-
Notifications
You must be signed in to change notification settings - Fork 111
Pull requests: sgl-project/sgl-kernel-npu
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Fix low_latency dispatch&combine checks with bs condition
#436
opened Apr 11, 2026 by
zuje123
Collaborator
Loading…
add dispatch_ffn_combine_bf16 kernel for deepep
#410
opened Mar 27, 2026 by
zuje123
Collaborator
Loading…
[WIP] add fuse_deep_moe_no_buffer for enable-torch-compile
#409
opened Mar 27, 2026 by
jiaming1130
Loading…
add fused_deep_moe test for dispatch_ffn_combine
#400
opened Mar 18, 2026 by
zuje123
Collaborator
Loading…
[WIP] Deepep A5 normal and low-latency operators
#381
opened Feb 24, 2026 by
oagniqgnat
Contributor
Loading…
MMLU benchmark for different inverse implementations
#374
opened Feb 11, 2026 by
gioelegott
Loading…
(tri_inv) (pto-isa) implement AIV triangular inverse using pto-isa
#369
opened Feb 6, 2026 by
zouzias
Contributor
Loading…
wrap triton_kernels into callable that can be traced into a graph
#368
opened Feb 5, 2026 by
lawtherWu
Loading…
deeepep normal support shmem with asymmetric tensor
#328
opened Jan 19, 2026 by
zuje123
Collaborator
Loading…
Dynamo do not support 'with torch.npu.device', delete it
#319
opened Jan 15, 2026 by
DevinP16
Loading…
Previous Next
ProTip!
Filter pull requests by the default branch with base:main.