constraint remains relative...
[ "Arya Mazumdar", "Soumyabrata Pal" ]
[ "cs.LG", "cs.IT", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2022-02-24T00:00:00
https://arxiv.org/abs/2202.11940
https://arxiv.org/pdf/2202.11940v2
2202.11940
10.1109/TIT.2024.3462937
1
0
false
null
IEEE Transactions on Information Theory
0.0753
6c4668e251cb0da1c5faf5d4eb06f2762f6a78d302ccef6cf9331f1112cfd821
[ "arxiv", "semantic_scholar" ]
Mixture-of-Experts with Expert Choice Routing
Sparsely-activated Mixture-of-experts (MoE) models allow the number of parameters to greatly increase while keeping the amount of computation for a given token or a given sample unchanged. However, a poor expert routing strategy (e.g. one resulting in load imbalance) can cause certain experts to be under-trained, leadi...
[ "Yanqi Zhou", "Tao Lei", "Hanxiao Liu", "Nan Du", "Yanping Huang", "Vincent Zhao", "Andrew Dai", "Zhifeng Chen", "Quoc Le", "James Laudon" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2022-02-18T00:00:00
https://arxiv.org/abs/2202.09368
https://arxiv.org/pdf/2202.09368v2
2202.09368
10.52202/068431-0515
775
43
false
null
Neural Information Processing Systems
0.8217
2aa3a854199d5839fd2b0d868d14d135b30fa1ced20db558570dfca5e6a6bf7b
[ "arxiv", "semantic_scholar" ]
ST-MoE: Designing Stable and Transferable Sparse Expert Models
Scale has opened new frontiers in natural language processing -- but at a high cost. In response, Mixture-of-Experts (MoE) and Switch Transformers have been proposed as an energy efficient path to even larger and more capable language models. But advancing the state-of-the-art across a broad set of natural language tas...
[ "Barret Zoph", "Irwan Bello", "Sameer Kumar", "Nan Du", "Yanping Huang", "Jeff Dean", "Noam Shazeer", "William Fedus" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2022-02-17T00:00:00
https://arxiv.org/abs/2202.08906
https://arxiv.org/pdf/2202.08906v2
2202.08906
null
411
26
false
null
null
0.7157
c6e4164642396f36eed0941b2001f06b4cc8a2fe9eb5144d99f27ceead60c77e
[ "arxiv", "semantic_scholar" ]
DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scale
As the training of giant dense models hits the boundary on the availability and capability of the hardware resources today, Mixture-of-Experts (MoE) models become one of the most promising model architectures due to their significant training cost reduction compared to a quality-equivalent dense model. Its training cos...
[ "Samyam Rajbhandari", "Conglong Li", "Zhewei Yao", "Minjia Zhang", "Reza Yazdani Aminabadi", "Ammar Ahmad Awan", "Jeff Rasley", "Yuxiong He" ]
[ "cs.LG", "cs.AI", "cs.DC" ]
[ "Computer Science" ]
2022-01-14T00:00:00
https://arxiv.org/abs/2201.05596
https://arxiv.org/pdf/2201.05596v2
2201.05596
null
563
58
false
null
International Conference on Machine Learning
0.8854
f278e5f6b2ca16a750dbc62da96f8f270528fa42bc8222d5446b5aec4445fe22
[ "arxiv", "semantic_scholar" ]
EvoMoE: An Evolutional Mixture-of-Experts Training Framework via Dense-To-Sparse Gate
Mixture-of-experts (MoE) is becoming popular due to its success in improving the model quality, especially in Transformers. By routing tokens with a sparse gate to a few experts (i.e., a small pieces of the full model), MoE can easily increase the model parameters to a very large scale while keeping the computation cos...
[ "Xiaonan Nie", "Xupeng Miao", "Shijie Cao", "Lingxiao Ma", "Qibin Liu", "Jilong Xue", "Youshan Miao", "Yi Liu", "Zhi Yang", "Bin Cui" ]
[ "cs.LG", "cs.AI" ]
[ "Computer Science" ]
2021-12-29T00:00:00
https://arxiv.org/abs/2112.14397
https://arxiv.org/pdf/2112.14397v2
2112.14397
null
39
4
false
null
null
0.4005
906ab457f419b7935aadb72a52c65f1b0b467b8a8175115f4d3d5755aad0fe82
[ "arxiv", "semantic_scholar" ]
Building a great multi-lingual teacher with sparsely-gated mixture of experts for speech recognition
The sparsely-gated Mixture of Experts (MoE) can magnify a network capacity with a little computational complexity. In this work, we investigate how multi-lingual Automatic Speech Recognition (ASR) networks can be scaled up with a simple routing algorithm in order to achieve better accuracy. More specifically, we apply ...
