Nikita Shulga
b56b002842
Fix NULL dereference in binary CPU ops ( #115183 )
...
Targeted fix for https://github.com/pytorch/pytorch/issues/113037
A more fundamental one, where those functions are not even called for
empty tensors are coming later
Pull Request resolved: https://github.com/pytorch/pytorch/pull/115183
Approved by: https://github.com/drisspg , https://github.com/atalman , https://github.com/huydhn
2023-12-06 03:37:47 +00:00
Isuru Fernando
2f536ff92c
Refactor values kwarg in foreach tests ( #112781 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112781
Approved by: https://github.com/lezcano
ghstack dependencies: #112778
2023-11-22 22:10:54 +00:00
Isuru Fernando
1f1ff629a8
Use parent class attribute supports_out for foreach_zero opinfo ( #112778 )
...
Instead of introducing a new has_no_out_of_place attribute
Also fixes foreach_copy tests
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112778
Approved by: https://github.com/lezcano
2023-11-22 18:00:44 +00:00
Jane Xu
deec2380c7
Add 0dim Tensor overload for _foreach_div ( #113688 )
...
This PR is ALMOST basically just following the steps from #106677 EXCEPT! We do add one feature. Similar to fused_adam(w), for the CUDA dispatches: when the scalar tensor is on CPU, we .item and redispatch to the normal scalar overload. Otherwise, the cuda kernel will complain about mismatch in devices between the scalar and the tensors.
Why do we add this feature? Our optimizers want to allow lr as a tensor, and lr could be a CPU tensor. lr is used with foreach_div_ in Adam, so our CI will break otherwise.
After this PR, `_foreach_mul` and `_foreach_div` will accept either a CPU or a GPU tensor for the scalar tensor (vs only a GPU tensor). They join the ranks of `fused_adam(w)` in this characteristic. I did not yet do the same thing for foreach_add (the only other foreach op with a .Tensor overload) because there is no use case and will be more involved.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113688
Approved by: https://github.com/mlazos , https://github.com/albanD
2023-11-15 20:59:32 +00:00
rraminen
44367c59b2
Update skip reason for failing unit tests on ROCm 5.7 ( #113286 )
...
Follow up to https://github.com/pytorch/pytorch/pull/110465 . Updated skip reason for failing unit tests on ROCm 5.7
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113286
Approved by: https://github.com/malfet
2023-11-13 19:29:04 +00:00
Isuru Fernando
3b915f9de0
[pt2] enable meta tests for foreach ops ( #113484 )
...
Try https://github.com/pytorch/pytorch/pull/113059 again.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113484
Approved by: https://github.com/lezcano
2023-11-11 02:43:41 +00:00
Nikita Shulga
d5eb9f725c
Fix test_add_scalar_with_empty_list_tensor ( #113262 )
...
By actually instantiating test method to a different types and devices rather than always creating it on CPU.
Also, remove `bool` from the list, as adding 1 to bool is not supported.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113262
Approved by: https://github.com/jeanschmidt , https://github.com/atalman , https://github.com/lezcano
2023-11-08 20:56:37 +00:00
rraminen
3a429423fc
Upgrade CI to ROCm5.7 ( #110465 )
...
This PR is to upgrade CI to ROCm5.7
Pull Request resolved: https://github.com/pytorch/pytorch/pull/110465
Approved by: https://github.com/pruthvistony , https://github.com/malfet
2023-11-08 06:11:10 +00:00
Nikita Shulga
236eff9531
[BE] Refactor repeated assets in test_foreach.py ( #112348 )
...
Tested conditions in `test_binary_op_list_error_cases` looks almost identical, although it tests method and in-place variants. Use for loop to make distinction a bit more explicit
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112348
Approved by: https://github.com/albanD
ghstack dependencies: #112349
2023-10-31 01:11:44 +00:00
Nikita Shulga
80de49653a
Prevent OOB access in foreach_list variants ( #112349 )
...
By checking that lists sizes are the same before computing forward gradients.
Before the change
```cpp
::std::vector<at::Tensor> _foreach_add_List(c10::DispatchKeySet ks, at::TensorList self, at::TensorList other, const at::Scalar & alpha) {
auto self_ = unpack(self, "self", 0);
auto other_ = unpack(other, "other", 1);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self, other );
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]) || isFwGradDefined(other[i]);
}
...
```
after the change:
```cpp
::std::vector<at::Tensor> _foreach_add_List(c10::DispatchKeySet ks, at::TensorList self, at::TensorList other, const at::Scalar & alpha) {
auto self_ = unpack(self, "self", 0);
auto other_ = unpack(other, "other", 1);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self, other );
TORCH_CHECK(
self.size() == other.size(),
"Tensor lists must have the same number of tensors, got ",
self.size(),
" and ",
other.size());
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]) || isFwGradDefined(other[i]);
}
```
Add regression test
Fixes https://github.com/pytorch/pytorch/issues/112305
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112349
Approved by: https://github.com/Chillee
2023-10-30 20:43:03 +00:00
Isuru Fernando
c7dcba9276
Remove passing disable_fastpath in kwargs ( #112250 )
...
Fixes an issue that came up in https://github.com/pytorch/pytorch/pull/112030
Pull Request resolved: https://github.com/pytorch/pytorch/pull/112250
Approved by: https://github.com/lezcano
2023-10-27 18:29:20 +00:00
Jane Xu
ca7d084ff9
Add ScalarTensor or 0dim overload for _foreach_add ( #111079 )
...
