Instructions to use kernels-community/megablocks with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use kernels-community/megablocks with Kernels:
# !pip install kernels from kernels import get_kernel # a version (or an explicit revision) is required; see the "Files and versions" tab for the available ones kernel = get_kernel("kernels-community/megablocks", version=1) - Notebooks
- Google Colab
- Kaggle
Commit History
-
update binned_copy kernels
3outeille5ba4d9f
-
fix: include torch compile flag
drbh8176cbe -
fix: support torch compile via fake tensors
drbhcd5b9c4
-
fix: add quickstart and avoid autotune when no cuda
drbh09e15a7
-
feat: support shared experts layer and tests
drbh89e2950
-
fix: extract expert device mesh for group from unused prehook
drbhaa23f77
-
fix: prefer using passed parallel group
drbh76c7de7 -
fix: adjust layer params in source
drbh9a1816c
-
fix: improve expert parallel implementation and refactors
drbhe47036a -
fix: add parallel forward functional logic
drbh13afbbe
-
fix: fully vendor stk and fix imports
drbh63599de -
fix: vendor stk decorators
drbh0586ba6 -
feat: add functional version of moe class and layer
drbhdabb815
-
fix: prefer relative imports
drbh484fde0 -
feat: vendor grouped gemm
drbh3224250
-
feat: validate build with original test suite
drbh9c4ca75
-
feat: initial port of megablocks to builder format
drbh2595c46