UniEP fuses MoE communication and computation into unified MegaKernels with deterministic token ordering, delivering 1.03x-1.38x speedups over prior work while preserving training accuracy.
Deepspeed-moe: Advancing mixture-of-experts inference and training to power next-generation AI scale
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
UniEP: Unified Expert-Parallel MoE MegaKernel for LLM Training
UniEP fuses MoE communication and computation into unified MegaKernels with deterministic token ordering, delivering 1.03x-1.38x speedups over prior work while preserving training accuracy.