Sieve dynamically schedules MoE experts across GPU and PIM hardware to handle bimodal token distributions, achieving 1.3x to 1.6x gains in throughput and interactivity over static prior PIM systems on three large models.
hub Canonical reference
DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models
Canonical reference. 88% of citing Pith papers cite this work as background.
87
Pith papers citing it
Background
88% of classified citations
abstract
In the era of large language models, Mixture-of-Experts (MoE) is a promising architecture for managing computational costs when scaling up model parameters. However, conventional MoE architectures like GShard, which activate the top-$K$ out of $N$ experts, face challenges in ensuring expert specialization, i.e. each expert acquires non-overlapping and focused knowledge. In response, we propose the DeepSeekMoE architecture towards ultimate expert specialization. It involves two principal strategies: (1) finely segmenting the experts into $mN$ ones and activating $mK$ from them, allowing for a more flexible combination of activated experts; (2) isolating $K_s$ experts as shared ones, aiming at capturing common knowledge and mitigating redundancy in routed experts. Starting from a modest scale with 2B parameters, we demonstrate that DeepSeekMoE 2B achieves comparable performance with GShard 2.9B, which has 1.5 times the expert parameters and computation. In addition, DeepSeekMoE 2B nearly approaches the performance of its dense counterpart with the same number of total parameters, which set the upper bound of MoE models. Subsequently, we scale up DeepSeekMoE to 16B parameters and show that it achieves comparable performance with LLaMA2 7B, with only about 40% of computations. Further, our preliminary efforts to scale up DeepSeekMoE to 145B parameters consistently validate its substantial advantages over the GShard architecture, and show its performance comparable with DeepSeek 67B, using only 28.5% (maybe even 18.2%) of computations.
hub tools
citation-role summary
background 21
baseline 1
dataset 1
method 1
citation-polarity summary
claims ledger
- abstract In the era of large language models, Mixture-of-Experts (MoE) is a promising architecture for managing computational costs when scaling up model parameters. However, conventional MoE architectures like GShard, which activate the top-$K$ out of $N$ experts, face challenges in ensuring expert specialization, i.e. each expert acquires non-overlapping and focused knowledge. In response, we propose the DeepSeekMoE architecture towards ultimate expert specialization. It involves two principal strategies: (1) finely segmenting the experts into $mN$ ones and activating $mK$ from them, allowing for a m
co-cited works
representative citing papers
The authors derive a Maximally Scale-Stable Parameterization (MSSP) for MoE models that achieves robust learning-rate transfer and monotonic performance gains with scale across co-scaling regimes of width, experts, and sparsity.
Routers in SMoE models form geometric alignments with their experts through shared gradient directions, enabling effective specialization that auxiliary load-balancing losses tend to disrupt.
EEP makes wide expert-parallel MoE serving survive single-rank failures with an 11s recovery pause, 8s reintegration pause, and throughput restored to 95% of pre-fault level within 52s while staying within 4.4% of a fixed-membership baseline in steady state.
Mixture of Layers replaces monolithic transformer blocks with routed thin parallel blocks using hybrid attention that combines a shared softmax block for global context with Gated DeltaNet linear attention in the routed blocks.
SDG-MoE introduces learned signed interaction graphs and disagreement-gated deliberation among experts in MoE architectures, yielding 19.8% better validation perplexity than the strongest baseline.
MISA routes to a small subset of indexer heads via block statistics, matching full DSA performance on LongBench with 4-8x fewer heads and 3.82x speedup while recovering over 92% of selected tokens.
MoE experts in pretrained Transformers exhibit functional decorrelation with near-zero Jacobian alignment yet occupy partially overlapping representation subspaces, with routing sparsity modulating the geometry.
OneTrackerV2 unifies multimodal tracking via Meta Merger and Dual Mixture-of-Experts to reach state-of-the-art results on five tasks and 12 benchmarks with efficiency and robustness when modalities are missing.
RouteHijack is a routing-aware jailbreak that identifies safety-critical experts via activation contrast and optimizes suffixes to suppress them, reaching 69.3% average attack success rate on seven MoE LLMs with strong transfer to variants and VLMs.
