fix flaky test due to input_jkt.weight dtype #2763
Open
+11
−2
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
facebook-github-bot
added
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D70126859 |
TroyGarden
added a commit
to TroyGarden/torchrec
that referenced
this pull request
Feb 24, 2025
Summary: # context * The [test_model_parallel_nccl](https://fb.workplace.com/groups/970281557043698/posts/1863456557726189/?comment_id=1867254224013089) has been reported to be flaky: [paste](https://www.internalfb.com/intern/everpaste/?color=0&handle=GJBrgxaEWkfR-ycEAP_fNV5sl_l1br0LAAAz) * after an in-depth investigation, the root cause is that when the dtype of the generated input KJT._weights is always `torch.float` (i.e., `torch.float32`), but the tested embedding table's dtype could be `torch.FP16`. # changes * added `dtype` argument to the `torch.rand` function in input generation so that the generated input kjt can have the correct dtype for the kjt._weights Differential Revision: D70126859
TroyGarden
force-pushed
the
export-D70126859
branch
from
aa37644 to
3477ac1
Compare
|
This pull request was exported from Phabricator. Differential Revision: D70126859 |
Summary: # context * The [test_model_parallel_nccl](https://fb.workplace.com/groups/970281557043698/posts/1863456557726189/?comment_id=1867254224013089) has been reported to be flaky: [paste](https://www.internalfb.com/intern/everpaste/?color=0&handle=GJBrgxaEWkfR-ycEAP_fNV5sl_l1br0LAAAz) * after an in-depth investigation, the root cause is that when the dtype of the generated input KJT._weights is always `torch.float` (i.e., `torch.float32`), but the test embedding table's dtype could be `torch.FP16`. # changes * convert the input_kjt._weights.dtype to be consistent with `EmbeddingBag.weight.dtype` in EBC (unsharded) Differential Revision: D70126859
TroyGarden
force-pushed
the
export-D70126859
branch
from
3477ac1 to
132725e
Compare
|
This pull request was exported from Phabricator. Differential Revision: D70126859 |
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.
Summary:
context
The test_model_parallel_nccl has been reported to be flaky: paste
after an in-depth investigation, the root cause is that when the dtype of the generated input KJT._weights is always
torch.float(i.e.,torch.float32), but the tested embedding table's dtype could betorch.FP16.changes
dtypeargument to thetorch.randfunction in input generation so that the generated input kjt can have the correct dtype for the kjt._weightsDifferential Revision: D70126859