Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/22175
- Rename AliasAnalysisKind::DEFAULT to AliasAnalysisKind::CONSERVATIVE
- Introduce AliasAnalysisKind::FROM_SCHEMA that means the alias annotations of the schema should be honored
- Introduce AliasAnalysisKind::INTERNAL_SPECIAL_CASE to be able to run assertions that internal special cased ops are treated correctly
- aten:: and prim:: ops are not treated as special cases anymore, but just use AliasAnalysisKind::FROM_SCHEMA
- There's a set of assertions to ensure that aten:: and prim:: ops are all correctly set up to use AliasAnalysisKind::FROM_SCHEMA. Once this PR lands and passes all tests, we will remove those assertions and open up for the possibility of different AliasAnalysisKind settings for aten:: and prim:: ops
Differential Revision: D15929595
fbshipit-source-id: 7c6a9d4d29e13b8c9a856062cd6fb3f8a46a2e0d
Summary:
As part of the Variable/Tensor merge work: https://github.com/pytorch/pytorch/issues/13638, we make the following changes in this PR:
1. Remove the `Variable::Impl` class and the `DifferentiableViewImpl` class
2. Change all `Variable.data()` call sites to either use `Variable` directly, or use `Variable.tensor_data()`
3. Remove `Variable.data()` API
3. Add `Variable.variable_data()` that matches `tensor.data` in Python API, which creates a new `Variable` that shares the same storage and tensor metadata with the original `Variable`, but with a completely new autograd history.
After this PR, Variable doesn't wrap a Tensor internally anymore, and both Variable and Tensor use the same TensorImpl class as its `impl_`. The only difference is that Variable always has AutogradMeta in its TensorImpl, but Tensor doesn't.
**Note that this PR is BC-breaking in the following use cases:**
**Use Case 1:**
Previously, `x.data = y` works even if `x` and `y` are of different TensorImpl type (e.g. `x` is a CPU dense tensor whose impl is of type TensorImpl, while `y` is a CPU sparse tensor whose impl is of type SparseTensorImpl). However, after this PR, `x.data = y` doesn't work anymore if `x` and `y` are of different TensorImpl type, because the underlying implementation `variable.set_data(tensor)` no longer works if `variable` and `tensor` have different TensorImpl type.
**Use Case 2:**
If a tensor `x`'s `grad` is sparse, accumulating dense gradients to `x` will change the tensor that `x.grad` is pointing to. This is better illustrated with the following example:
```python
params = torch.tensor([1.5, 1.5]).requires_grad_()
with torch.no_grad():
# Change gradient to a sparse tensor
params.grad = torch.sparse_coo_tensor(torch.tensor([[1, 1]]).long(), torch.tensor([1., 1.]))
grad_saved = params.grad
params.backward(torch.tensor([1.5, 1.5]))
assert id(grad_saved) == id(params.grad) # This will fail after this PR
```
The assertion in the last line will fail after this PR, because adding dense gradients to sparse gradients will change the `params.grad` tensor reference.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17072
Differential Revision: D14075257
Pulled By: yf225
fbshipit-source-id: 0e681df641270dea586042dd26db59f2e76b5957
Summary:
This patch specializes `Optional[Tensor]` graph inputs to either a `DimensionedTensorType` (if a Tensor is passed) or `NoneType`. Other `Optional[T]` are specialized to `T` or `None`.
- For unwrapping (checked and unchecked) we need to keep the output type, as IR code that follows unwrapping may not work with NoneType (just as it doesn't deal with Optional). While it would not be hit during execution, it will run against the (legitimate) assumptions of the analysis passes.
- Function lookup currently will not match NoneType when it expects optional (I'm not entirely sure why this doesn't lead to unhappyness currently, but hey), I amend this at the level of the function matching code (`operator.cpp`), but see Adam's comments. We would run into trouble if we needed to select between functions whose signature only differs in Optional types with different subtypes, but we would have the same problem when calling them directly, so I would think this is OK.
- It would enable throwing away branches we can't hit. This also reduces the "blockyness" of the graph, so it may be easier to apply optimizations (e.g. fuse things in `if t is None: ...` and outside the `if`.