Adding a Tensor overload will allow us to:
- optimize in more cases than before
- increase coverage for scalarTensor instead of just scalars in our foreach APIs
The main complication in this PR was that add.Tensor has a scalar overload, so I've now built out support for that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/111079
Approved by: https://github.com/albanD
2023-10-20 01:34:07 +00:00
Jane Xu
0a60219fe3
[foreach] Fix 0-size handling for real for real ( #109402 )
...
@crcrpar's last attempt to fix the 0-size problem unfortunately did not pass all cases. See my comment in https://github.com/pytorch/pytorch/issues/100701 . When we have a tail tensor of size 0, the old code would mess with the chunk logic to check the previous tensor's length. This is flawed because:
1. if the previous tensor was also 0 sized, (so a tensor list of [tensor, tensor, tensor, ..., 0-sized tensor, 0-sized tensor],) chunks would still be 0 and the nested for loop would be missed.
2. the nested forloop pronounces side effects on tensorListMeta that _shouldn't_ be there! This can mess up the compute in unexpected ways that I haven't really needed to reason through.
We noticed that the problem had not been fixed due to an internal report. This PR solves the issue by:
- removing the finagling of chunks when the tail tensor is 0-sized
- adding a surefire way for the kernel to be launched in the case where the last tensor is 0-sized AND there's content in the metadata, signifying there is stuff to compute still.
## test plan
As I went through the code, I also added some comments explaining what's up and modified our tensor inputs to ensure that this case is tested in the test_parity test in test_foreach.py. Yes, I do realize there is quite a bit of duplication and that this file could be due for a refactor. That said, the primary goal of this PR is to fix the pretty egregious bug and refactoring can be a followup.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109402
Approved by: https://github.com/albanD
2023-09-26 17:38:20 +00:00
Jane Xu
4b0281b32c
[BE][foreach] name tests correctly. noncontiguous inputs != fastpath ( #109771 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/109771
Approved by: https://github.com/soulitzer
2023-09-22 19:16:14 +00:00
Masaki Kozuki
602413a0a0
Refactor test_foreach.py ( #107869 )
...
## Summary
- Change the default of `supports_autograd` and `supports_forward_ad` of `ForeachFuncInfo` to `True`
- Add `test_zero_size_tensor_inputs` to make sure that foreach functions can handle 0-size Tensor inputs
- Add `test_parity` to check the consistency between outputs of foreach and for-loop of native function.
- Add `test_autodiff` to check forward-mode and reverse-mode AD
- Keep the corner cases that are not covered by the newly introduced methods
rel:
- #58833
Pull Request resolved: https://github.com/pytorch/pytorch/pull/107869
Approved by: https://github.com/janeyx99
2023-09-14 19:39:26 +00:00
Masaki Kozuki
5814380e7b
Revert "Revert "Reland "Add forward mode AD to out-place foreach functions ( #102409 ) ( #106043 )""" ( #106320 )
...
Fixed a typo specifying the number of tensors and elements in the test having failed in slow gradcheck
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106320
Approved by: https://github.com/soulitzer
2023-08-18 23:01:42 +00:00
Masaki Kozuki
b234b94760
Add in-place _foreach_copy ( #107226 )
...
Fixes #107162
Pull Request resolved: https://github.com/pytorch/pytorch/pull/107226
Approved by: https://github.com/janeyx99
2023-08-17 00:11:18 +00:00
PyTorch MergeBot
354484ea6d
Revert "Add _foreach_clamp ( #106574 )"
...
This reverts commit 2b560d3c3ahttps://github.com/pytorch/pytorch/pull/106574 on behalf of https://github.com/kit1980 due to breaking internal windows builds ([comment](https://github.com/pytorch/pytorch/pull/106574#issuecomment-1675400335 ))
2023-08-11 21:05:04 +00:00
Masaki Kozuki
2b560d3c3a
Add _foreach_clamp ( #106574 )
...
Rel:
- #106221
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106574
Approved by: https://github.com/janeyx99
2023-08-10 05:26:09 +00:00
Masaki Kozuki
9e4e0ecdd9
Add 0-dim Tensor overload to _foreach_mul ( #106677 )
...
rel:
- https://github.com/pytorch/pytorch/issues/106427
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106677
Approved by: https://github.com/janeyx99
2023-08-08 03:00:01 +00:00
PyTorch MergeBot
2b427ae3a7
Revert "Reland "Add forward mode AD to out-place foreach functions ( #102409 ) ( #106043 )"
...
This reverts commit e773f28ee3https://github.com/pytorch/pytorch/pull/106043 on behalf of https://github.com/DanilBaibak due to Break slow tests ([comment](https://github.com/pytorch/pytorch/pull/106043#issuecomment-1658642734 ))
2023-07-31 15:50:36 +00:00
Masaki Kozuki
e773f28ee3
Reland "Add forward mode AD to out-place foreach functions ( #102409 ) ( #106043 )
...
forward-mode AD of out-of-place foreach functions, finally.
