mirror of
https://github.com/hpcaitech/ColossalAI.git
synced 2025-06-25 15:01:43 +00:00
[cli] hotfix launch command for multi-nodes (#4165)
This commit is contained in:
parent
2ac24040eb
commit
1908caad38
@ -164,9 +164,7 @@ def get_launch_command(
|
||||
]
|
||||
else:
|
||||
# extra launch args for torch distributed launcher with torch >= 1.9
|
||||
default_torchrun_rdzv_args = dict(rdzv_backend="c10d",
|
||||
rdzv_endpoint=f"{master_addr}:{master_port}",
|
||||
rdzv_id="colossalai-default-job")
|
||||
default_torchrun_rdzv_args = dict(master_addr=master_addr, master_port=master_port)
|
||||
|
||||
# update rdzv arguments
|
||||
for key in default_torchrun_rdzv_args.keys():
|
||||
|
Loading…
Reference in New Issue
Block a user