wassname/pytorch-lightning
1
0
Fork 0
mirror of https://github.com/wassname/pytorch-lightning.git synced 2026-09-11 12:31:23 +08:00
Code Issues Packages Projects Releases Wiki Activity

fixed nccl init

Browse Source
This commit is contained in:
William Falcon
2019-07-13 22:16:10 -04:00
parent 4f5eef2e78
commit dc87a4fc91
1 changed files with 0 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
pytorch_lightning/models/trainer.py
-3
View File
@@ -379,9 +379,6 @@ class Trainer(TrainerIO):
:param tries:
:return:
"""
# hack to get nccl to stop throwing error... seems to be an nccl race condition
if self.proc_rank > 0:
sleep(10.0)
root_node = os.environ['SLURM_NODELIST'].split(' ')[0]
os.environ['MASTER_ADDR'] = root_node