Skip to content

Simplify dask serializers #8443

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

Merged
rapids-bot merged 1 commit into from
Aug 4, 2026
Merged

Simplify dask serializers #8443

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
44 changes: 13 additions & 31 deletions python/cuml/cuml/comm/serialize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,8 +2,6 @@
# SPDX-FileCopyrightText: Copyright (c) 2020-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0
#


import cudf.comm.serialize # noqa: F401

import cuml
Expand All @@ -18,38 +16,22 @@
from distributed.protocol.serialize import pickle_dumps, pickle_loads

from cuml.ensemble import RandomForestClassifier, RandomForestRegressor
from cuml.naive_bayes import MultinomialNB

# Registering RF Regressor and Classifier to use pickling even when
# Base is serialized with Dask or CUDA serializations
@dask_serialize.register(RandomForestRegressor)
@cuda_serialize.register(RandomForestRegressor)
def rfr_serialize(rf):
return pickle_dumps(rf)

@dask_deserialize.register(RandomForestRegressor)
@cuda_deserialize.register(RandomForestRegressor)
def rfr_deserialize(header, frames):
return pickle_loads(header, frames)

@dask_serialize.register(RandomForestClassifier)
@cuda_serialize.register(RandomForestClassifier)
def rfc_serialize(rf):
return pickle_dumps(rf)

@dask_deserialize.register(RandomForestClassifier)
@cuda_deserialize.register(RandomForestClassifier)
def rfc_deserialize(header, frames):
return pickle_loads(header, frames)

register_generic(cuml.Base, "cuda", cuda_serialize, cuda_deserialize)

register_generic(cuml.Base, "dask", dask_serialize, dask_deserialize)
# These classes require pickling instead of automatically traversing
# __dict__. They contain non-serializable internal attributes, and their
# representation is better serialized via pickle anyway.
pickle_only_classes = (RandomForestRegressor, RandomForestClassifier)

register_generic(MultinomialNB, "cuda", cuda_serialize, cuda_deserialize)
for name, dumps, loads in [
("cuda", cuda_serialize, cuda_deserialize),
("dask", dask_serialize, dask_deserialize),
]:
# Generic implementation for `cuml.Base` estimators
register_generic(cuml.Base, name, dumps, loads)

register_generic(MultinomialNB, "dask", dask_serialize, dask_deserialize)
# Overrides for pickle-only classes
dumps.register(pickle_only_classes, pickle_dumps)
loads.register(pickle_only_classes, pickle_loads)

except ImportError:
# distributed is probably not installed on the system
pass
Loading