MASCing uses an LSTM surrogate and optimized steering masks to enable flexible, inference-time control over MoE expert routing for safety objectives, improving jailbreak defense and content generation success rates substantially across multiple models.
PathMoE constrains expert paths in MoE models by sharing router parameters across layer blocks, yielding more concentrated paths, better performance on perplexity and tasks, and no need for auxiliary losses.
Split-MoPE integrates split learning with predefined-expert routing to maximize usable data in vertical federated learning under sample misalignment, delivering state-of-the-art accuracy in one communication round plus built-in robustness and per-sample contribution scores.
A new sequential interaction framework lets LLMs propose questions to forums, with simulations on real Stack Exchange data showing players can reach roughly half the utility of an ideal full-information scenario despite incentive misalignment.
S²GR adds stepwise thinking tokens with contrastive supervision on codebook clusters to balance computational focus and ground reasoning paths in generative recommendation.
DuoServe-MoE decouples prefill and decode phases in MoE LLM inference with a two-stream CUDA pipeline for prefill and an offline-trained predictor for decode, reporting up to 5.34x TTFT and 7.55x end-to-end latency gains.
Loss-Free Balancing keeps expert loads balanced in MoE models by dynamically adjusting routing-score biases based on recent usage, avoiding auxiliary-loss interference and yielding better performance.
DeepSeek-V2 delivers top-tier open-source LLM performance using only 21B active parameters by compressing the KV cache 93.3% and cutting training costs 42.5% via MLA and DeepSeekMoE.
A framework extracts a latent state machine from logs, induces a multi-table relational schema, and uses it as a generative prior to create synthetic data that augments real logs for better anomaly detection.
NASiC fuses CAM-based expert selection and multibit CIM computation in 3D NAND into one cycle for MoE LLM inference, claiming 4-114.8x performance and 3.9-70x energy efficiency gains over prior designs with high accuracy.
GEMQ applies global LP-based expert importance estimation and router fine-tuning within progressive quantization to cut memory and speed inference in MoE LLMs with little accuracy loss.
PALS adds dynamic GPU power capping to LLM serving frameworks like vLLM, jointly tuning it with batch size via offline models and feedback control to improve energy efficiency up to 26.3% and cut QoS violations 4-7x on dense and MoE models.
DODOCO measurements show MoE routing imbalance is intrinsic to architecture and real text, not correctable by EP scaling or represented by mock tokens, forming two persistent Gini bands.
HDMoE uses hierarchical MoE and RFR modules to address redundant information and fine-grained intra/inter-modality relationships in multimodal cancer survival prediction, with positive results on private liver cancer and TCGA datasets.
citing papers explorer
-
Sieve: Dynamic Expert-Aware PIM Acceleration for Evolving Mixture-of-Experts Models
Sieve dynamically schedules MoE experts across GPU and PIM hardware to handle bimodal token distributions, achieving 1.3x to 1.6x gains in throughput and interactivity over static prior PIM systems on three large models.
-
How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization
The authors derive a Maximally Scale-Stable Parameterization (MSSP) for MoE models that achieves robust learning-rate transfer and monotonic performance gains with scale across co-scaling regimes of width, experts, and sparsity.
-
Routers Learn the Geometry of Their Experts: Geometric Coupling in Sparse Mixture-of-Experts
Routers in SMoE models form geometric alignments with their experts through shared gradient directions, enabling effective specialization that auxiliary load-balancing losses tend to disrupt.
-
Surviving Partial Rank Failures in Wide Expert-Parallel MoE Inference
EEP makes wide expert-parallel MoE serving survive single-rank failures with an 11s recovery pause, 8s reintegration pause, and throughput restored to 95% of pre-fault level within 52s while staying within 4.4% of a fixed-membership baseline in steady state.
-
Mixture of Layers with Hybrid Attention
Mixture of Layers replaces monolithic transformer blocks with routed thin parallel blocks using hybrid attention that combines a shared softmax block for global context with Gated DeltaNet linear attention in the routed blocks.
