UltraEP is the first exact-load real-time expert balancer for large-EP MoE training and serving on rack-scale nodes, reaching 94.3% of ideal throughput and 1.49x over no-balancing.
LAER-MoE: Load-adaptive expert re-layout for efficient mixture- of-experts training
6 Pith papers cite this work. Polarity classification is still indexing.
6
Pith papers citing it
years
2026 6representative citing papers
UBEP replaces BSP All-to-All for MoE on multi-tier superpods with dependency-driven kernel decomposition, topology-aware token scheduling, and Data-as-Flag atomics, cutting All-to-All latency up to 52.4% and TPOT up to 11.1%.
Agents often fix reported LLVM missed-optimization cases yet misalign with developer-intended scope; historical PR retrieval and distillation improve alignment and real-world IR hits.
GF-DiT dynamically adapts parallelism during DiT serving via trajectory tasks and group-free collectives, reporting up to 6x throughput and 95% latency reduction versus static configurations.
NanoCP introduces request-level dynamic context parallelism to decouple MoE communication from KV cache placement in hybrid data-expert parallel serving, reporting up to 3.27x higher request rates and 2.12x lower P99 latency under TPOT SLOs.
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 persistently concentrated (MLA, GDN) classes.
citing papers explorer
-
UltraEP: Unleash MoE Training and Inference on Rack-Scale Nodes with Near-Optimal Load Balancing
UltraEP is the first exact-load real-time expert balancer for large-EP MoE training and serving on rack-scale nodes, reaching 94.3% of ideal throughput and 1.49x over no-balancing.
-
UBEP: Re-architecting Expert Parallelism Communication Library for Production Superpods
UBEP replaces BSP All-to-All for MoE on multi-tier superpods with dependency-driven kernel decomposition, topology-aware token scheduling, and Data-as-Flag atomics, cutting All-to-All latency up to 52.4% and TPOT up to 11.1%.
-
Understanding Agent-Based Patching of Compiler Missed Optimizations
Agents often fix reported LLVM missed-optimization cases yet misalign with developer-intended scope; historical PR retrieval and distillation improve alignment and real-world IR hits.
-
GF-DiT: Scheduling Parallelism for Diffusion Transformer Serving
GF-DiT dynamically adapts parallelism during DiT serving via trajectory tasks and group-free collectives, reporting up to 6x throughput and 95% latency reduction versus static configurations.
-
NanoCP: Request-Level Dynamic Context Parallelism for Data-Expert Parallel Decoding
NanoCP introduces request-level dynamic context parallelism to decouple MoE communication from KV cache placement in hybrid data-expert parallel serving, reporting up to 3.27x higher request rates and 2.12x lower P99 latency under TPOT SLOs.
-
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 persistently concentrated (MLA, GDN) classes.