rel:
- #102409
- #105504
- #58833
- #100695
---
# Generated Foreach
```c++
::std::vector<at::Tensor> _foreach_sinh(c10::DispatchKeySet ks, at::TensorList self) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]);
}
std::shared_ptr<ForeachSinhBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<ForeachSinhBackward0>(new ForeachSinhBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->self_ = make_saved_variable_list(self);
grad_fn->self_size_ = self.size();
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::_foreach_sinh(ks & c10::after_autograd_keyset, self_);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
std::vector<c10::optional<at::Tensor>> result_new_fw_grad_opts(self.size(), c10::nullopt);
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
if (_any_has_forward_grad_result[i]) {
auto self_t_raw = toNonOptFwGrad(self[i]);
auto self_tensor = toNonOptTensor(self[i]);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self[i]);
result_new_fw_grad_opts[i] = (self_t.conj() * self_p.cosh().conj()).conj();
}
}
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
auto& result_new_fw_grad_opt = result_new_fw_grad_opts[i];
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result[i].defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result[i]._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
}
return result;
}
::std::vector<at::Tensor> _foreach_norm_Scalar(c10::DispatchKeySet ks, at::TensorList self, const at::Scalar & ord) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]);
}
std::shared_ptr<ForeachNormBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<ForeachNormBackward0>(new ForeachNormBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->ord = ord;
grad_fn->self_ = make_saved_variable_list(self);
grad_fn->self_size_ = self.size();
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::_foreach_norm(ks & c10::after_autograd_keyset, self_, ord);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
std::vector<c10::optional<at::Tensor>> result_new_fw_grad_opts(self.size(), c10::nullopt);
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
if (_any_has_forward_grad_result[i]) {
auto self_t_raw = toNonOptFwGrad(self[i]);
auto self_tensor = toNonOptTensor(self[i]);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self[i]);
result_new_fw_grad_opts[i] = norm_jvp(self_p, self_t, ord, result[i]);
}
}
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
auto& result_new_fw_grad_opt = result_new_fw_grad_opts[i];
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result[i].defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result[i]._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
}
if (grad_fn) {
grad_fn->result = result;
}
return result;
}
```
# Reference
```c++
at::Tensor sinh(c10::DispatchKeySet ks, const at::Tensor & self) {
auto& self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
[[maybe_unused]] auto _any_has_forward_grad_result = (isFwGradDefined(self));
std::shared_ptr<SinhBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<SinhBackward0>(new SinhBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->self_ = SavedVariable(self, false);
}
#ifndef NDEBUG
c10::optional<Storage> self__storage_saved =
self_.has_storage() ? c10::optional<Storage>(self_.storage()) : c10::nullopt;
c10::intrusive_ptr<TensorImpl> self__impl_saved;
if (self_.defined()) self__impl_saved = self_.getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::sinh(ks & c10::after_autograd_keyset, self_);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
if (self__storage_saved.has_value() &&
!at::impl::dispatch_mode_enabled() &&
!at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved.value().is_alias_of(self_.storage()));
if (self__impl_saved && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved == self_.getIntrusivePtr());
if (result.has_storage() && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result)) {
TORCH_INTERNAL_ASSERT(result.storage().use_count() == 1, "function: sinh");
}
if (!at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result))
TORCH_INTERNAL_ASSERT(result.use_count() <= 1, "function: sinh");
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
c10::optional<at::Tensor> result_new_fw_grad_opt = c10::nullopt;
if (_any_has_forward_grad_result && (result.defined())) {
auto self_t_raw = toNonOptFwGrad(self);
auto self_tensor = toNonOptTensor(self);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self);
result_new_fw_grad_opt = (self_t.conj() * self_p.cosh().conj()).conj();
}
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result.defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
return result;
}
at::Tensor norm_Scalar(c10::DispatchKeySet ks, const at::Tensor & self, const at::Scalar & p) {
auto& self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
[[maybe_unused]] auto _any_has_forward_grad_result = (isFwGradDefined(self));
std::shared_ptr<NormBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<NormBackward0>(new NormBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->p = p;
grad_fn->self_ = SavedVariable(self, false);
}
#ifndef NDEBUG
c10::optional<Storage> self__storage_saved =
self_.has_storage() ? c10::optional<Storage>(self_.storage()) : c10::nullopt;
c10::intrusive_ptr<TensorImpl> self__impl_saved;
if (self_.defined()) self__impl_saved = self_.getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::norm(ks & c10::after_autograd_keyset, self_, p);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
if (self__storage_saved.has_value() &&
!at::impl::dispatch_mode_enabled() &&
!at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved.value().is_alias_of(self_.storage()));
if (self__impl_saved && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved == self_.getIntrusivePtr());
if (result.has_storage() && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result)) {
TORCH_INTERNAL_ASSERT(result.storage().use_count() == 1, "function: norm_Scalar");
}
if (!at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result))
TORCH_INTERNAL_ASSERT(result.use_count() <= 1, "function: norm_Scalar");
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
throw_error_for_complex_autograd(result, "norm");
c10::optional<at::Tensor> result_new_fw_grad_opt = c10::nullopt;
if (_any_has_forward_grad_result && (result.defined())) {
auto self_t_raw = toNonOptFwGrad(self);
auto self_tensor = toNonOptTensor(self);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self);
result_new_fw_grad_opt = norm_jvp(self_p, self_t, p, result);
}
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result.defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
if (grad_fn) {
grad_fn->result_ = SavedVariable(result, true);
}
return result;
}
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/106043
Approved by: https://github.com/soulitzer
2023-07-27 03:13:24 +00:00
Masaki Kozuki
72f2c87a5a
[foreach] Set SavedVariable.is_output to true for grad_fn->result_ ( #105504 )
...
