Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 4 additions & 5 deletions train_diloco.py
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@
ProcessGroupGloo,
ProcessGroupNCCL,
)
from torchft.checkpointing.pg_transport import PGTransport
from torchft.checkpointing.http_transport import HTTPTransport
from torchft.local_sgd import DiLoCo

logging.basicConfig(level=logging.INFO)
Expand Down Expand Up @@ -67,13 +67,12 @@ def state_dict():
timeout=timedelta(seconds=10),
)
if torch.cuda.is_available() and USE_NCCL
else ProcessGroupGloo(timeout=timedelta(seconds=5))
else ProcessGroupGloo(timeout=timedelta(seconds=10))
)

transport = PGTransport(
pg,
transport = HTTPTransport(
timeout=timedelta(seconds=10),
device=device,
num_chunks=0,
)

manager = Manager(
Expand Down