Skip to content
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

[dask] run one training task on each worker #4132

Merged
merged 4 commits into from
Mar 29, 2021
Merged

[dask] run one training task on each worker #4132

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
c3eac4d
[dask] run one training task on each worker
jameslamb Mar 29, 2021
c893d25
add comment on pure
jameslamb Mar 29, 2021
3a80201
missing ticks
jameslamb Mar 29, 2021
ed75fb5
empty commit
jameslamb Mar 29, 2021
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
10 changes: 10 additions & 0 deletions python-package/lightgbm/dask.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -380,6 +380,13 @@ def _train(
num_machines = len(worker_address_to_port)

# Tell each worker to train on the parts that it has locally
#
# This code treates ``_train_part()`` calls as not "pure" because:
# 1. there is randomness in the training process unless parameters ``seed``
# and ``is_deterministic`` are set
# 2. even with those parameters set, the output of one ``_train_part()`` call
# relies on global state (it and all the other LightGBM training processes
# coordinate with each other)
futures_classifiers = [
client.submit(
_train_part,
Expand All @@ -391,6 +398,9 @@ def _train(
num_machines=num_machines,
time_out=params.get('time_out', 120),
return_model=(worker == master_worker),
workers=[worker],
allow_other_workers=False,
pure=False,
**kwargs
)
for worker, list_of_parts in worker_map.items()
Expand Down