- Arguments passed into `Optional[Tensor]` arguments will get shape information, which is very handy.
- It get's rid of the problem that tensors passed into Optional arguments get requires_grad set erroneously #18270 (though that also affects lists, which aren't fixed here).
- `Optional[List[int]]` is needed for #18697.
- We're changing typing in a more subtle way than the `TensorType`->`DimensionedTensorType`.
- In particular, specializing to NoneType loses the Type information captured in the `OptionalType` element type.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18407
Reviewed By: zdevito
Differential Revision: D15216808
Pulled By: eellison
fbshipit-source-id: 01f1a7643deaf4962c3f55eff2070d54b0e54b69
Summary:
Tested locally this fix#19039, did not add a test since there's no way to create a script module in the cpp world.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/19700
Differential Revision: D15094195
Pulled By: wanchaol
fbshipit-source-id: fcc2c1e5efbc160d976ae485ba2457442f62f065
Summary:
Add input information into generated RecordFunction calls in
VariableType wrappers, JIT operators and a few more locations
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18717
Differential Revision: D14729156
Pulled By: ilia-cher
fbshipit-source-id: 811ac4cbfd85af5c389ef030a7e82ef454afadec
Summary:
This PR propagates where we use first-class modules objects into the compiler. This creates a transitionary state where:
* compiler.cpp creates Graphs where `self` is a Module class and attributes/parameters/buffers/submodules are looked up with `prim::GetAttr`
* GraphExecutor still runs "lowered graphs" where the self object has been removed by a compiler pass `lower_first_class_method`.
* Tracing still creates "lowered graphs", and a pass "lift_lowered_method" creates a first-class method graph for things.
* This PR separates out Method and Function. A script::Function is a pure Graph with no `self` bound. Similar to Python, a script::Method is just a bound `self` and its underlying `script::Function`.
* This PR also separates CompilationUnit from Module. A CompilationUnit is just a list of named script::Functions. Class's have a CompilationUnit holding the class methods, and Modules also have a CompilationUnit holding their Methods. This avoids the weird circular case Module --has a-> Class -> has a -> Module ...
Details:
* In this transitionary state, we maintain two copies of a Graph, first-class module and lowered. Th first-class one has a self argument that is the module's class type. The lowered one is the lowered graph that uses the initial_ivalues inputs.
* When defining lowered methods using `_defined_lowered` we immediately create the first-class equivalent. The reverse is done lazily, creating lowered_methods on demand from the class.
* The two way conversions will be deleted in a future PR when the executor itself runs first-class objects. However this requires more changes to (1) the traces, (2) the python bindings, and (3) the onnx export pass and would make this PR way to large.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/19167
Differential Revision: D14891966
Pulled By: zdevito
fbshipit-source-id: 0b5f03118aa65448a15c7a7818e64089ec93d7ea
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18711
ghimport-source-id: c9caedc0660b2b7ba3730cd0e1a2e0e9c3cf422b
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18711 [jit] fix side-effects and aliasing for custom ops**
Previously we didn't track aliasing, mutation, or side effects for
custom ops. This PR adds in guards with the most conservative
assumptions possible: the op will
1) have side effects,
2) write to everything
3) produce a wildcard.
In order to tell whether a given operator is a custom op, this PR introduces
the concept of a "reserved" namespace (basically all our builtin namespaces).
Custom ops live in non-reserved namespaces, so a check on the namespace
is sufficient to tell whether a schema/node is "custom" or not.
This is just to get things correct for now. Follow-ups to this:
- Users should be able to specify aliasing/mutability without having to learn
the whole alias annotation schema.
- Relax assumptions a bit. In particular outputs can only alias input tensors,
they don't have to be wildcards.
