QMoE: Practical Sub-1-Bit Compression of Trillion-Parameter Models
Paper
•
2310.16795
•
Published
•
26
MoE
-
-
Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference
Paper • 2308.12066 • Published • 4 -
Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference
Paper • 2303.06182 • Published • 1 -
EvoMoE: An Evolutional Mixture-of-Experts Training Framework via Dense-To-Sparse Gate
Paper • 2112.14397 • Published • 1 -
From Sparse to Soft Mixtures of Experts
Paper • 2308.00951 • Published • 20 -
Experts Weights Averaging: A New General Training Scheme for Vision Transformers
Paper • 2308.06093 • Published • 2 -
ShiftAddViT: Mixture of Multiplication Primitives Towards Efficient Vision Transformer
Paper • 2306.06446 • Published • 1 -
Sparse Upcycling: Training Mixture-of-Experts from Dense Checkpoints
Paper • 2212.05055 • Published • 5 -
SMILE: Scaling Mixture-of-Experts with Efficient Bi-level Routing
Paper • 2212.05191 • Published • 1 -
Patch-level Routing in Mixture-of-Experts is Provably Sample-efficient for Convolutional Neural Networks
Paper • 2306.04073 • Published • 2 -
Multi-Head Adapter Routing for Cross-Task Generalization
Paper • 2211.03831 • Published • 2 -
Improving Visual Prompt Tuning for Self-supervised Vision Transformers
Paper • 2306.05067 • Published • 2 -
A Unified View of Long-Sequence Models towards Modeling Million-Scale Dependencies
Paper • 2302.06218 • Published • 1 -
Alternating Gradient Descent and Mixture-of-Experts for Integrated Multimodal Perception
Paper • 2305.06324 • Published • 1 -
Sparse Backpropagation for MoE Training
Paper • 2310.00811 • Published • 2 -
Zorro: the masked multimodal transformer
Paper • 2301.09595 • Published • 2 -
DeepSpeed-MoE: Advancing Mixture-of-Experts Inference and Training to Power Next-Generation AI Scale
Paper • 2201.05596 • Published • 2 -
Approximating Two-Layer Feedforward Networks for Efficient Transformers
Paper • 2310.10837 • Published • 10 -
Sparsifiner: Learning Sparse Instance-Dependent Attention for Efficient Vision Transformers
Paper • 2303.13755 • Published • 1 -
Mixture of Tokens: Efficient LLMs through Cross-Example Aggregation
Paper • 2310.15961 • Published • 1 -
LoRA ensembles for large language model fine-tuning
Paper • 2310.00035 • Published • 2 -
Build a Robust QA System with Transformer-based Mixture of Experts
Paper • 2204.09598 • Published • 1 -
PaCE: Unified Multi-modal Dialogue Pre-training with Progressive and Compositional Experts
Paper • 2305.14839 • Published • 1 -
A Mixture-of-Expert Approach to RL-based Dialogue Management
Paper • 2206.00059 • Published • 1 -
Spatial Mixture-of-Experts
Paper • 2211.13491 • Published • 1 -
FastMoE: A Fast Mixture-of-Expert Training System
Paper • 2103.13262 • Published • 2 -
SE-MoE: A Scalable and Efficient Mixture-of-Experts Distributed Training and Inference System
Paper • 2205.10034 • Published • 1 -
Eliciting and Understanding Cross-Task Skills with Task-Level Mixture-of-Experts
Paper • 2205.12701 • Published • 1 -
FEAMOE: Fair, Explainable and Adaptive Mixture of Experts
Paper • 2210.04995 • Published • 1 -
On the Adversarial Robustness of Mixture of Experts
Paper • 2210.10253 • Published • 1 -
HMOE: Hypernetwork-based Mixture of Experts for Domain Generalization
Paper • 2211.08253 • Published • 1 -
Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity
Paper • 2101.03961 • Published • 14 -
A Hybrid Tensor-Expert-Data Parallelism Approach to Optimize Mixture-of-Experts Training
Paper • 2303.06318 • Published • 1 -
Mixture-of-Supernets: Improving Weight-Sharing Supernet Training with Architecture-Routed Mixture-of-Experts
Paper • 2306.04845 • Published • 4 -
