Add ops(unflatten) | feat(atenlib) #612
Conversation
titaiwangms
added
the
module: torchlib
Codecov Report
@@ Coverage Diff @@
## main #612 +/- ##
==========================================
+ Coverage 74.35% 74.38% +0.03%
==========================================
Files 107 107
Lines 11209 11226 +17
Branches 1161 1162 +1
==========================================
+ Hits 8334 8351 +17
Misses 2560 2560
Partials 315 315
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
| dim = self_rank + dim | ||
|
|
||
| head_start_idx = op.Constant(value_ints=[0]) | ||
| head_end_idx = op.Reshape(dim, op.Constant(value_ints=[1])) |
There was a problem hiding this comment.
@gramalingam @justinchuby
ORTEvaluator doesn't seem to understand:
head_end_idx = op.Reshape(dim, [1]))
| skip( | ||
| "unflatten", | ||
| matcher=lambda sample: any(dim == 0 for dim in sample.input.shape), | ||
| reason="0 dim in ONNX is undefined behavior.", | ||
| ), |
There was a problem hiding this comment.
Can we create a condition to handle this case?
There was a problem hiding this comment.
To raise a warning? Not sure what we should give back to users.
There was a problem hiding this comment.
What is pytorch's behavior?
To raise a warning? Not sure what we should give back to users.
I was suggesting handling it like pytorch does if possible
There was a problem hiding this comment.
PyTorch takes this because 0-d is in their spec.
For example,
a = torch.randn(4, 0)
# tensor([], size=(4, 0))
torch.unflatten(a, 0, (2, 2))
# tensor([], size=(2, 2, 0))There was a problem hiding this comment.
I wonder if torch_libs has a certain way to inform users when the difference between specs is hit? In TorchScript exporter, we could raise UnsupportedONNXError
There was a problem hiding this comment.
I see. We could skip this if it is too complicated and come back to it as needed. I would update the skip reason to say something to the effect of "fixme: logic not implemented for size 0 inputs"
There was a problem hiding this comment.
We don't raise errors in atenlib. The exporter can raise errors when it sees a difference after OpSchema is implemented
| @@ -100,9 +100,6 @@ def aten_addmm( | |||
| ) -> TFloat: | |||
| """addmm(Tensor self, Tensor mat1, Tensor mat2, *, Scalar beta=1, Scalar alpha=1) -> Tensor""" | |||
|
|
|||
| # TODO(titaiwang): op.Gemm seems needed to take care of corner case according to old symbolic_fn. | |||
| # Currently, it shows op-level validation failing on bloom. | |||
|
|
|||
There was a problem hiding this comment.
Remove line 110 down below as well?
…nnx-script into titaiwang/add_unflatten
Fixes #599
xfail graph test to wait microsoft/onnxruntime#15409