4 people 7f9173dfa2 [ROCm][Perf] Kimi-K3 Fused kernel for KDA decode (#50654) ...

Signed-off-by: kliuae <[email protected]>
Co-authored-by: Claude <[email protected]>
Co-authored-by: TJian <[email protected]>

2026-08-11 20:39:34 -07:00

..

__init__.py

[ROCm][Perf] Kimi-K3 Shard Latent MoE up-projection for ROCm path (#51253)

2026-08-07 06:11:18 +00:00

test_amd_attn_res.py

[Kimi-K3][AMD] Fuse AttnRes state updates and norms (#50593)

2026-08-04 12:32:13 -04:00

test_amd_kda_decode.py

[ROCm][Perf] Kimi-K3 Fused kernel for KDA decode (#50654)

2026-08-11 20:39:34 -07:00

test_amd_latent_moe_runner.py

[ROCm][Perf] Kimi-K3 Shard Latent MoE up-projection for ROCm path (#51253)

2026-08-07 06:11:18 +00:00

test_attn_res.py

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

test_eagle3.py

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

test_kda_metadata.py

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

test_kda.py

[Bugfix][Kimi-K3] Give the AMD packed KDA decode kernel the state-index stride (#51682)

2026-08-11 00:05:20 +08:00

test_latent_moe_tail.py

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

test_sequence_parallel.py

[Kimi K3 Perf] option to shard the shared expert for non mega case, 16.98 GiB memory/GPU saved (#50912)

2026-08-04 14:00:44 -04:00