-
SDG-MoE: Signed Debate Graph Mixture-of-Experts
SDG-MoE introduces learned signed interaction graphs and disagreement-gated deliberation among experts in MoE architectures, yielding 19.8% better validation perplexity than the strongest baseline.
-
MISA: Mixture of Indexer Sparse Attention for Long-Context LLM Inference
MISA routes to a small subset of indexer heads via block statistics, matching full DSA performance on LongBench with 4-8x fewer heads and 3.82x speedup while recovering over 92% of selected tokens.
-
Geometric Asymmetry in MoE Specialization: Functional Decorrelation and Representational Overlap
MoE experts in pretrained Transformers exhibit functional decorrelation with near-zero Jacobian alignment yet occupy partially overlapping representation subspaces, with routing sparsity modulating the geometry.
-
Unified Multimodal Visual Tracking with Dual Mixture-of-Experts
OneTrackerV2 unifies multimodal tracking via Meta Merger and Dual Mixture-of-Experts to reach state-of-the-art results on five tasks and 12 benchmarks with efficiency and robustness when modalities are missing.
-
RouteHijack: Routing-Aware Attack on Mixture-of-Experts LLMs
RouteHijack is a routing-aware jailbreak that identifies safety-critical experts via activation contrast and optimizes suffixes to suppress them, reaching 69.3% average attack success rate on seven MoE LLMs with strong transfer to variants and VLMs.
-
MASCing: Configurable Mixture-of-Experts Behavior via Activation Steering Masks
MASCing uses an LSTM surrogate and optimized steering masks to enable flexible, inference-time control over MoE expert routing for safety objectives, improving jailbreak defense and content generation success rates substantially across multiple models.
-
Path-Constrained Mixture-of-Experts
PathMoE constrains expert paths in MoE models by sharing router parameters across layer blocks, yielding more concentrated paths, better performance on perplexity and tasks, and no need for auxiliary losses.
-
Mixture of Predefined Experts: Maximizing Data Usage on Vertical Federated Learning
Split-MoPE integrates split learning with predefined-expert routing to maximize usable data in vertical federated learning under sample misalignment, delivering state-of-the-art accuracy in one communication round plus built-in robustness and per-sample contribution scores.
-
From Competition to Collaboration: Designing Sustainable Mechanisms Between LLMs and Online Forums
A new sequential interaction framework lets LLMs propose questions to forums, with simulations on real Stack Exchange data showing players can reach roughly half the utility of an ideal full-information scenario despite incentive misalignment.
-
S$^2$GR: Stepwise Semantic-Guided Reasoning in Latent Space for Generative Recommendation
S²GR adds stepwise thinking tokens with contrastive supervision on codebook clusters to balance computational focus and ground reasoning paths in generative recommendation.
-
DuoServe-MoE: Dual-Phase Expert Prefetch and Caching for LLM Inference QoS Assurance
DuoServe-MoE decouples prefill and decode phases in MoE LLM inference with a two-stream CUDA pipeline for prefill and an offline-trained predictor for decode, reporting up to 5.34x TTFT and 7.55x end-to-end latency gains.
-
Auxiliary-Loss-Free Load Balancing Strategy for Mixture-of-Experts
Loss-Free Balancing keeps expert loads balanced in MoE models by dynamically adjusting routing-score biases based on recent usage, avoiding auxiliary-loss interference and yielding better performance.
-
DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
DeepSeek-V2 delivers top-tier open-source LLM performance using only 21B active parameters by compressing the KV cache 93.3% and cutting training costs 42.5% via MLA and DeepSeekMoE.
-
State Machine Guided Multi-Relational Synthetic Data from Logs for Anomaly Detection
A framework extracts a latent state machine from logs, induces a multi-table relational schema, and uses it as a generative prior to create synthetic data that augments real logs for better anomaly detection.
-
NASiC: 3D NAND-based CAM-Selected Multibit CIM Architecture for Efficient On-Device Mixture-of-Experts LLM Inference
NASiC fuses CAM-based expert selection and multibit CIM computation in 3D NAND into one cycle for MoE LLM inference, claiming 4-114.8x performance and 3.9-70x energy efficiency gains over prior designs with high accuracy.
