Skip to content

[CUDA] Support SwiGlu in MoE and qMoE #25530

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 6 commits into from
Jul 28, 2025
Merged

[CUDA] Support SwiGlu in MoE and qMoE #25530

Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
755e33c
support swiglu
tianleiwu Jul 24, 2025
514b072
only build interleaved version
tianleiwu Jul 24, 2025
63d8808
Update qMoE
tianleiwu Jul 28, 2025
e24271e
Update operator doc
tianleiwu Jul 28, 2025
4c07bcf
static cast
tianleiwu Jul 28, 2025
360f4d0
add test for 4 bits
tianleiwu Jul 28, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
static cast
  • Loading branch information
@tianleiwu
tianleiwu committed Jul 28, 2025
commit 4c07bcf6dc65c9919c1af3e7a708e1a5efe509de
2 changes: 1 addition & 1 deletion onnxruntime/contrib_ops/cuda/moe/ft_moe/moe_kernel.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1001,7 +1001,7 @@ void CutlassMoeFCRunner<T, WeightType, Enable>::run_moe_fc(
swiglu_output_buffer + total_past_rows_ * inter_size,
gemm1_output_buffer + total_past_rows_ * 2 * inter_size,
inter_size,
total_covered_rows_,
static_cast<int>(total_covered_rows_),
swiglu_alpha,
stream);

Expand Down
Loading