fixes #105502
The scope of this pull request is out-of-place foreach functions that depend on their output tensorlist for backward such as `_foreach_exp`. An example of the generated code with this update is as follows:
```c++
variable_list ForeachExpBackward0::apply(variable_list&& grads) {
std::lock_guard<std::mutex> lock(mutex_);
TORCH_CHECK(!result_released_, ERR_BACKWARD_TWICE);
IndexRangeGenerator gen;
auto self_ix = gen.range(self_size_);
variable_list grad_inputs(gen.size());
auto result = unpack_list(result_, shared_from_this());
if (task_should_compute_output({ self_ix })) {
std::vector<Tensor> grad_result;
grad_result.reserve(grads.size());
for (const auto & i : c10::irange(grads.size())) {
if (grads[i].defined()) {
grad_result.emplace_back(grads[i] * result[i].conj());
} else {
grad_result.emplace_back(Tensor());
}
}
copy_range(grad_inputs, self_ix, grad_result);
}
return grad_inputs;
}
::std::vector<at::Tensor> _foreach_exp(c10::DispatchKeySet ks, at::TensorList self) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::shared_ptr<ForeachExpBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<ForeachExpBackward0>(new ForeachExpBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->self_size_ = self.size();
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
auto _tmp = ([&]() {
if ((isFwGradDefinedTensorList(self))) {
static c10::OperatorName full_name("aten::_foreach_exp", "");
static c10::optional<c10::OperatorHandle> opt_op = c10::Dispatcher::singleton().findSchema(full_name);
return impl::run_jit_decomposition_with_args_for_jvp<::std::vector<at::Tensor>>("_foreach_exp", *opt_op, ks, self);
} else {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::_foreach_exp(ks & c10::after_autograd_keyset, self_);
}
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
if (grad_fn) {
grad_fn->result_ = make_saved_variable_list(result, true);
}
return result;
}
```
A bit of context:
- https://github.com/pytorch/pytorch/pull/105368#issuecomment-1640912479
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105504
Approved by: https://github.com/soulitzer
2023-07-26 14:29:32 +00:00
Jane Xu
803d42e457
add lerp cpu support for half ( #105607 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105607
Approved by: https://github.com/albanD
2023-07-21 20:29:05 +00:00
Justin Chu
73e1455327
[BE] Enable ruff's UP rules and autoformat test/ ( #105434 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/105434
Approved by: https://github.com/albanD
2023-07-19 20:36:06 +00:00
PyTorch MergeBot
8958f041be
Revert "Add forward mode AD to out-place foreach functions ( #102409 )"
...
This reverts commit e2ec0ba404https://github.com/pytorch/pytorch/pull/102409 on behalf of https://github.com/huydhn due to Sorry for reverting your PR but it is failing some tests in trunk e799f565ebhttps://github.com/pytorch/pytorch/pull/102409#issuecomment-1615254393 ))
2023-06-30 22:46:57 +00:00
Masaki Kozuki
e2ec0ba404
Add forward mode AD to out-place foreach functions ( #102409 )
...
The major difference from in-place support is that some out-place functions have their derivatives spelled out in derivatives.yaml, which requires some changes in `load_derivatives.py` and some handlings in various places due to the others whose derivatives are generated by `torchgen`.
rel:
- #58833
- #100695
---
# Generated Foreach
```c++
::std::vector<at::Tensor> _foreach_sinh(c10::DispatchKeySet ks, at::TensorList self) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]);
}
std::shared_ptr<ForeachSinhBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<ForeachSinhBackward0>(new ForeachSinhBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->self_ = make_saved_variable_list(self);
grad_fn->self_size_ = self.size();
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::_foreach_sinh(ks & c10::after_autograd_keyset, self_);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
std::vector<c10::optional<at::Tensor>> result_new_fw_grad_opts(self.size(), c10::nullopt);
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
if (_any_has_forward_grad_result[i]) {
auto self_t_raw = toNonOptFwGrad(self[i]);
auto self_tensor = toNonOptTensor(self[i]);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self[i]);
result_new_fw_grad_opts[i] = (self_t.conj() * self_p.cosh().conj()).conj();
}
}
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
auto& result_new_fw_grad_opt = result_new_fw_grad_opts[i];
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result[i].defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result[i]._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
}
return result;
}
::std::vector<at::Tensor> _foreach_norm_Scalar(c10::DispatchKeySet ks, at::TensorList self, const at::Scalar & ord) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::vector<bool> _any_has_forward_grad_result(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_result[i] = isFwGradDefined(self[i]);
}
std::shared_ptr<ForeachNormBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<ForeachNormBackward0>(new ForeachNormBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->ord = ord;
grad_fn->self_ = make_saved_variable_list(self);
grad_fn->self_size_ = self.size();
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::_foreach_norm(ks & c10::after_autograd_keyset, self_, ord);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
std::vector<c10::optional<at::Tensor>> result_new_fw_grad_opts(self.size(), c10::nullopt);
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
if (_any_has_forward_grad_result[i]) {
auto self_t_raw = toNonOptFwGrad(self[i]);
auto self_tensor = toNonOptTensor(self[i]);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self[i]);
result_new_fw_grad_opts[i] = norm_jvp(self_p, self_t, ord, result[i]);
}
}
for (const auto& i : c10::irange(result_new_fw_grad_opts.size())) {
auto& result_new_fw_grad_opt = result_new_fw_grad_opts[i];
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result[i].defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result[i]._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
}