-
GEMQ: Global Expert-Level Mixed-Precision Quantization for MoE LLMs
GEMQ applies global LP-based expert importance estimation and router fine-tuning within progressive quantization to cut memory and speed inference in MoE LLMs with little accuracy loss.
-
PALS: Power-Aware LLM Serving for Mixture-of-Experts Models
PALS adds dynamic GPU power capping to LLM serving frameworks like vLLM, jointly tuning it with batch size via offline models and feedback control to improve energy efficiency up to 26.3% and cut QoS violations 4-7x on dense and MoE models.
-
Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory
DODOCO measurements show MoE routing imbalance is intrinsic to architecture and real text, not correctable by EP scaling or represented by mock tokens, forming two persistent Gini bands.
-
HDMoE: A Hierarchical Decoupling-Fusion Mixture-of-Experts Framework for Multimodal Cancer Survival Prediction
HDMoE uses hierarchical MoE and RFR modules to address redundant information and fine-grained intra/inter-modality relationships in multimodal cancer survival prediction, with positive results on private liver cancer and TCGA datasets.
-
What Really Improves Mathematical Reasoning: Structured Reasoning Signals Beyond Pure Code
Controlled experiments show structured reasoning traces and higher-density math-domain samples improve mathematical reasoning more than pure executable code, with internal routing patterns reflecting these data effects.
-
TFGN: Task-Free, Replay-Free Continual Pre-Training Without Catastrophic Forgetting at LLM Scale
TFGN is an architectural overlay for transformers enabling task-free, replay-free continual pre-training across heterogeneous domains at LLM scale with near-zero backward transfer and high gradient orthogonality.
-
Combining pre-trained models via localized model averaging
Localized model averaging with covariate-dependent weights achieves asymptotic optimality and weight consistency for combining pre-trained models under a general loss framework.
-
MLCommons Chakra: Advancing Performance Benchmarking and Co-design using Standardized Execution Traces
Chakra introduces a standardized graph-based execution trace representation for distributed ML workloads along with supporting tools to enable benchmarking, analysis, generation, and co-design across simulators and hardware.
-
DisagMoE: Computation-Communication overlapped MoE Training via Disaggregated AF-Pipe Parallelism
DisagMoE achieves up to 1.8x faster MoE training by disaggregating attention and FFN layers into disjoint GPU groups with a multi-stage uni-directional pipeline and roofline-based bandwidth balancing.
-
Hierarchical Mixture-of-Experts with Two-Stage Optimization
Hi-MoE uses two-level hierarchical routing objectives to enforce group-level balance while promoting within-group specialization, yielding better perplexity and expert utilization than prior MoE baselines in NLP and vision tasks.
-
TAS-LoRA: Transformer Architecture Search with Mixture-of-LoRA Experts
TAS-LoRA attaches a mixture of LoRA experts to a supernet and uses a dynamic router plus group-wise initialization to let different architecture subnets learn distinct features, yielding higher accuracy than prior TAS methods on ImageNet and transfer datasets.
-
UniPool: A Globally Shared Expert Pool for Mixture-of-Experts
A shared global expert pool in MoE improves validation loss over per-layer experts and allows sublinear expert-parameter growth with depth.
-
MoE-Hub: Taming Software Complexity for Seamless MoE Overlap with Hardware-Accelerated Communication on Multi-GPU Systems
MoE-Hub enables seamless MoE communication overlap via hardware-accelerated destination-agnostic data transmission, delivering 1.40x-3.08x per-layer and 1.21x-1.98x end-to-end speedups over prior systems.
-
Piper: Efficient Large-Scale MoE Training via Resource Modeling and Pipelined Hybrid Parallelism
Piper introduces resource modeling and pipelined hybrid parallelism for MoE training, delivering 2-3.5X higher MFU than prior frameworks and 1.2-9X better all-to-all bandwidth.
-
SMoES: Soft Modality-Guided Expert Specialization in MoE-VLMs
SMoES improves MoE-VLM performance and efficiency via soft modality-guided expert routing and inter-bin mutual information regularization, yielding 0.9-4.2% task gains and 56% communication reduction.
