Bolin SunandClaude Fable 5 d410fc12f3 [Kernel] Enable Kimi-K3 SiTU on the CuteDSL MoE backend and the SM107 low-latency GEMM plan (#54606) ...

Signed-off-by: Bolin Sun <bolins@nvidia.com>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>

2026-09-03 16:49:31 +00:00

..

ops

[Kimi-K3] Overlap low-M TP8 KDA projections (#54697)

2026-09-02 03:35:28 +00:00

__init__.py

[Kimi-K3] Add AttnRes kernels (#50090)

2026-07-28 00:22:09 -07:00

dspark_mla.py

Reduce AutoWeightsLoader kwargs (#53106)

2026-08-20 15:40:11 +00:00

kda_metadata.py

[Kimi Bug] Fix cannot access local variable 'active_non_spec_mask_cpu' (#54781)

2026-09-01 21:08:50 +00:00

kda.py

[Kimi-K3] Overlap low-M TP8 KDA projections (#54697)

2026-09-02 03:35:28 +00:00

latent_moe_runner.py

[Bugfix][Kimi K3] Enable deferred MoE finalization before weight loading (#53327)

2026-08-22 11:23:32 +08:00

low_latency_gemm.py

[Kernel] Enable Kimi-K3 SiTU on the CuteDSL MoE backend and the SM107 low-latency GEMM plan (#54606)

2026-09-03 16:49:31 +00:00

mla.py

[Kimi-K3] Merge MLA gate into QKV-A projection (#54015)

2026-08-27 17:47:18 -07:00

model.py

[Kimi-K3] Overlap low-M TP8 KDA projections (#54697)

2026-09-02 03:35:28 +00:00

mtp.py

[Kimi-K3] Merge MLA gate into QKV-A projection (#54015)

2026-08-27 17:47:18 -07:00