if (grad_fn) {
grad_fn->result = result;
}
return result;
}
```
# Reference
```c++
at::Tensor sinh(c10::DispatchKeySet ks, const at::Tensor & self) {
auto& self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
[[maybe_unused]] auto _any_has_forward_grad_result = (isFwGradDefined(self));
std::shared_ptr<SinhBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<SinhBackward0>(new SinhBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->self_ = SavedVariable(self, false);
}
#ifndef NDEBUG
c10::optional<Storage> self__storage_saved =
self_.has_storage() ? c10::optional<Storage>(self_.storage()) : c10::nullopt;
c10::intrusive_ptr<TensorImpl> self__impl_saved;
if (self_.defined()) self__impl_saved = self_.getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::sinh(ks & c10::after_autograd_keyset, self_);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
if (self__storage_saved.has_value() &&
!at::impl::dispatch_mode_enabled() &&
!at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved.value().is_alias_of(self_.storage()));
if (self__impl_saved && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved == self_.getIntrusivePtr());
if (result.has_storage() && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result)) {
TORCH_INTERNAL_ASSERT(result.storage().use_count() == 1, "function: sinh");
}
if (!at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result))
TORCH_INTERNAL_ASSERT(result.use_count() <= 1, "function: sinh");
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
c10::optional<at::Tensor> result_new_fw_grad_opt = c10::nullopt;
if (_any_has_forward_grad_result && (result.defined())) {
auto self_t_raw = toNonOptFwGrad(self);
auto self_tensor = toNonOptTensor(self);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self);
result_new_fw_grad_opt = (self_t.conj() * self_p.cosh().conj()).conj();
}
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result.defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
return result;
}
at::Tensor norm_Scalar(c10::DispatchKeySet ks, const at::Tensor & self, const at::Scalar & p) {
auto& self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
[[maybe_unused]] auto _any_has_forward_grad_result = (isFwGradDefined(self));
std::shared_ptr<NormBackward0> grad_fn;
if (_any_requires_grad) {
grad_fn = std::shared_ptr<NormBackward0>(new NormBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self ));
grad_fn->p = p;
grad_fn->self_ = SavedVariable(self, false);
}
#ifndef NDEBUG
c10::optional<Storage> self__storage_saved =
self_.has_storage() ? c10::optional<Storage>(self_.storage()) : c10::nullopt;
c10::intrusive_ptr<TensorImpl> self__impl_saved;
if (self_.defined()) self__impl_saved = self_.getIntrusivePtr();
#endif
auto _tmp = ([&]() {
at::AutoDispatchBelowADInplaceOrView guard;
return at::redispatch::norm(ks & c10::after_autograd_keyset, self_, p);
})();
auto result = std::move(_tmp);
#ifndef NDEBUG
if (self__storage_saved.has_value() &&
!at::impl::dispatch_mode_enabled() &&
!at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved.value().is_alias_of(self_.storage()));
if (self__impl_saved && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved == self_.getIntrusivePtr());
if (result.has_storage() && !at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result)) {
TORCH_INTERNAL_ASSERT(result.storage().use_count() == 1, "function: norm_Scalar");
}
if (!at::impl::dispatch_mode_enabled() && !at::impl::tensor_has_dispatch(result))
TORCH_INTERNAL_ASSERT(result.use_count() <= 1, "function: norm_Scalar");
#endif
if (grad_fn) {
set_history(flatten_tensor_args( result ), grad_fn);
}
throw_error_for_complex_autograd(result, "norm");
c10::optional<at::Tensor> result_new_fw_grad_opt = c10::nullopt;
if (_any_has_forward_grad_result && (result.defined())) {
auto self_t_raw = toNonOptFwGrad(self);
auto self_tensor = toNonOptTensor(self);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::_efficientzerotensor(self_tensor.sizes(), self_tensor.options());
auto self_p = toNonOptPrimal(self);
result_new_fw_grad_opt = norm_jvp(self_p, self_t, p, result);
}
if (result_new_fw_grad_opt.has_value() && result_new_fw_grad_opt.value().defined() && result.defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
result._set_fw_grad(result_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ false);
}
if (grad_fn) {
grad_fn->result_ = SavedVariable(result, true);
}
return result;
}
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/102409
Approved by: https://github.com/soulitzer
2023-06-30 04:51:43 +00:00
Nikita Shulga
6d2887cc06
Reland "Move tensor grouping to ATen" ( #103912 )
...
This is a reland of https://github.com/pytorch/pytorch/pull/100007 with a build fix for Windows debug builds.
`at::native::ParamsHash` only works on structs with standard layout, but `std::string` isn't one in Visual C++ debug builds, which one can easily verified by running something like:
```cpp
#define _DEBUG
#include <type_traits>
#include <string>
static_assert(std::is_standard_layout_v<std::string>, "Oh noes");
```
If above conditon is not met, instead of printing a static_assert output, VC++ raises a very cryptic compilation errors, see https://github.com/pytorch/pytorch/pull/100007#discussion_r1227116292 for more detail.
Also, using `std::hash` for string should result in a faster hash function.
(cherry picked from commit 74b7a6c75e🤖 Generated by Copilot at 5914771</samp>
This pull request introduces a new function `_group_tensors_by_device_and_dtype` that can group tensors by their device and dtype, and updates the `foreach` utilities and several optimizers to use this function. The goal is to improve the performance, readability, and compatibility of the code that handles tensors with different properties. The pull request also adds a test case and type annotations for the new function, and some error checks for the `fused` argument in Adam and AdamW.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/103912
Approved by: https://github.com/janeyx99
2023-06-21 09:26:33 +00:00
PyTorch MergeBot
0cb5bc3b04
Revert "Move tensor grouping to ATen ( #100007 )"
...