AutoMoE: Heterogeneous Mixture-of-Experts with Adaptive Computation for Efficient Neural Machine Translation
Paper • 2210.07535 • Published • 1 -
Optimizing Mixture of Experts using Dynamic Recompilations
Paper • 2205.01848 • Published • 1 -
Towards Understanding Mixture of Experts in Deep Learning
Paper • 2208.02813 • Published • 1 -
Learning Factored Representations in a Deep Mixture of Experts
Paper • 1312.4314 • Published • 1 -
Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer
Paper • 1701.06538 • Published • 4 -
GLaM: Efficient Scaling of Language Models with Mixture-of-Experts
Paper • 2112.06905 • Published • 1 -
Contextual Mixture of Experts: Integrating Knowledge into Predictive Modeling
Paper • 2211.00558 • Published • 1 -
Taming Sparsely Activated Transformer with Stochastic Experts
Paper • 2110.04260 • Published • 2 -
Heterogeneous Multi-task Learning with Expert Diversity
Paper • 2106.10595 • Published • 1 -
SteloCoder: a Decoder-Only LLM for Multi-Language to Python Code Translation
Paper • 2310.15539 • Published • 1 -
LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Composition
Paper • 2307.13269 • Published • 31 -
SkillNet-NLG: General-Purpose Natural Language Generation with a Sparsely Activated Approach
Paper • 2204.12184 • Published • 1 -
SkillNet-NLU: A Sparsely Activated Model for General-Purpose Natural Language Understanding
Paper • 2203.03312 • Published • 1 -
Residual Mixture of Experts
Paper • 2204.09636 • Published • 1 -
Parameter-Efficient Mixture-of-Experts Architecture for Pre-trained Language Models
Paper • 2203.01104 • Published • 2 -
Emergent Mixture-of-Experts: Can Dense Pre-trained Transformers Benefit from Emergent Modular Structures?
Paper • 2310.10908 • Published • 1 -
One Student Knows All Experts Know: From Sparse to Dense
Paper • 2201.10890 • Published • 1 -
HetuMoE: An Efficient Trillion-scale Mixture-of-Expert Distributed Training System
Paper • 2203.14685 • Published • 1 -
Edge-MoE: Memory-Efficient Multi-Task Vision Transformer Architecture with Task-level Sparsity via Mixture-of-Experts
Paper • 2305.18691 • Published • 1 -
An Efficient General-Purpose Modular Vision Model via Multi-Task Heterogeneous Training
Paper • 2306.17165 • Published • 1 -
MegaBlocks: Efficient Sparse Training with Mixture-of-Experts
Paper • 2211.15841 • Published • 7 -
SpeechMoE: Scaling to Large Acoustic Models with Dynamic Routing Mixture of Experts
Paper • 2105.03036 • Published • 2 -
Language-Routing Mixture of Experts for Multilingual and Code-Switching Speech Recognition
Paper • 2307.05956 • Published • 1 -
M6-T: Exploring Sparse Expert Models and Beyond
Paper • 2105.15082 • Published • 1 -
Cross-token Modeling with Conditional Computation
Paper • 2109.02008 • Published • 1 -
Sparsely-gated Mixture-of-Expert Layers for CNN Interpretability
Paper • 2204.10598 • Published • 2 -
Efficient Language Modeling with Sparse all-MLP
Paper • 2203.06850 • Published • 1 -
Efficient Large Scale Language Modeling with Mixtures of Experts
Paper • 2112.10684 • Published • 2 -
TAME: Task Agnostic Continual Learning using Multiple Experts
Paper • 2210.03869 • Published • 1 -
Learning an evolved mixture model for task-free continual learning
Paper • 2207.05080 • Published • 1 -
Model Spider: Learning to Rank Pre-Trained Models Efficiently
Paper • 2306.03900 • Published • 1 -
Task-Specific Expert Pruning for Sparse Mixture-of-Experts
Paper • 2206.00277 • Published • 1 -
SiRA: Sparse Mixture of Low Rank Adaptation
Paper • 2311.09179 • Published • 8 -
Pushing Mixture of Experts to the Limit: Extremely Parameter Efficient MoE for Instruction Tuning
Paper • 2309.05444 • Published • 1 -
MoEC: Mixture of Expert Clusters
Paper • 2207.09094 • Published • 1 -
