Skip to content

PipelineRL -- keep cache on weight update #1694

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
hari-hm wants to merge 1 commit into
base: main
Choose a base branch
from
Open

PipelineRL -- keep cache on weight update #1694

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 2 additions & 1 deletion slime/backends/megatron_utils/update_weight/update_weight_from_distributed.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,8 @@ def update_weights(self) -> None:

if dist.get_rank() == 0:
ray.get([engine.pause_generation.remote() for engine in self.rollout_engines])
ray.get([engine.flush_cache.remote() for engine in self.rollout_engines])
if not getattr(self.args, "keep_cache_on_weight_update", False):
ray.get([engine.flush_cache.remote() for engine in self.rollout_engines])

# int4/fp4 pre_process
if self.quantization_config and self.quantization_config["quant_method"] in ["compressed-tensors"]:
Expand Down
6 changes: 6 additions & 0 deletions slime/utils/arguments.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -434,6 +434,12 @@ def add_rollout_arguments(parser):
default=1,
help="Interval for updating the weights",
)
parser.add_argument(
"--keep-cache-on-weight-update",
action="store_true",
default=False,
help="If set, skip flushing the rollout engine cache during weight updates.",
)
parser.add_argument(
"--keep-old-actor",
action="store_true",
Expand Down