-
Beyond N-gram: Data-Aware X-GRAM Extraction for Efficient Embedding Parameter Scaling
X-GRAM applies data-aware dynamic token injection with hybrid hashing and local feature extraction to achieve up to 4.4 accuracy point gains over vanilla backbones and 3.2 over retrieval baselines at 0.73B-1.15B scales using 50% smaller tables.
-
Train Separately, Merge Together: Modular Post-Training with Mixture-of-Experts
BAR trains independent domain experts via separate mid-training, SFT, and RL pipelines then composes them with a MoE router to match monolithic retraining performance at lower cost and without catastrophic forgetting.
-
Geometric Routing Enables Causal Expert Control in Mixture of Experts
Cosine-similarity routing in low-dimensional space makes MoE experts monosemantic by construction and enables direct causal control via centroid interventions.
-
The Myth of Expert Specialization in MoEs: Why Routing Reflects Geometry, Not Necessarily Domain Expertise
Expert specialization in MoEs is an emergent effect of hidden state geometry due to linear routers, not domain expertise, as confirmed empirically across models and explained by a proof on load-balancing effects.
-
MP-ISMoE: Mixed-Precision Interactive Side Mixture-of-Experts for Efficient Transfer Learning
MP-ISMoE uses Gaussian noise perturbed iterative quantization and interactive side mixture-of-experts to deliver higher accuracy than prior memory-efficient transfer learning methods while keeping similar parameter and memory usage.
-
Adaptive Semantic Communication for Wireless Image Transmission Leveraging Mixture-of-Experts Mechanism
A novel adaptive MoE-based semantic communication system jointly routes experts using real-time CSI and semantic image content for improved MIMO wireless image transmission.
-
Mixture of Sequence: Theme-Aware Mixture-of-Experts for Long-Sequence Recommendation
MoS applies theme-aware routing to extract multi-scale theme-specific subsequences from noisy long user sequences, achieving state-of-the-art recommendation performance with fewer FLOPs than comparable MoE models.
-
L2R: Low-Rank and Lipschitz-Controlled Routing for Mixture-of-Experts
L2R improves MoE performance by routing in a low-rank space with Lipschitz-controlled saturated inner-product scoring and multi-anchor mechanisms.
-
Continually Evolving Skill Knowledge in Vision Language Action Model
Stellar VLA achieves continual learning in VLA models by maintaining a growing knowledge space and routing tasks to specialized experts conditioned on semantic relations, delivering strong LIBERO benchmark results with only 1% data replay and successful real-world transfer on dual-arm hardware.
-
Patterns behind Chaos: Forecasting Data Movement for Efficient Large-Scale MoE LLM Inference
Comprehensive profiling of expert selection in frontier MoE models reveals temporal and spatial patterns that enable 6.6x speedup on wafer-scale GPUs and 1.25x on existing systems via targeted optimizations.
-
ShinkaEvolve: Towards Open-Ended And Sample-Efficient Program Evolution
ShinkaEvolve improves sample efficiency in LLM-driven program evolution via parent sampling, code novelty rejection-sampling, and bandit LLM ensemble selection, achieving new SOTA circle packing with 150 samples and gains on math reasoning and competitive programming tasks.
-
Mixture-of-Experts Can Surpass Dense LLMs Under Strictly Equal Resource
MoE models with activation rates in an optimal region outperform dense LLMs of identical total parameter count, training compute, and data budget, with the optimal region consistent across scales.
-
DriveMoE: Mixture-of-Experts for Vision-Language-Action Model in End-to-End Autonomous Driving
DriveMoE applies scene-specialized Vision MoE and skill-specialized Action MoE to a VLA baseline to achieve SOTA closed-loop performance on Bench2Drive.
-
SoftHGNN: Soft Hypergraph Neural Networks for General Visual Recognition
SoftHGNN introduces differentiable soft hyperedges via learnable prototypes and top-k sparse selection to model high-order visual interactions and improve recognition accuracy.
-
TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate
TurboQuant achieves near-optimal vector quantization distortion for both MSE and inner products via random rotation and per-coordinate scalar quantization, with a formal proof that it matches lower bounds within a factor of approximately 2.7.