Fix no-golden TS rows: feed golden a 0-dim tensor scalar#13
Merged
Conversation
The tensor-scalar path passed a python scalar to both the device op and the torch golden. Several goldens (maximum, minimum, logical_and/or/xor, ldexp, logaddexp/2, div_no_nan and their in-place forms) reject a python scalar 2nd operand, so call_golden returned None -> 605 rows reported "no-golden". Feed the golden a 0-dim tensor of the scalar instead (device still gets the python scalar); torch broadcasts it identically. All 605 no-golden TS rows now produce real pass/fail measurements; TT rows are unchanged. Co-authored-by: Cursor <cursoragent@cursor.com>
Deploying with
|
| Status | Name | Latest Commit | Preview URL | Updated (UTC) |
|---|---|---|---|---|
| ✅ Deployment successful! View logs |
ttnn-ops-coverage | cb58cfb | Commit Preview URL Branch Preview URL |
Jul 03 2026, 10:36 AM |
Aswincloud-Bot
approved these changes
Jul 3, 2026
Aswincloud-Bot
left a comment
Aswincloud-Bot
left a comment
There was a problem hiding this comment.
Auto-approved: @Aswinmcw is a member of @Aswincloud/admins.
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Follow-up to the tensor-scalar (TS) variants work. The TS path passed a python scalar to both the device op and the torch golden, but several goldens reject a python scalar 2nd operand, so
call_goldenreturnedNoneand 605 rows were reported asno-golden(accuracy not measured).Affected ops (all
<op> TS):maximum,minimum,logical_and/or/xor(+ in-place),ldexp(+ in-place),logaddexp,logaddexp2(+ in-place),div_no_nan.Fix
Feed the golden a 0-dim tensor of the scalar (
torch.tensor(s, dtype=int32/float32)); the device op still receives the python scalar. torch broadcasts a 0-dim tensor identically, so those goldens now evaluate.Impact
no-goldenTS rows now produce real pass/fail + PCC.Test plan
bfloat16/tile/dram):maximum/minimum/logaddexp/ldexp/logical_and/div_no_nanTS now report pass/PCC;0no-goldenrows remain.add TSunchanged.Made with Cursor