Skip to content

Add the support of profiling and multi-GPU running #636

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

Jump to bottom
Merged
merged 13 commits into from
Feb 7, 2018
Merged

Add the support of profiling and multi-GPU running #636

merged 13 commits into from
Feb 7, 2018

Conversation

@kuke
Copy link
Collaborator

@kuke kuke commented Feb 5, 2018 •
edited
Loading

Resolve #622
@kuke kuke changed the title Add the support of profiling and multi-GPUs running [WIP]Add the support of profiling and multi-GPUs running Feb 5, 2018
@kuke kuke added the DeepASR label Feb 5, 2018
@kuke kuke mentioned this pull request Feb 5, 2018
Closed
@kuke kuke mentioned this pull request Feb 5, 2018
Closed
3 tasks
zhxfl
zhxfl previously approved these changes Feb 6, 2018
View reviewed changes
@kuke kuke changed the title [WIP]Add the support of profiling and multi-GPUs running Add the support of profiling and multi-GPUs running Feb 6, 2018
@kuke kuke requested a review from pkuyym February 6, 2018 12:05
@kuke kuke dismissed zhxfl’s stale review February 6, 2018 12:06

Updated after approved

pkuyym
pkuyym requested changes Feb 6, 2018
View reviewed changes
fluid/DeepASR/train.py Outdated
trans_splice.TransSplice()
]

res_feature = fluid.LoDTensor()
Copy link
Contributor

Choose a reason for hiding this comment

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

Please think a better name.
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
fluid/DeepASR/train.py Outdated
test_data_reader.batch_iterator(args.batch_size,
args.minimum_batch_size)):
# load_data
(bat_feature, bat_label, lod) = batch_data
Copy link
Contributor

Choose a reason for hiding this comment

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

Same as above
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
kuke
kuke commented Feb 7, 2018
View reviewed changes
Copy link
Collaborator Author

@kuke kuke left a comment

Choose a reason for hiding this comment

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

All done, thx
fluid/DeepASR/train.py Outdated
trans_splice.TransSplice()
]

res_feature = fluid.LoDTensor()
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
fluid/DeepASR/train.py Outdated
test_data_reader.batch_iterator(args.batch_size,
args.minimum_batch_size)):
# load_data
(bat_feature, bat_label, lod) = batch_data
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
@zhxfl
Copy link
Member

zhxfl commented Feb 7, 2018

LGTM
pkuyym
pkuyym requested changes Feb 7, 2018
View reviewed changes
fluid/DeepASR/model.py Outdated
stacked_num,
parallel=False,
is_train=True,
class_num=1749):
Copy link
Contributor

Choose a reason for hiding this comment

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

I think we should not give a default value for class_num
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
fluid/DeepASR/model.py
def _net_conf(feature, label):
seq_conv1 = fluid.layers.sequence_conv(
input=feature,
num_filters=1024,
Copy link
Contributor

Choose a reason for hiding this comment

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

We can make this configurable later, just mark here.
fluid/DeepASR/model.py
size=class_num,
act='softmax')

cost = fluid.layers.cross_entropy(input=prediction, label=label)
Copy link
Contributor

Choose a reason for hiding this comment

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

If is_traing==False, cost is unnecessary, since there maybe no label.
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

cost and acc will be pruned when saving inference model
@pkuyym
Copy link
Contributor

pkuyym commented Feb 7, 2018

Please consider reorganize the directory structure, for example:

model_utils/model.py train.py utils/profile.py test.py
kuke
kuke commented Feb 7, 2018
View reviewed changes
Copy link
Collaborator Author

@kuke kuke left a comment

Choose a reason for hiding this comment

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

Updated. Thx
fluid/DeepASR/model.py Outdated
stacked_num,
parallel=False,
is_train=True,
class_num=1749):
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
fluid/DeepASR/model.py
size=class_num,
act='softmax')

cost = fluid.layers.cross_entropy(input=prediction, label=label)
Copy link
Collaborator Author

Choose a reason for hiding this comment

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

cost and acc will be pruned when saving inference model
pkuyym
pkuyym approved these changes Feb 7, 2018
View reviewed changes
Copy link
Contributor

@pkuyym pkuyym left a comment

Choose a reason for hiding this comment

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

LGTM
@kuke kuke merged commit 2738ca1 into PaddlePaddle:develop Feb 7, 2018
@kuke kuke changed the title Add the support of profiling and multi-GPUs running Add the support of profiling and multi-GPU running Feb 8, 2018
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

DeepASR

3 participants

@kuke @zhxfl @pkuyym