Convert input/output in stats to be MB per iteration #178
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
Differential Revision: D34758134 fbshipit-source-id: dadbc62dc2fb92ff4c68eb6177c46df65d04982d
Summary: Instead of naively showing input as just the pooling factor and output as just the embedding dimension this diff changes planner stats to use the actual size of input & output in terms of megabytes per iteration **input**: global_batch_size * pooling factor * sizeof(dtype of input) **output**: global_batch_size * (output size (1 in pooled)) * sizeof(dytpe of emb) * emb_dim This provides a sense of scale for data coming in and out, and additionally makes plans with multiple sharding types directly comparable. Also fixes a bug with TWCW, we incorrectly specified the ranks as entire world size when it should be limited to the local world of the host that the parameter is sharded on. Differential Revision: D35153224 fbshipit-source-id: 531626ef9d1a6ddf8e9feb756f778d9aa2e406a5
facebook-github-bot
added
CLA Signed
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D35153224 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.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.
Summary:
Instead of naively showing input as just the pooling factor and output as just the embedding dimension
this diff changes planner stats to use the actual size of input & output in terms of megabytes per iteration
input: global_batch_size * pooling factor * sizeof(dtype of input)
output: global_batch_size * (output size (1 in pooled)) * sizeof(dytpe of emb) * emb_dim
This provides a sense of scale for data coming in and out, and additionally makes plans with multiple sharding types directly comparable.
Also fixes a bug with TWCW, we incorrectly specified the ranks as entire world size when it should be limited to the local world of the host that the parameter is sharded on.
Differential Revision: D35153224