hahazhky 0b12c2acf7 [Kernel] Remove cumsum in groupedmatmul (#987) ...

### What this PR does / why we need it remove cumsum operator in MOE to improve performance

### How was this patch tested?
it should be tested on a case with mc2 operator and graph mode enabled

Signed-off-by: zhky <hahazhky@163.com>
Co-authored-by: 洪炜杰 <hongweijie1@huawei.com>

2025-06-06 19:17:27 +08:00

..

__init__.py

[aclgraph] implentment NPUPiecewiseBackend to enable aclgraph (#836)

2025-05-29 11:58:26 +08:00

activation.py

Optimize qwen2_vl and qwen2_5_vl (#701)

2025-04-30 14:22:38 +08:00

attention.py

[1/N][UT][v1 MTP] add basic v1 mtp features (#890)

2025-05-30 08:59:58 +08:00

cache.py

port deepseekv2 and mtp to main branch (#429)

2025-04-19 17:38:18 +08:00

common_fused_moe.py

[Attention][Kernel]moe support for llama4 and mllama4 (#740)

2025-05-13 19:12:40 +08:00

fused_moe.py

[Kernel] Remove cumsum in groupedmatmul (#987)

2025-06-06 19:17:27 +08:00

layernorm.py

[CI]Add model basic accuracy test(Qwen2.5-0.5B-Instruct) (#460)

2025-04-17 14:59:56 +08:00

rotary_embedding.py

[Bugfix] Correct method call for _set_cos_sin_cache (#774)

2025-05-09 12:55:57 +08:00

vocab_parallel_embedding.py

port deepseekv2 and mtp to main branch (#429)

2025-04-19 17:38:18 +08:00