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.

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

Enable ParallelExecutor in SSD-MobileNet and Refine code. #832

Merged
merged 3 commits into from
Apr 12, 2018

Conversation

qingqing01
Copy link
Collaborator

No description provided.

batch_size,
num_passes,
model_save_dir,
pretrained_model=None):
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

parallel_do need have an argument to define using nccl or not. The default value does not use nccl.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done.

return os.path.exists(os.path.join(pretrained_model, var.name))
fluid.io.load_vars(exe, pretrained_model, predicate=if_exist)

train_exe = fluid.ParallelExecutor(use_cuda=True, loss_name=loss.name)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

line 218 and line 227 is not consistent, currently ParallelExecutor only allows to use GPU.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here change use_cuda=arg.use_gpu. Thanks!

for _, data in enumerate(test_reader()):
test_map = exe.run(test_program,
feed=feeder.feed(data),
fetch_list=[accum_map])
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

exe.run is not the format of parallel_exe.

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The testing use Executor, not ParallelExecutor.

Copy link
Contributor

@chengduoZH chengduoZH left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@qingqing01 qingqing01 merged commit 6fa3050 into PaddlePaddle:develop Apr 12, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants