add sparse update section in fluid dist doc #8997
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
typhoonzero
reviewed
Mar 12, 2018
|
|
||
| ### Sparse Update | ||
|
|
||
| For an embedding layer, the gradient maybe be very sparse(upper 90% is zero) for each mini-batch. |
Contributor
There was a problem hiding this comment.
There should be a space between sparse and (.
Contributor
There was a problem hiding this comment.
the gradient maybe be very sparse => the gradient may have many rows containing only 0
| ### Sparse Update | ||
|
|
||
| For an embedding layer, the gradient maybe be very sparse(upper 90% is zero) for each mini-batch. | ||
| Fluid use [SelectedRows](../selected_rows.md) to support the sparse variable. Distributed training support `Sparse Update`, |
Contributor
There was a problem hiding this comment.
the sparse variable => sparse variables.
Contributor
There was a problem hiding this comment.
Distributed training support Sparse Update, which sends a SelectedRows variable to the parameter server to run parameter updates.
typhoonzero
reviewed
Mar 12, 2018
| It would save a lot of bandwidth and make the distributed training job have better performance. | ||
| For embedding layers, the gradient may have many rows containing only 0 when training, | ||
| if the gradient use a dense tensor to do parameter optimization, | ||
| it could spend unnessesary memory, slow down the calculations and waste |
Contributor
There was a problem hiding this comment.
unnessesary => unnecessary
helinwang
reviewed
Mar 12, 2018
| ### Sparse Update | ||
|
|
||
| For embedding layers, the gradient may have many rows containing only 0 when training, | ||
| if the gradient use a dense tensor to do parameter optimization, |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixed #8996