Fixes#18490
Differential Revision: D14730978
fbshipit-source-id: 540b47a24ccf24145051609bdcc99c97e46e0fe0
Summary:
Further breakup test_misc.h. The remaining tests don't directly map to a jit file so I left them in test_misc.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18191
Differential Revision: D14533442
Pulled By: eellison
fbshipit-source-id: 7f538ce0aea208b6b55a4716dfcf039548305041
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18121
ghimport-source-id: 70c273bfbcb68f7b25cf87f5614c662960864758
Stack from [ghstack](https://github.com/ezyang/ghstack):
* **#18121 [jit] fix double free in test_jit**
These definitions used to be in anonymous namespace so they weren't exported from the translation unit. #18071 put those in a `test` namespace so I guess they were getting their destructors called twice on exit somehow. Making them static again fixes the problem.
Reviewed By: ezyang
Differential Revision: D14498349
fbshipit-source-id: f969781695dcbebdfcfce667fce5b986222a373e
Summary:
Breakup test_misc so that a test for a file is in test_filename. I think we might want to wait on moving test files into the source directory, since that would involve moving some tests over to the C10 folder, and this goes 99% of the way for test discoverability IMO anyway.
I added a file test_utils for common functions invoked in the tests.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/18071
Differential Revision: D14485787
Pulled By: eellison
fbshipit-source-id: dcb20d1978d490999d435ea20c1d0503413a5c80
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17585
Create a sugared value that represents a class during initialization. This is
so that assignments to attributes correctly define attributes in __init__ but
raise an error elsewhere.
Reviewed By: shannonzhu
Differential Revision: D14263403
fbshipit-source-id: 09b2feeb272302f00a79c2a0302fbdf5483aed6a
Summary:
Last batch of IR expect files removed. Includes some removal of expect files that are no longer used.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17886
Differential Revision: D14414435
Pulled By: eellison
fbshipit-source-id: 0bfd7ce66ac2f72a57f15f45ebd60b95e80b6c16
Summary:
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17480
This was always part of our "spec" but not implemented
Reviewed By: houseroad
Differential Revision: D14214301
fbshipit-source-id: 118db320b43ec099dc3e730c67d39487474c23ea
Summary:
This PR removes a few size of `self` that passed from forward pass to backward pass when `self` is already required in backward pass. This could be reason that cause the potential slow down in #16689 . I will attach a few perf numbers (still a bit volatile among runs tho) I got in the comment.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17187
Differential Revision: D14179512
Pulled By: ailzhang
fbshipit-source-id: 5f3b1f6f26a3fef6dec15623b940380cc13656fa
Summary:
Trying to land again, make prim::None into a case of prim::Constant. Reverted the previous landing because it broke an important onnx export test.
https://github.com/pytorch/pytorch/pull/16160
Pull Request resolved: https://github.com/pytorch/pytorch/pull/17186
Differential Revision: D14115304
Pulled By: eellison
fbshipit-source-id: 161435fc30460b4e116cdd62c7b2e5b94581dcb7
Summary:
This change simplifies analysis done on constants since prim::None does not need to be handled separately now. To check if a constant node is None, use node->isNone().
Next step will be to remove prim::Undefined.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/16160
Differential Revision: D14109636
Pulled By: eellison
fbshipit-source-id: d26fd383976163a2ddd4c24984bd672a541cc876
Summary:
- Moved a few functions from `autograd` namespace to `aten` namespace to be visible from JIT nativeResolver.
- Added a hack to loop up keyword only argument. Will add proper support for kw only later
- Simulate function overload in aten using `_<number>` as function name suffix.
- Even `forward` returns multiple outputs like in `kthvalue`, there's at most one requires grad that we currently support.
- Removed the `TensorList` related ops here since partial `TensorList` support is prone to bugs. Our symbolic diff for `cat` was never tested with autodiff, and it seems broken. Need to find another proper way to support these ops(either by properly supporting `TensorList` or sth like `prim::ConstantChunk` and leave them for next PR.
Ops supported in this PR:
```
erf
expand_as
index
kthvalue
mean
permute
pow
rsub
select
sqrt
squeeze
t
to
topk
transpose
view
var
embedding
logsumexp
// grad is None
_dim_arange
contiguous
nonzero
ones_like
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/16689
Differential Revision: D14020806
Pulled By: ailzhang
fbshipit-source-id: a5e2c144a7be5a0d39d7ac5f93cb402ec12503a5
