Repository navigation
Qualcomm: backport remaining YOLO26 QNN fixes - #21
Merged
Merged
Conversation
… 0D select Remove0DTensor replaced a 0D select with its whole source, which is only right for a 1-element source: y + x[1] became y + x, and torch.full(shape, t[i]) became a copy of all of t (lowering then failed with "expand: ... length 3 -> 1"). A 1-element slice of the selected element replaces it instead. Backport of pytorch/executorch#24078. 1.4.1 has no rework pass tests.
…loat An integer cast to a quantized float tensor lowered to QNN Cast, which copies the integer's raw value into the fixed-point tensor (56 became 56 * scale), and the HTP rejects Convert there. Cast to fp32, then Quantize. Backport of pytorch/executorch#24079.
AnnotateUnbind read the quant attrs from the quantize node feeding unbind's dequantize, but a folded quantized constant feeds the dequantize directly, and the pass crashed with "'str' object has no attribute '_schema'". Take them from the dequantize then. Backport of pytorch/executorch#24080. 1.4.1 has no rework pass tests, so the test is an end-to-end lowering test in test_passes.py here.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Backports of pytorch/executorch#24078, #24079, #24080. Follow-up to #14.
New tests fail on the base and pass here;
test_passes.pyand the x86-simulator FP/quantized operator and FP model suites pass.