This reverts commit 74b7a6c75ehttps://github.com/pytorch/pytorch/pull/100007 on behalf of https://github.com/izaitsevfb due to Breaks internal builds, see D46629727 ([comment](https://github.com/pytorch/pytorch/pull/100007#issuecomment-1587861598 ))
2023-06-12 18:30:33 +00:00
Masaki Kozuki
74b7a6c75e
Move tensor grouping to ATen ( #100007 )
...
rel: #94344
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100007
Approved by: https://github.com/janeyx99
2023-06-09 15:44:46 +00:00
Masaki Kozuki
0bb2b01541
Add forward mode AD to in-place foreach functions ( #100695 )
...
Awkwardly implement fwd AD by
- adding a few `CodeTemplate`s
- allowing for the cases where a variable is initialized with i-th element of TensorList
<!--
### TODOs:
- [x] ~~remove the first `_any_has_forward_grad_self`~~ make it a vector of bool
- [ ] clean up mapping of names from reference impl to foreach impl
- [x] add tests
-->
### Rel:
- #58833
- #96405
---
`_foreach_addcmul_.ScalarList` from `VariableType`
```c++
void _foreach_addcmul__ScalarList(c10::DispatchKeySet ks, at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars) {
auto self_ = unpack(self, "self", 0);
auto tensor1_ = unpack(tensor1, "tensor1", 1);
auto tensor2_ = unpack(tensor2, "tensor2", 2);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self, tensor1, tensor2 );
std::vector<bool> _any_has_forward_grad_self(self.size());
for (const auto& i : c10::irange(self.size())) {
_any_has_forward_grad_self[i] = isFwGradDefined(self[i]) || isFwGradDefined(tensor1[i]) || isFwGradDefined(tensor2[i]);
}
std::vector<c10::optional<at::Tensor>> original_selfs(self.size());
std::vector<std::shared_ptr<AddcmulBackward0>> grad_fns;
if (_any_requires_grad) {
for (const auto& i : c10::irange( self.size() )) {
const auto ith_requires_grad = compute_requires_grad(self[i], tensor1[i], tensor2[i]);
check_inplace(self[i], ith_requires_grad);
grad_fns.push_back([&]() -> std::shared_ptr<AddcmulBackward0> {
if (!ith_requires_grad) {
return nullptr;
} else {
auto grad_fn = std::shared_ptr<AddcmulBackward0>(new AddcmulBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self[i], tensor1[i], tensor2[i] ));
return grad_fn;
}
}());
}
if (!grad_fns.empty()) {
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
grad_fn->self_scalar_type = self[i].scalar_type();
grad_fn->tensor1_scalar_type = tensor1[i].scalar_type();
if (grad_fn->should_compute_output(1)) {
grad_fn->tensor2_ = SavedVariable(tensor2[i], false);
}
grad_fn->value = scalars[i];
if (grad_fn->should_compute_output(2)) {
grad_fn->tensor1_ = SavedVariable(tensor1[i], false);
}
grad_fn->tensor2_scalar_type = tensor2[i].scalar_type();
}
}
}
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
std::vector<c10::optional<Storage>> tensor1__storage_saved(tensor1_.size());
for (const Tensor& tensor : tensor1_)
tensor1__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> tensor1__impl_saved(tensor1_.size());
for (size_t i=0; i<tensor1_.size(); i++)
if (tensor1_[i].defined()) tensor1__impl_saved[i] = tensor1_[i].getIntrusivePtr();
std::vector<c10::optional<Storage>> tensor2__storage_saved(tensor2_.size());
for (const Tensor& tensor : tensor2_)
tensor2__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> tensor2__impl_saved(tensor2_.size());
for (size_t i=0; i<tensor2_.size(); i++)
if (tensor2_[i].defined()) tensor2__impl_saved[i] = tensor2_[i].getIntrusivePtr();
#endif
{
at::AutoDispatchBelowAutograd guard;
at::redispatch::_foreach_addcmul_(ks & c10::after_autograd_keyset, self_, tensor1_, tensor2_, scalars);
}
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
for (size_t i=0; i<tensor1_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor1__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(tensor1_))
TORCH_INTERNAL_ASSERT(tensor1__storage_saved[i].value().is_alias_of(tensor1_[i].storage()));
}
for (size_t i=0; i<tensor1_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor1__impl_saved[i] && !at::impl::tensorlist_has_dispatch(tensor1_))
TORCH_INTERNAL_ASSERT(tensor1__impl_saved[i] == tensor1_[i].getIntrusivePtr());
}
for (size_t i=0; i<tensor2_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor2__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(tensor2_))
TORCH_INTERNAL_ASSERT(tensor2__storage_saved[i].value().is_alias_of(tensor2_[i].storage()));
}
for (size_t i=0; i<tensor2_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor2__impl_saved[i] && !at::impl::tensorlist_has_dispatch(tensor2_))
TORCH_INTERNAL_ASSERT(tensor2__impl_saved[i] == tensor2_[i].getIntrusivePtr());
