[PT FE]: Add support for prim::data operation #32167
Conversation
|
hello @mvafin @PiotrKrzem @p-wysocki @mlukasze @mitruska ,I have submitted this pull request to Add support for prim::data operation.please let me know your feedback on my PR , Thanks. Best Regards |
|
build_jenkins |
| {"prim::TupleIndex", op::translate_tuple_index}, | ||
| // prim::TupleUnpack - Supported in limited set of patterns | ||
| {"prim::type", op::skip_node}, // Used with prim::device, pass PtFrameworkNode. | ||
| {"prim::data", op::translate_data}, |
There was a problem hiding this comment.
I think it can be op::skip_node. Doesn't think you need any special handling for complex types in this case.
| tensor = torch.from_numpy(data).to(self.dtype) | ||
| return (tensor.numpy(),) | ||
|
|
||
| @pytest.mark.parametrize("dtype", [torch.float32, torch.float64, torch.int32, torch.int64]) |
There was a problem hiding this comment.
Add test case for complex numbers too
… complex tests xfail
|
hey @mvafin ,thanks for the feedback.I have addressed all of your suggestions and added test cases for complex-types,changed to op::skip_node.Please let me know if any other changes are needed. |
|
build_jenkins |
|
Hi @mlukasze @mvafin @PiotrKrzem @p-wysocki , just following up to see if you’ve had a chance to look at this PR. Let me know if any changes are needed or if there are questions I can clarify. Thanks for your time! |
this solves issue #29721
Details:
-added support for prim::data
Tickets: