Hugging Face's logo Hugging Face
  • Models
  • Datasets
  • Spaces
  • Buckets new
  • Docs
  • Enterprise
  • Pricing
    • Website
      • Tasks
      • HuggingChat
      • Collections
      • Languages
      • Organizations
    • Community
      • Blog
      • Posts
      • Daily Papers
      • Hardware
      • Learn
      • Discord
      • Forum
      • GitHub
    • Solutions
      • Team & Enterprise
      • Hugging Face PRO
      • Enterprise Support
      • Inference Providers
      • Inference Endpoints
      • Storage Buckets

  • Log In
  • Sign Up

Kernels:
phanerozoic
/
grouped-gemm-moe

Kernel card Files Files and versions
xet
Community
grouped-gemm-moe
7.48 MB
Ctrl+K
Ctrl+K
  • 2 contributors
History: 12 commits
phanerozoic's picture
phanerozoic
Card: standardized form with hero
f5bbb16 verified 17 days ago
  • build
    Uploaded using `kernel-builder`. 24 days ago
  • grouped_gemm_moe_cuda
    grouped-gemm-moe: cp.async-pipelined mma.sync bf16 forward (19x over the tiled path) 24 days ago
  • tests
    Add fused MoE backward: trainable end to end (dx, dW, routing-weight grads) 25 days ago
  • torch-ext
    Add fused MoE backward: trainable end to end (dx, dW, routing-weight grads) 25 days ago
  • .gitattributes
    127 Bytes
    initial commit 25 days ago
  • CARD.md
    5.31 kB
    Card: standardized form with hero 17 days ago
  • README.md
    5.31 kB
    Card: standardized form with hero 17 days ago
  • build.toml
    428 Bytes
    Add grouped-gemm-moe source, tests, and model card 25 days ago
  • flake.nix
    262 Bytes
    Add grouped-gemm-moe source, tests, and model card 25 days ago