pytorch/torch/_export
Avik Chaudhuri bceedeec2b fix checking non-trivial input constraints (#143442)
A bunch of auto dynamic shape tests would fail non-strict retraceability because when checking input constraints, we'd compare non-trivial expressions, which would require / affect shape env.
```
... is not tracked with proxy for <torch.fx.experimental.proxy_tensor._ModuleStackTracer object ...
```

I've also observed this bug internally.

This PR does an early check on whether args passed have concrete shapes, and only then proceeds: as before, we
1. try to unify / solve with the arg dim when the corresponding placeholder node dim is symbolic in one symbol
2. check directly if the placeholder node dim is concrete
3. otherwise defer to run time.

Differential Revision: [D67359596](https://our.internmc.facebook.com/intern/diff/D67359596/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/143442
Approved by: https://github.com/tugsbayasgalan
2024-12-18 07:29:08 +00:00
..
db Remove unused Python variables in torch/[_-a]* (#133492) 2024-12-12 17:39:14 +00:00
pass_infra
passes remove allow-untyped-defs for torch/_export/passes/remove_runtime_assertions.py (#143435) 2024-12-18 03:05:20 +00:00
serde Add float8 support in serde schema (#143343) 2024-12-18 05:07:21 +00:00
__init__.py Revert "Kill capture_pre_autograd_graph API (#143224)" 2024-12-17 00:47:24 +00:00
converter.py Remove unused Python variables in torch/[_-a]* (#133492) 2024-12-12 17:39:14 +00:00
error.py
non_strict_utils.py support slicing with symints in non-strict (#143217) 2024-12-14 10:27:45 +00:00
pass_base.py
tools.py
utils.py fix checking non-trivial input constraints (#143442) 2024-12-18 07:29:08 +00:00
verifier.py Add truediv support in export serializer (#136364) 2024-12-05 17:33:33 +00:00
wrappers.py