3 accepted papers
Traditional Mixture-of-Experts (MoE) training typically proceeds without any structural priors, effectively requiring the model to simultaneously train expert weights while searching for an optimal routing policy within a vast combinatorial space. This entanglement often leads to sluggish convergenc…
Sparse Mixture-of-Experts (MoE) models scale Transformers efficiently but suffer from expert overlap, where different experts process similar tokens and learn redundant functions, resulting in ambiguous routing and underutilized capacity. While architectural solutions like DeepSeek-style shared expe…
As distributed optimization scales to meet the demands of Large Language Model (LLM) training, hardware failures become increasingly non-negligible. Existing fault-tolerant training methods often introduce significant computational or memory overhead, demanding additional resources. To address this…