Expert-parallel communication for Attention–FFN Disaggregation (AFD), built on DeepEP V2
-
Updated
Nov 2, 2026 - Python
#
deepep
Here are 2 public repositories matching this topic...
Hand-built MoE expert parallelism: routing-skew capture, fused grouped-GEMM Triton kernels, 2-all2all dispatch/combine, overlap ablations — proving by counter-example why DeepEP exists.
cuda triton moe nccl mixture-of-experts all-to-all deepseek gpu-profiling distributed-inference kernel-optimization deepep grouped-gemm expert-parallelism
-
Updated
Oct 12, 2026 - Python
Add this topic to your repo
To associate your repository with the deepep topic, visit your repo's landing page and select "manage topics."