Image Super-resolution Via Latent Diffusion: A Sampling-space Mixture Of Experts And Frequency-augmented Decoder Approach
Paper • 2310.12004 • Published • 2 -
A General Theory for Softmax Gating Multinomial Logistic Mixture of Experts
Paper • 2310.14188 • Published • 1 -
Extending Mixture of Experts Model to Investigate Heterogeneity of Trajectories: When, Where and How to Add Which Covariates
Paper • 2007.02432 • Published • 1 -
Mixture of experts models for multilevel data: modelling framework and approximation theory
Paper • 2209.15207 • Published • 1 -
ComPEFT: Compression for Communicating Parameter Efficient Updates via Sparsification and Quantization
Paper • 2311.13171 • Published • 1 -
The Information Pathways Hypothesis: Transformers are Dynamic Self-Ensembles
Paper • 2306.01705 • Published • 1 -
Exponentially Faster Language Modelling
Paper • 2311.10770 • Published • 118 -
Scaling Expert Language Models with Unsupervised Domain Discovery
Paper • 2303.14177 • Published • 2 -
Hash Layers For Large Sparse Models
Paper • 2106.04426 • Published • 2 -
Memory-efficient NLLB-200: Language-specific Expert Pruning of a Massively Multilingual Machine Translation Model
Paper • 2212.09811 • Published • 1 -
Exploiting Transformer Activation Sparsity with Dynamic Inference
Paper • 2310.04361 • Published • 1 -
Mixture of Quantized Experts (MoQE): Complementary Effect of Low-bit Quantization and Robustness
Paper • 2310.02410 • Published • 1 -
Punica: Multi-Tenant LoRA Serving
Paper • 2310.18547 • Published • 2 -
Merging Experts into One: Improving Computational Efficiency of Mixture of Experts
Paper • 2310.09832 • Published • 1 -
Adaptive Gating in Mixture-of-Experts based Language Models
Paper • 2310.07188 • Published • 2 -
Making Small Language Models Better Multi-task Learners with Mixture-of-Task-Adapters
Paper • 2309.11042 • Published • 2 -
SwitchHead: Accelerating Transformers with Mixture-of-Experts Attention
Paper • 2312.07987 • Published • 40 -
Omni-SMoLA: Boosting Generalist Multimodal Models with Soft Mixture of Low-rank Experts
Paper • 2312.00968 • Published • 1 -
Memory Augmented Language Models through Mixture of Word Experts
Paper • 2311.10768 • Published • 16 -
Routing to the Expert: Efficient Reward-guided Ensemble of Large Language Models
Paper • 2311.08692 • Published • 12 -
DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models
Paper • 2401.06066 • Published • 42 -
Mixture of Attention Heads: Selecting Attention Heads Per Token
Paper • 2210.05144 • Published • 2 -
Direct Neural Machine Translation with Task-level Mixture of Experts models
Paper • 2310.12236 • Published • 2 -
Mixture-of-Linguistic-Experts Adapters for Improving and Interpreting Pre-trained Language Models
Paper • 2310.16240 • Published • 1 -
Parameter-Efficient Sparsity Crafting from Dense to Mixture-of-Experts for Instruction Tuning on General Tasks
Paper • 2401.02731 • Published • 2 -
Parameter-Efficient Conformers via Sharing Sparsely-Gated Experts for End-to-End Speech Recognition
Paper • 2209.08326 • Published • 1 -
Mixture-of-experts VAEs can disregard variation in surjective multimodal data
Paper • 2204.05229 • Published • 1 -
One Model, Multiple Modalities: A Sparsely Activated Approach for Text, Sound, Image, Video and Code
Paper • 2205.06126 • Published • 1 -
Specialized Language Models with Cheap Inference from Limited Domain Data
Paper • 2402.01093 • Published • 45 -
BlackMamba: Mixture of Experts for State-Space Models
Paper • 2402.01771 • Published • 22 -
OpenMoE: An Early Effort on Open Mixture-of-Experts Language Models
Paper • 2402.01739 • Published • 26 -
Fast Inference of Mixture-of-Experts Language Models with Offloading
