Skip to content

Conversation

@jacquesqiao
Copy link
Member

@jacquesqiao jacquesqiao commented Mar 26, 2018

project: #9211
fix: #9372

@jacquesqiao jacquesqiao mentioned this pull request Mar 26, 2018
15 tasks
Copy link
Contributor

@typhoonzero typhoonzero left a comment

Choose a reason for hiding this comment

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

Just one comment, all others LGTM


const size_t shard_num = outs.size();

std::vector<std::vector<T>> out_ids;
Copy link
Contributor

Choose a reason for hiding this comment

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

Maybe can remove this temp vector and push data directly to outs[i] to reduce one for loop.

Copy link
Member Author

@jacquesqiao jacquesqiao Mar 26, 2018

Choose a reason for hiding this comment

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

The problem is that we don't know the shape of outs[i], so we cannot call mutable_data to allocate memory for it. The first loop is used to get the shape of each outs.

Copy link
Contributor

Choose a reason for hiding this comment

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

True.

@typhoonzero
Copy link
Contributor

Don't know why build android on travis-ci stucks.

@jacquesqiao
Copy link
Member Author

@typhoonzero Ok, I will restart the CI.

Copy link
Contributor

@typhoonzero typhoonzero left a comment

Choose a reason for hiding this comment

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

LGTM!

@jacquesqiao jacquesqiao merged commit f3dc311 into PaddlePaddle:develop Mar 26, 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.

add split_ids_op

2 participants