Updated vadv2 files#422
Open
Suhana-TT wants to merge 1 commit into
Open
Conversation
Suhana-TT
requested review from
jajuTT,
lyen1,
ramamalladiTT,
ssapreTT and
vmgeorgeTT
as code owners
Contributor
There was a problem hiding this comment.
Pull request overview
This PR updates the VADV2 TTNN workload codepaths and tests, primarily refactoring around tensor slicing/indexing and reducing torch interop in several attention/transformer components.
Changes:
- Replaced several tensor slicing/indexing patterns with shape-based tensor construction.
- Updated VADV2 attention/encoder components to allocate intermediate tensors directly as TTNN tensors (and adjusted compare-threshold handling in
tt_head). - Updated PCC tests’ debug logging formatting and modified the backbone test’s device initialization.
Reviewed changes
Copilot reviewed 11 out of 12 changed files in this pull request and generated 13 comments.
Show a summary per file
| File | Description |
|---|---|
| workloads/ttnn/vadv2/tt/tt_utils.py | Refactors deformable attention sampling grid construction (now shape-based). |
| workloads/ttnn/vadv2/tt/tt_transformer.py | Refactors object/map query embedding splitting logic (now shape-based). |
| workloads/ttnn/vadv2/tt/tt_temporal_self_attention.py | Refactors slice/shape handling for value concatenation and spatial shape normalization. |
| workloads/ttnn/vadv2/tt/tt_spatial_cross_attention.py | Refactors bev_mask per-camera handling and allocates rebatch tensors as TTNN. |
| workloads/ttnn/vadv2/tt/tt_mha.py | Refactors Q/K/V projection splitting from combined weights/biases (now shape-based). |
| workloads/ttnn/vadv2/tt/tt_head.py | Wraps scalar thresholds for ttnn.compare and changes pad tensor construction. |
| workloads/ttnn/vadv2/tt/tt_encoder.py | Refactors lidar2img construction and multiple slice-based computations (now shape-based). |
| workloads/ttnn/vadv2/tt/tt_deformable_attention.py | Refactors spatial shape column extraction (now shape-based). |
| workloads/ttnn/vadv2/tests/pcc/test_tt_head.py | Fixes logger debug output formatting using f-strings. |
| workloads/ttnn/vadv2/tests/pcc/test_tt_backbone.py | Changes test signature and opens a device inside the test. |
| workloads/DiffusionDrive/navsim/agents/diffusiondrive/reference/Transfuser_backbone/comparison_results.md | Updates recorded comparison results content and timestamp. |
| .gitignore | Removes the *.npy ignore entry. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
ramamalladiTT
requested changes
Apr 28, 2026
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 14 out of 14 changed files in this pull request and generated 5 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 14 out of 14 changed files in this pull request and generated 2 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
ramamalladiTT
requested changes
May 12, 2026
Contributor
ramamalladiTT
left a comment
ramamalladiTT
left a comment
There was a problem hiding this comment.
please limit changes to workloads directory. go ahead with the change in workloads/ttnn/vadv2/tt/tt_deformable_attention.py for s0 and s1. resolve the comments. I would we approve post these changes.
Comment on lines
+217
to
+222
|
|
||
| divisor = ttnn.maximum(ref_slice, ones_like * eps_tensor) | ||
| divisor = ttnn.Tensor(shape=divisor.shape, dtype=ttnn.bfloat16, device=self.device, layout=ttnn.Layout.TILE_LAYOUT, data=divisor.data) | ||
| # DEBUG: Check types before the failing line | ||
|
|
||
|
|
Comment on lines
+137
to
+140
| num_levels = spatial_shapes.shape[0] # This is 1 | ||
| slice_shape = (num_levels,) # Shape should be [1] for each column | ||
| # slice_sp_s0 = spatial_shapes[..., 0].shape | ||
| # slice_sp_s1 = spatial_shapes[..., 1].shape |
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
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.
No description provided.