Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[BUG] inference ops unit tests are failing #6839

Open
oelayan7 opened this issue Dec 9, 2024 · 2 comments
Open

[BUG] inference ops unit tests are failing #6839

oelayan7 opened this issue Dec 9, 2024 · 2 comments
Assignees
Labels
bug Something isn't working inference

Comments

@oelayan7
Copy link
Contributor

oelayan7 commented Dec 9, 2024

It was seen that tests under unit/ops/transformer/inference are not being run in any CI job.
Some tests are failing in that directory (examples will be provided below), I have talked to @loadams about it and he tried running them on a V100 setup.
The results he got were 440 failed, 2598 passed, 8 skipped for those tests.

Example for the tests we saw them failing were:

  • unit/ops/transformer/inference/test_bias_geglu.py::test_gated_silu and the failure was on different results than the reference.
  • unit/ops/transformer/inference/test_layer_norm.py::test_layer_norm and the failure was Feature '.bf16' requires .target sm_80 or higher

A hint that could help, those tests are permutated over the supported dtypes, and the failures are always in dtype2 (I assume it is bf16).

@oelayan7 oelayan7 added bug Something isn't working inference labels Dec 9, 2024
@oelayan7
Copy link
Contributor Author

oelayan7 commented Dec 9, 2024

test_layer_norm_residual, test_residual_add, test_bias_geglu, test_moe_residual_matmul, test_pre_norm, test_rms_norm

@loadams loadams self-assigned this Dec 9, 2024
@loadams
Copy link
Contributor

loadams commented Dec 16, 2024

Thanks @oelayan7 - updating this, when we currently run off of the master branch, I see the following:

 68 failed, 1064 passed, 2545 skipped, 2103 deselected, 4 warnings in 226.45s (0:03:46) 

Will create and link a PR that reproduces this.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working inference
Projects
None yet
Development

No branches or pull requests

2 participants