Paper • 2312.17238 • Published • 7 -
Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference
Paper • 2401.08383 • Published • 1 -
A Review of Sparse Expert Models in Deep Learning
Paper • 2209.01667 • Published • 3 -
Robust Mixture-of-Expert Training for Convolutional Neural Networks
Paper • 2308.10110 • Published • 2 -
On the Representation Collapse of Sparse Mixture of Experts
Paper • 2204.09179 • Published • 1 -
StableMoE: Stable Routing Strategy for Mixture of Experts
Paper • 2204.08396 • Published • 1 -
DSelect-k: Differentiable Selection in the Mixture of Experts with Applications to Multi-Task Learning
Paper • 2106.03760 • Published • 3 -
CPM-2: Large-scale Cost-effective Pre-trained Language Models
Paper • 2106.10715 • Published • 1 -
Demystifying Softmax Gating Function in Gaussian Mixture of Experts
Paper • 2305.03288 • Published • 1 -
Statistical Perspective of Top-K Sparse Softmax Gating Mixture of Experts
Paper • 2309.13850 • Published • 1 -
Sparse Mixture-of-Experts are Domain Generalizable Learners
Paper • 2206.04046 • Published • 1 -
Unified Scaling Laws for Routed Language Models
Paper • 2202.01169 • Published • 2 -
Multimodal Contrastive Learning with LIMoE: the Language-Image Mixture of Experts
Paper • 2206.02770 • Published • 3 -
Merge, Then Compress: Demystify Efficient SMoE with Hints from Its Routing Policy
Paper • 2310.01334 • Published • 3 -
ST-MoE: Designing Stable and Transferable Sparse Expert Models
Paper • 2202.08906 • Published • 2 -
Mobile V-MoEs: Scaling Down Vision Transformers via Sparse Mixture-of-Experts
Paper • 2309.04354 • Published • 13 -
A non-asymptotic approach for model selection via penalization in high-dimensional mixture of experts models
Paper • 2104.02640 • Published • 1 -
Non-asymptotic oracle inequalities for the Lasso in high-dimensional mixture of experts
Paper • 2009.10622 • Published • 1 -
Fast Feedforward Networks
Paper • 2308.14711 • Published • 2 -
Mixture-of-Experts with Expert Choice Routing
Paper • 2202.09368 • Published • 3 -
Go Wider Instead of Deeper
Paper • 2107.11817 • Published • 1 -
Sparse Mixers: Combining MoE and Mixing to build a more efficient BERT
Paper • 2205.12399 • Published • 1 -
AdaMix: Mixture-of-Adaptations for Parameter-efficient Model Tuning
Paper • 2205.12410 • Published • 1 -
Mixtures of Experts Unlock Parameter Scaling for Deep RL
Paper • 2402.08609 • Published • 34 -
Fiddler: CPU-GPU Orchestration for Fast Inference of Mixture-of-Experts Models
Paper • 2402.07033 • Published • 16 -
MoAI: Mixture of All Intelligence for Large Language and Vision Models
Paper • 2403.07508 • Published • 75 -
Scattered Mixture-of-Experts Implementation
Paper • 2403.08245 • Published • 1 -
Sparse Universal Transformer
Paper • 2310.07096 • Published -
Multi-Head Mixture-of-Experts
Paper • 2404.15045 • Published • 59 -
LoRAMoE: Revolutionizing Mixture of Experts for Maintaining World Knowledge in Language Model Alignment
Paper • 2312.09979 • Published • 1 -
JetMoE: Reaching Llama2 Performance with 0.1M Dollars
Paper • 2404.07413 • Published • 36 -
Learning to Route Among Specialized Experts for Zero-Shot Generalization
Paper • 2402.05859 • Published • 5 -
MoELoRA: Contrastive Learning Guided Mixture of Experts on Parameter-Efficient Fine-Tuning for Large Language Models
Paper • 2402.12851 • Published • 2 -
MoEUT: Mixture-of-Experts Universal Transformers
Paper • 2405.16039 • Published -
Yuan 2.0-M32: Mixture of Experts with Attention Router
Paper • 2405.17976 • Published • 18 -
Enhancing Fast Feed Forward Networks with Load Balancing and a Master Leaf Node
Paper • 2405.16836 • Published -
Mixture of A Million Experts
Paper • 2407.04153 • Published • 4