}
#endif
if (!grad_fns.empty()) {
auto differentiable_outputs = flatten_tensor_args( self );
TORCH_INTERNAL_ASSERT(differentiable_outputs.size() == grad_fns.size());
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
rebase_history(differentiable_outputs[i], grad_fns[i]);
}
}
}
std::vector<c10::optional<at::Tensor>> self_new_fw_grad_opts(self.size(), c10::nullopt);
for (const auto& i : c10::irange(self_new_fw_grad_opts.size())) {
if (_any_has_forward_grad_self[i]) {
auto self_t_raw = toNonOptFwGrad(self[i]);
auto self_tensor = toNonOptTensor(self[i]);
auto self_t = (self_t_raw.defined() || !self_tensor.defined())
? self_t_raw : at::zeros(self_tensor.sizes(), self_tensor.options());
auto tensor1_t_raw = toNonOptFwGrad(tensor1[i]);
auto tensor1_tensor = toNonOptTensor(tensor1[i]);
auto tensor1_t = (tensor1_t_raw.defined() || !tensor1_tensor.defined())
? tensor1_t_raw : at::_efficientzerotensor(tensor1_tensor.sizes(), tensor1_tensor.options());
auto tensor1_p = toNonOptPrimal(tensor1[i]);
auto tensor2_t_raw = toNonOptFwGrad(tensor2[i]);
auto tensor2_tensor = toNonOptTensor(tensor2[i]);
auto tensor2_t = (tensor2_t_raw.defined() || !tensor2_tensor.defined())
? tensor2_t_raw : at::_efficientzerotensor(tensor2_tensor.sizes(), tensor2_tensor.options());
auto tensor2_p = toNonOptPrimal(tensor2[i]);
self_t = GradMode::is_enabled() ? self_t.clone() : self_t;
self_new_fw_grad_opts[i] = self_t_raw.defined() ? self_t_raw.copy_(self_t + maybe_multiply(tensor1_t * tensor2_p, scalars[i]) + maybe_multiply(tensor2_t * tensor1_p, scalars[i])) : self_t + maybe_multiply(tensor1_t * tensor2_p, scalars[i]) + maybe_multiply(tensor2_t * tensor1_p, scalars[i]);
}
}
for (const auto& i : c10::irange(self_new_fw_grad_opts.size())) {
auto& self_new_fw_grad_opt = self_new_fw_grad_opts[i];
if (self_new_fw_grad_opt.has_value() && self_new_fw_grad_opt.value().defined() && self[i].defined()) {
// The hardcoded 0 here will need to be updated once we support multiple levels.
self[i]._set_fw_grad(self_new_fw_grad_opt.value(), /* level */ 0, /* is_inplace_op */ true);
}
}
}
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100695
Approved by: https://github.com/soulitzer
2023-05-25 15:39:48 +00:00
Masaki Kozuki
ba2bc7df8f
Enable backward on _foreach_zero_ ( #101149 )
...
Currently torchgen cannot find an appropriate `DifferentiabilityInfo` for `_foreach_zero_` because `gen_foreach_derivativeinfo` doesn't correctly make use of `functional_info_by_signature` and `differentiability_infos`, and `is_reference_for_foreach` a bit too strict to `_foreach_zero_`.
Generated code in `VariableType`
```c++
void _foreach_zero_(c10::DispatchKeySet ks, at::TensorList self) {
auto self_ = unpack(self, "self", 0);
[[maybe_unused]] auto _any_requires_grad = compute_requires_grad( self );
std::vector<c10::optional<at::Tensor>> original_selfs(self.size());
std::vector<std::shared_ptr<ZeroBackward0>> grad_fns;
if (_any_requires_grad) {
for (const auto& i : c10::irange( self.size() )) {
const auto ith_requires_grad = compute_requires_grad(self[i]);
check_inplace(self[i], ith_requires_grad);
grad_fns.push_back([&]() -> std::shared_ptr<ZeroBackward0> {
if (!ith_requires_grad) {
return nullptr;
} else {
auto grad_fn = std::shared_ptr<ZeroBackward0>(new ZeroBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self[i] ));
return grad_fn;
}
}());
}
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
{
at::AutoDispatchBelowAutograd guard;
at::redispatch::_foreach_zero_(ks & c10::after_autograd_keyset, self_);
}
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (!grad_fns.empty()) {
auto differentiable_outputs = flatten_tensor_args( self );
TORCH_INTERNAL_ASSERT(differentiable_outputs.size() == grad_fns.size());
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
rebase_history(differentiable_outputs[i], grad_fns[i]);
}
}
}
}
```
Rel:
- #58833
- #96405
Pull Request resolved: https://github.com/pytorch/pytorch/pull/101149
Approved by: https://github.com/soulitzer
2023-05-17 03:10:13 +00:00
Masaki Kozuki
6c934a89a7
Skip invalid grads in outplace foreachs' backward ( #100256 )
...
Fixes #100248
Pull Request resolved: https://github.com/pytorch/pytorch/pull/100256
Approved by: https://github.com/soulitzer , https://github.com/albanD
2023-04-29 22:45:26 +00:00
Masaki Kozuki
674018903d
per-Tensor grad_fn for in-place foreach functions ( #96405 )
...
Generate a `grad_fn` for each (tuple of) `Tensor`(s) of the same index for `_foreach_foo_` and each `grad_fn` is `FooBackward`.
