[npu] use extension for op builder (#5172)

* update extension

* update cpu adam

* update is

* add doc for cpu adam

* update kernel

* update commit

* update flash

* update memory efficient

* update flash attn

* update flash attention loader

* update api

* fix

* update doc

* update example time limit

* reverse change

* fix doc

* remove useless kernel

* fix

* not use warning

* update

* update
This commit is contained in:
Xuanlei Zhao
2024-01-08 11:39:16 +08:00
committed by GitHub
parent d6df19bae7
commit dd2c28a323
35 changed files with 1067 additions and 274 deletions

View File

@@ -1,7 +1,14 @@
from .cpu_adam_loader import CPUAdamLoader
from .cuda_native import FusedScaleMaskSoftmax, LayerNorm, MultiHeadAttention
from .extensions.flash_attention import AttnMaskType
from .flash_attention_loader import ColoAttention, FlashAttentionLoader
__all__ = [
"LayerNorm",
"FusedScaleMaskSoftmax",
"MultiHeadAttention",
"CPUAdamLoader",
"FlashAttentionLoader",
"ColoAttention",
"AttnMaskType",
]