REVIEW 13 cited by
Tutel: Adaptive Mixture-of-Experts at Scale
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
Signed reviews
read the original abstract
Sparsely-gated mixture-of-experts (MoE) has been widely adopted to scale deep learning models to trillion-plus parameters with fixed computational cost. The algorithmic performance of MoE relies on its token routing mechanism that forwards each input token to the right sub-models or experts. While token routing dynamically determines the amount of expert workload at runtime, existing systems suffer inefficient computation due to their static execution, namely static parallelism and pipelining, which does not adapt to the dynamic workload. We present Flex, a highly scalable stack design and implementation for MoE with dynamically adaptive parallelism and pipelining. Flex designs an identical layout for distributing MoE model parameters and input data, which can be leveraged by all possible parallelism or pipelining methods without any mathematical inequivalence or tensor migration overhead. This enables adaptive parallelism/pipelining optimization at zero cost during runtime. Based on this key design, Flex also implements various MoE acceleration techniques. Aggregating all techniques, Flex finally delivers huge speedup at any scale -- 4.96x and 5.75x speedup of a single MoE layer over 16 and 2,048 A100 GPUs, respectively, over the previous state-of-the-art. Our evaluation shows that Flex efficiently and effectively runs a real-world MoE-based model named SwinV2-MoE, built upon Swin Transformer V2, a state-of-the-art computer vision architecture. On efficiency, Flex accelerates SwinV2-MoE, achieving up to 1.55x and 2.11x speedup in training and inference over Fairseq, respectively. On effectiveness, the SwinV2-MoE model achieves superior accuracy in both pre-training and down-stream computer vision tasks such as COCO object detection than the counterpart dense model, indicating the readiness of Flex for end-to-end real-world model training and inference.
Forward citations
Cited by 13 Pith papers
-
Hecate: Unlocking Efficient Sparse Model Training via Fully Sharded Sparse Data Parallelism
Fully Sharded Sparse Data Parallelism materializes expert parameters on the fly with SparseAllGather and SparseReduceScatter, avoiding the overhead of expert rearrangement in MoE training.
-
Adaptive Matrix Multiplication for Dynamic Shapes on Ascend NPUs
An adaptive MatMul framework for Ascend NPUs selects optimal tiling and instruction schedules via an analytical model, delivering 1.85x mean speedup over the vendor library on 80,000 dynamic shapes.
-
Relax Within, Balance Across: Geometry-Guided Load Balancing for Vision-Language Mixture-of-Experts
The standard mixed image-text load-balancing loss lets modality errors cancel at one token mix; ReBA balances each modality separately and per image, reducing load across resolutions and tiles with comparable accuracy.
-
A Training-Memory Regression in MLA Sequence Parallelism: Why Megatron-Core Forbids Absorption, and LAGA -- a Communication-Efficient Fix
LAGA replaces MLA training's per-head K/V all-to-all with a latent all-gather and local up-projection, matching explicit-form memory while cutting collective communication ~1.98x.
-
Communication-Aware Placement and Pruning for Efficient Mixture-of-Experts Inference
Communication-aware expert placement plus device-level pruning yields 1.23–1.86× MoE inference throughput and better accuracy at equal speedup than load-balance or sequential baselines.
-
Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory
Expert-parallel scaling leaves per-expert routing imbalance flat; mock-token benchmarks overestimate real-text imbalance and fake a batch-size trend; architectures split into data-resilient (MHA, Mamba-2) and persiste...
-
Physics of Language Models: Part 4.1, Architecture Design and the Magic of Canon Layers
Canon layers—residual 1-d causal convolutions over adjacent tokens—boost synthetic reasoning depth 2-4x, lift NoPE to RoPE level, and bring GLA up to Mamba2/GDN, with qualitative real-world confirmation.
-
MPipeMoE: Memory Efficient MoE for Pre-trained Models with Adaptive Pipeline Parallelism
MPipeMoE speeds up MoE training by adaptively pipelining token batches and reusing memory buffers across partitions, achieving up to 2.8x speedup and 47% memory reduction over FasterMoE.
-
Learning Heterogeneous Mixture of Scene Experts for Large-scale Neural Radiance Fields
A mixture-of-experts NeRF with hash-based gating and heterogeneous grid-resolution experts improves large-scale scene rendering accuracy and training efficiency over prior decomposition-based NeRFs.
-
Maximum Score Routing For Mixture-of-Experts
MaxScore casts MoE routing as min-cost max-flow with SoftTopk and claims better loss and eval scores at equal FLOPs; unverified because the full text is unreadable.
-
The Evolution of Mixture-of-Experts Architectures in Large Language Models: Routing, Topology, Load Balancing, and Expert Parallelism
A survey proposes eight architectural milestones and four control planes for MoE LLMs, arguing the field is moving toward decoupling routing, compute budgets, and physical execution.
-
Expert Merging in Sparse Mixture of Experts with Nash Bargaining
Expert merging in MoE models can be improved by setting per-expert weights with the Nash bargaining solution instead of simple averaging.
-
Taming the Titans: A Survey of Efficient LLM Inference Serving
A survey that classifies LLM inference serving techniques into instance, cluster, emerging-scenario, and miscellaneous categories, with a companion GitHub index and four research directions.
Discussion (0). Continue with ORCID to comment.