fix: replace torch.device("cuda") with torch.device("cuda:0") in devices initialization #3184
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
vblagoje
requested review from
masci and
sjrl
and removed request for
a team
|
wdyt @sjrl ? This is what I had in mind ☝️ |
sjrl
reviewed
Sep 8, 2022
haystack/modeling/utils.py
Outdated
| @@ -96,6 +96,12 @@ def initialize_device_settings( | |||
| n_gpu = 1 | |||
| # Initializes the distributed backend which will take care of sychronizing nodes/GPUs | |||
| torch.distributed.init_process_group(backend="nccl") | |||
|
|
|||
| # HF transformers v4.21.2 pipeline object doesn't accept torch.device("cuda"), it has to be indexed cuda device | |||
There was a problem hiding this comment.
small typo add "an" to "to be an indexed ..."
sjrl
reviewed
Sep 8, 2022
| @@ -96,6 +96,12 @@ def initialize_device_settings( | |||
| n_gpu = 1 | |||
| # Initializes the distributed backend which will take care of sychronizing nodes/GPUs | |||
| torch.distributed.init_process_group(backend="nccl") | |||
|
|
|||
| # HF transformers v4.21.2 pipeline object doesn't accept torch.device("cuda"), it has to be indexed cuda device | |||
| # TODO eventually remove once the limitation is fixed in HF transformers | |||
There was a problem hiding this comment.
Can you make this TODO an issue, so we can keep track in GitHub?
sjrl
approved these changes
Sep 8, 2022
|
Corrected comment and opened #3185 |
masci
changed the title
brandenchan
pushed a commit
that referenced
this pull request
Sep 21, 2022
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Outdated suggestions cannot be applied.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.
Related Issues
Proposed Changes:
As version 4.21.2 in their pipelines, HF transformers only accept indexed Cuda devices. Therefore, we needed to replace instances of
torch.device("cuda")withtorch.device("cuda:0")in devices initialization util function.How did you test it?
A fix is trivial; I tried a small unit test in an interpreter but didn't add any additional unit tests
Notes for the reviewer
Think of a potential scenario where the proposed fix breaks.