Using acc_t for log_softmax #143896
Using acc_t for log_softmax #143896
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/143896
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New Failure, 1 Cancelled Job, 4 Unrelated FailuresAs of commit 44e69b2 with merge base 28796f7 ( NEW FAILURE - The following job has failed:
CANCELLED JOB - The following job was cancelled. Please retry:
FLAKY - The following job failed but was likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but was present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
UNSTABLE - The following jobs are marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
@pytorchbot rebase |
|
@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here |
|
Successfully rebased |
|
@pytorchbot rebase |
|
@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here |
|
Successfully rebased |
|
@pytorchmergebot merge -i |
Merge failedReason: Approvers from one of the following sets are needed:
|
|
Hi @cyyever, thanks for the approval! Do you have any suggestions of the reviewers to land this PR? |
There was a problem hiding this comment.
Copilot reviewed 1 out of 2 changed files in this pull request and generated no comments.
Files not reviewed (1)
- aten/src/ATen/native/cpu/SoftMaxKernel.cpp: Language not supported
Comments suppressed due to low confidence (1)
test/test_nn.py:13260
- [nitpick] The variable name 'input' shadows the built-in function; consider renaming it (e.g., to 'input_tensor' or 'inp') to avoid potential issues.
input = torch.randn(2, 2, 2, 2, 2, dtype=torch.bfloat16)
|
@Skylion007 May help review? |
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
|
@pytorchbot rebase -b main |
|
@pytorchbot started a rebase job onto refs/remotes/origin/main. Check the current status here |
|
Successfully rebased |
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
This PR is to fix #140222. Using high precision as the accumulate type for log_softmax forward. Reproducer in the issue can pass now.
Stack from ghstack (oldest at bottom):
cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @jerryzh168