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

Unsqueeze pytorch operator test plan #986

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

vobojevicTT
Copy link
Contributor

@vobojevicTT vobojevicTT commented Dec 30, 2024

Test plan for pytorch unsqueeze operator.

Local run (no failing rules):
16 failed, 2582 passed in 938.44s (0:15:38)

@vobojevicTT vobojevicTT self-assigned this Dec 30, 2024
@vobojevicTT vobojevicTT added the Ops Support new op in tt-forge and tt-mlir label Dec 30, 2024
@vobojevicTT vobojevicTT linked an issue Dec 30, 2024 that may be closed by this pull request
@vobojevicTT vobojevicTT force-pushed the vobojevic/pytorch-unsqueeze-test branch from 34ac560 to 0fe6a0e Compare January 10, 2025 11:07
@vobojevicTT vobojevicTT marked this pull request as ready for review January 10, 2025 11:08
dims = list(range(-dim - 1, dim + 1))

for i in dims:
yield {"dim": i}
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@dgolubovicTT @vbrkicTT Here, we generate all possible dim values, which creates 2598 tests. The test run takes around 20 minutes. Should we skip some dims?

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It doesn't sound yet critical however we might introduce some limits for number of tests on PR/nightly

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I wouldn't skip some dim for all tests. Maybe sampling subset of tests will provide uniform testing for all shapes and dimensions? We could do that by choosing subset of dims randomly with seed based on input shape (therefore we would get the same subset for the same shape). What do you think?

@vobojevicTT vobojevicTT force-pushed the vobojevic/pytorch-unsqueeze-test branch from 0fe6a0e to 0f80fed Compare January 10, 2025 11:14
@vobojevicTT vobojevicTT linked an issue Jan 10, 2025 that may be closed by this pull request
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Ops Support new op in tt-forge and tt-mlir
Projects
None yet
Development

Successfully merging this pull request may close these issues.

[Test Plan] Unsqueeze operator
3 participants