github
/
ColossalAI
mirror of https://github.com/hpcaitech/ColossalAI
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

[doc] update docstring in ProcessGroup (#1468)

pull/1469/head
Jiarui Fang 2022-08-19 13:41:57 +08:00 committed by GitHub
parent b73fb7a077
commit 1b491ad7de
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 3 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
colossalai/tensor/process_group.py
View File

@ -31,7 +31,7 @@ PYTORCHPGDICT_ = PyTorchProcessGroupDict()
class ProcessGroup: class ProcessGroup:
"""ProcessGroup """ProcessGroup
Process Group contains group partition for Tensor Parallel and Data Parallel. Process Group indicates how processes are organized in groups for parallel execution using Tensor Parallelism and Data Parallelism.
NOTE, the ProcessGroup must be used after `torch.distributed.initialize()` NOTE, the ProcessGroup must be used after `torch.distributed.initialize()`
@ -40,8 +40,8 @@ class ProcessGroup:
rank: the global rank of the current process. rank: the global rank of the current process.
ranks: List[int], a list of rank id belongings to this process group. ranks: List[int], a list of rank id belongings to this process group.
backend: str, the backend of the process group. backend: str, the backend of the process group.
tp_degree: Optional[int], tensor parallelism degree, default None means 1 tp_degree: Optional[int], tensor parallelism degree. How many processes are inside a tp process group. default None means 1.
dp_degree: Optional[int], data parallelism degree, default None means len(ranks) dp_degree: Optional[int], data parallelism degree. How many processes are inside a dp process group. . default None means len(ranks).
""" """
def __init__(self, def __init__(self,