Pytorch parallelism #5916
Open
Pytorch parallelism #5916
Conversation
This file contains 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
Radhika-okhade
added
status: under review
|
Hey! @andersooi Please correct the file path. This is the correct path |
| @@ -0,0 +1,104 @@ | |||
| --- | |||
| Title: 'Distributed Data Parallelism' | |||
| Description: 'An overview of distributed data parallelism in PyTorch.' | |||
There was a problem hiding this comment.
The description is too generic. Explain what distributed data parallelism is within 1-2 lines.
https://github.com/Codecademy/docs/blob/main/documentation/style-guide.md
Comment on lines
+8
to
+10
| - 'PyTorch' | ||
| - 'Data' | ||
| - 'Data Parallelism' |
There was a problem hiding this comment.
Suggested change
| - 'PyTorch' | |
| - 'Data' | |
| - 'Data Parallelism' | |
| - 'Data' | |
| - 'Data Parallelism' | |
| - 'PyTorch' |
|
|
||
| ## Introduction to Distributed Data Parallelism | ||
|
|
||
| Distributed Data Parallelism (DDP) in PyTorch is a module that enables users to efficiently train models across multiple GPUs and machines. By splitting the training process across multiple machines, DDP helps reduce training time and facilitates scaling to larger models and datasets. It achieves parallelism by splitting the input data into smaller chunks, processing them on different GPUs, and aggregating results for updates. Compared to `DataParallel`, DDP offers better performance and scalability by minimising device communication overhead. |
There was a problem hiding this comment.
Suggested change
| Distributed Data Parallelism (DDP) in PyTorch is a module that enables users to efficiently train models across multiple GPUs and machines. By splitting the training process across multiple machines, DDP helps reduce training time and facilitates scaling to larger models and datasets. It achieves parallelism by splitting the input data into smaller chunks, processing them on different GPUs, and aggregating results for updates. Compared to `DataParallel`, DDP offers better performance and scalability by minimising device communication overhead. | |
| Distributed Data Parallelism (DDP) in PyTorch is a module that enables users to efficiently train models across multiple GPUs and machines. By splitting the training process across multiple machines, DDP helps reduce training time and facilitates scaling to larger models and datasets. | |
| It achieves parallelism by splitting the input data into smaller chunks, processing them on different GPUs, and aggregating results for updates. Compared to `DataParallel`, DDP offers better performance and scalability by minimizing device communication overhead. |
|
|
||
| To use DDP, a distributed process group needs to be initialised and wrapped to a model with `torch.nn.parallel.DistributedDataParallel`. | ||
|
|
||
| ```py |
There was a problem hiding this comment.
Suggested change
| ```py | |
| ```pseudo |
| def setup(rank, world_size): | ||
| os.environ['MASTER_ADDR'] = 'localhost' | ||
| os.environ['MASTER_PORT'] = '8000' | ||
| dist.init_process("nccl", rank=rank, world_size=world_size) |
There was a problem hiding this comment.
Suggested change
| dist.init_process("nccl", rank=rank, world_size=world_size) | |
| dist.init_process_group("nccl", rank=rank, world_size=world_size) |
|
Hey! @andersooi, Thank you for contributing to Codecademy docs. I have made a few suggestions; please go through them and make the necessary changes. |
Radhika-okhade
added
status: waiting for author
and removed
status: under review
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Description
Issue Solved
Closes #5871
Type of Change
Checklist
mainbranch.Issues Solvedsection.