[ "Kenichi Kumatani", "Robert Gmyr", "Felipe Cruz Salinas", "Linquan Liu", "Wei Zuo", "Devang Patel", "Eric Sun", "Yu Shi" ]
[ "cs.CL", "cs.AI", "cs.LG", "eess.AS" ]
[ "Computer Science", "Engineering" ]
2021-12-10T00:00:00
https://arxiv.org/abs/2112.05820
https://arxiv.org/pdf/2112.05820v3
2112.05820
null
22
0
false
null
arXiv.org
0.3404
58ee66f6ea3c2745c9a01ef1f66ab21d1a2b331b950c593d56c27345a5f997ba
[ "arxiv", "semantic_scholar" ]
SpeechMoE2: Mixture-of-Experts Model with Improved Routing
Mixture-of-experts based acoustic models with dynamic routing mechanisms have proved promising results for speech recognition. The design principle of router architecture is important for the large model capacity and high computational efficiency. Our previous work SpeechMoE only uses local grapheme embedding to help r...
[ "Zhao You", "Shulin Feng", "Dan Su", "Dong Yu" ]
[ "eess.AS", "cs.CL", "cs.SD" ]
[ "Computer Science", "Engineering" ]
2021-11-23T00:00:00
https://arxiv.org/abs/2111.11831
https://arxiv.org/pdf/2111.11831v1
2111.11831
10.1109/icassp43922.2022.9747065
41
1
false
null
IEEE International Conference on Acoustics, Speech, and Signal Processing
0.4058
c0e84cd18c452c9716ff9052c5643cd465a9e0da8c8ec5d8917a4976cfcf2bec
[ "arxiv", "semantic_scholar" ]
Taming Sparsely Activated Transformer with Stochastic Experts
Sparsely activated models (SAMs), such as Mixture-of-Experts (MoE), can easily scale to have outrageously large amounts of parameters without significant increase in computational cost. However, SAMs are reported to be parameter inefficient such that larger models do not always lead to better performance. While most on...
[ "Simiao Zuo", "Xiaodong Liu", "Jian Jiao", "Young Jin Kim", "Hany Hassan", "Ruofei Zhang", "Tuo Zhao", "Jianfeng Gao" ]
[ "cs.CL", "cs.LG" ]
[ "Computer Science" ]
2021-10-08T00:00:00
https://arxiv.org/abs/2110.04260
https://arxiv.org/pdf/2110.04260v3
2110.04260
null
149
14
true
https://github.com/microsoft/Stochastic-Mixture-of-Experts
International Conference on Learning Representations
0.588
b889c7cae9a0b241006e9c9e7f2a330aa07cae85796e4f2e97f03e6ee4f9a98e
[ "arxiv", "semantic_scholar" ]
Sparse MoEs meet Efficient Ensembles
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 models: ensembles of neural networks and sparse mixture of experts (sparse MoEs). Fir...
[ "James Urquhart Allingham", "Florian Wenzel", "Zelda E Mariet", "Basil Mustafa", "Joan Puigcerver", "Neil Houlsby", "Ghassen Jerfel", "Vincent Fortuin", "Balaji Lakshminarayanan", "Jasper Snoek", "Dustin Tran", "Carlos Riquelme Ruiz", "Rodolphe Jenatton" ]
[ "cs.LG", "cs.CV", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2021-10-07T00:00:00
https://arxiv.org/abs/2110.03360
https://arxiv.org/pdf/2110.03360v2
2110.03360
null
24
0
false
null
null
0.3495
58fec9c4a70cac02837664f8342a863cd2f4c725421fd98b76aaabb533354c5f
[ "arxiv", "semantic_scholar" ]
Scaling Vision with Sparse Mixture of Experts
Sparsely-gated Mixture of Experts networks (MoEs) have demonstrated excellent scalability in Natural Language Processing. In Computer Vision, however, almost all performant networks are "dense", that is, every input is processed by every parameter. We present a Vision MoE (V-MoE), a sparse version of the Vision Transfo...
[ "Carlos Riquelme", "Joan Puigcerver", "Basil Mustafa", "Maxim Neumann", "Rodolphe Jenatton", "André Susano Pinto", "Daniel Keysers", "Neil Houlsby" ]
[ "cs.CV", "cs.LG", "stat.ML" ]
[ "Computer Science", "Mathematics" ]
2021-06-10T00:00:00
https://arxiv.org/abs/2106.05974
https://arxiv.org/pdf/2106.05974v1
2106.05974
null
1,061
66
false
null
Neural Information Processing Systems
0.913
f97ebcc08ebf096aebd868609a3bd1b079fbb9d12f89e0135581500ff90442e4
[ "arxiv", "semantic_scholar" ]
M6-T: Exploring Sparse Expert Models and Beyond
Mixture-of-Experts (MoE) models can achieve promising results with outrageous large amount of parameters but constant computation cost, and thus it has become a trend in model scaling. Still it is a mystery how MoE layers bring quality gains by leveraging the parameters with sparse activation. In this work, we investig...
[ "An Yang", "Junyang Lin", "Rui Men", "Chang Zhou", "Le Jiang", "Xianyan Jia", "Ang Wang", "Jie Zhang", "Jiamang Wang", "Yong Li", "Di Zhang", "Wei Lin", "Lin Qu", "Jingren Zhou", "Hongxia Yang" ]
[ "cs.LG", "cs.CL" ]
[ "Computer Science" ]
2021-05-31T00:00:00
https://arxiv.org/abs/2105.15082
https://arxiv.org/pdf/2105.15082v5
2105.15082
null
24
3
false
null
arXiv.org
0.3495