Alloc-MoE allocates a fixed expert activation budget using layer-level dynamic programming based on sensitivity and token-level score-based redistribution, delivering 1.15x prefill and 1.34x decode speedups on DeepSeek-V2-Lite at half the original budget while preserving performance.
Krishna Teja Chitty-Venkata, Sandeep Madireddy, Mu- rali Emani, and Venkatram Vishwanath
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
representative citing papers
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.
citing papers explorer
-
Alloc-MoE: Budget-Aware Expert Activation Allocation for Efficient Mixture-of-Experts Inference
Alloc-MoE allocates a fixed expert activation budget using layer-level dynamic programming based on sensitivity and token-level score-based redistribution, delivering 1.15x prefill and 1.34x decode speedups on DeepSeek-V2-Lite at half the original budget while preserving performance.
-
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.