[npu] change device to accelerator api (#5239)

* update accelerator

* fix timer

* fix amp

* update

* fix

* update bug

* add error raise

* fix autocast

* fix set device

* remove doc accelerator

* update doc

* update doc

* update doc

* use nullcontext

* update cpu

* update null context

* change time limit for example

* udpate

* update

* update

* update

* [npu] polish accelerator code

---------

Co-authored-by: Xuanlei Zhao <xuanlei.zhao@gmail.com>
Co-authored-by: zxl <43881818+oahzxl@users.noreply.github.com>
This commit is contained in:
Hongxin Liu
2024-01-09 10:20:05 +08:00
committed by GitHub
parent dd2c28a323
commit d202cc28c0
128 changed files with 1773 additions and 868 deletions

View File

@@ -1,9 +1,9 @@
import torch
from colossalai.accelerator import get_accelerator
from colossalai.legacy.context import ParallelMode
from colossalai.legacy.core import global_context as gpc
from colossalai.legacy.nn import TransformerSelfAttentionRing
from colossalai.utils import get_current_device
def check_selfattention():
@@ -13,10 +13,10 @@ def check_selfattention():
HIDDEN_SIZE = 16
layer = TransformerSelfAttentionRing(16, 8, 8, 0.1)
layer = layer.to(get_current_device())
layer = layer.to(get_accelerator().get_current_device())
hidden_states = torch.rand(SUB_SEQ_LENGTH, BATCH, HIDDEN_SIZE).to(get_current_device())
hidden_states = torch.rand(SUB_SEQ_LENGTH, BATCH, HIDDEN_SIZE).to(get_accelerator().get_current_device())
attention_mask = torch.randint(low=0, high=2, size=(BATCH, 1, 1, 1, SUB_SEQ_LENGTH * WORLD_SIZE)).to(
get_current_device()
get_accelerator().get_current_device()
)
layer(hidden_states, attention_mask)