The current status of foreach functions' backward support for the record:
- out-place: Implemented, but no optimized implementations like their forward path
- in-place: not implemented. I think this check 7eaaefafb3/torchgen/api/autograd.py (L309-L311)https://github.com/pytorch/pytorch/pull/96405#discussion_r1154690940 ) would prevent in-place from using out-place versions (the logic is around 7eaaefafb3/torchgen/api/autograd.py (L495-L500)#95431
- #95765 for multiple `grad_fn`s logic
---
Examples: outputs of `_foreach_add_.List`, `_foreach_addcmul_.ScalarList`, and `_foreach_exp`
```c++
void _foreach_addcmul__ScalarList(c10::DispatchKeySet ks, at::TensorList self, at::TensorList tensor1, at::TensorList tensor2, at::ArrayRef<at::Scalar> scalars) {
auto self_ = unpack(self, "self", 0);
auto tensor1_ = unpack(tensor1, "tensor1", 1);
auto tensor2_ = unpack(tensor2, "tensor2", 2);
auto _any_requires_grad = compute_requires_grad( self, tensor1, tensor2 );
(void)_any_requires_grad;
std::vector<c10::optional<at::Tensor>> original_selfs(self.size());
std::vector<std::shared_ptr<AddcmulBackward0>> grad_fns;
if (_any_requires_grad) {
for (const auto& i : c10::irange( self.size() )) {
const auto ith_requires_grad = compute_requires_grad(self[i], tensor1[i], tensor2[i]);
check_inplace(self[i], ith_requires_grad);
grad_fns.push_back([&]() -> std::shared_ptr<AddcmulBackward0> {
if (!ith_requires_grad) {
return nullptr;
} else {
auto grad_fn = std::shared_ptr<AddcmulBackward0>(new AddcmulBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self[i], tensor1[i], tensor2[i] ));
return grad_fn;
}
}());
}
if (!grad_fns.empty()) {
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
grad_fn->self_scalar_type = self[i].scalar_type();
grad_fn->tensor1_scalar_type = tensor1[i].scalar_type();
if (grad_fn->should_compute_output(1)) {
grad_fn->tensor2_ = SavedVariable(tensor2[i], false);
}
grad_fn->value = scalars[i];
if (grad_fn->should_compute_output(2)) {
grad_fn->tensor1_ = SavedVariable(tensor1[i], false);
}
grad_fn->tensor2_scalar_type = tensor2[i].scalar_type();
}
}
}
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
std::vector<c10::optional<Storage>> tensor1__storage_saved(tensor1_.size());
for (const Tensor& tensor : tensor1_)
tensor1__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> tensor1__impl_saved(tensor1_.size());
for (size_t i=0; i<tensor1_.size(); i++)
if (tensor1_[i].defined()) tensor1__impl_saved[i] = tensor1_[i].getIntrusivePtr();
std::vector<c10::optional<Storage>> tensor2__storage_saved(tensor2_.size());
for (const Tensor& tensor : tensor2_)
tensor2__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> tensor2__impl_saved(tensor2_.size());
for (size_t i=0; i<tensor2_.size(); i++)
if (tensor2_[i].defined()) tensor2__impl_saved[i] = tensor2_[i].getIntrusivePtr();
#endif
{
at::AutoDispatchBelowAutograd guard;
at::redispatch::_foreach_addcmul_(ks & c10::after_autograd_keyset, self_, tensor1_, tensor2_, scalars);
}
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
for (size_t i=0; i<tensor1_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor1__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(tensor1_))
TORCH_INTERNAL_ASSERT(tensor1__storage_saved[i].value().is_alias_of(tensor1_[i].storage()));
}
for (size_t i=0; i<tensor1_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor1__impl_saved[i] && !at::impl::tensorlist_has_dispatch(tensor1_))
TORCH_INTERNAL_ASSERT(tensor1__impl_saved[i] == tensor1_[i].getIntrusivePtr());
}
for (size_t i=0; i<tensor2_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor2__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(tensor2_))
TORCH_INTERNAL_ASSERT(tensor2__storage_saved[i].value().is_alias_of(tensor2_[i].storage()));
}
for (size_t i=0; i<tensor2_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (tensor2__impl_saved[i] && !at::impl::tensorlist_has_dispatch(tensor2_))
TORCH_INTERNAL_ASSERT(tensor2__impl_saved[i] == tensor2_[i].getIntrusivePtr());
}
#endif
if (!grad_fns.empty()) {
auto differentiable_outputs = flatten_tensor_args( self );
TORCH_INTERNAL_ASSERT(differentiable_outputs.size() == grad_fns.size());
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
rebase_history(differentiable_outputs[i], grad_fns[i]);
}
}
}
}
```
```c++
void _foreach_add__List(c10::DispatchKeySet ks, at::TensorList self, at::TensorList other, const at::Scalar & alpha) {
auto self_ = unpack(self, "self", 0);
auto other_ = unpack(other, "other", 1);
auto _any_requires_grad = compute_requires_grad( self, other );
(void)_any_requires_grad;
std::vector<c10::optional<at::Tensor>> original_selfs(self.size());
std::vector<std::shared_ptr<AddBackward0>> grad_fns;
if (_any_requires_grad) {
for (const auto& i : c10::irange( self.size() )) {
const auto ith_requires_grad = compute_requires_grad(self[i], other[i]);
check_inplace(self[i], ith_requires_grad);
grad_fns.push_back([&]() -> std::shared_ptr<AddBackward0> {
if (!ith_requires_grad) {
return nullptr;
} else {
auto grad_fn = std::shared_ptr<AddBackward0>(new AddBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self[i], other[i] ));
return grad_fn;
}
}());
}
if (!grad_fns.empty()) {
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
grad_fn->other_scalar_type = other[i].scalar_type();
grad_fn->alpha = alpha;
grad_fn->self_scalar_type = self[i].scalar_type();
}
}
}
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
std::vector<c10::optional<Storage>> other__storage_saved(other_.size());
for (const Tensor& tensor : other_)
other__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> other__impl_saved(other_.size());
for (size_t i=0; i<other_.size(); i++)
if (other_[i].defined()) other__impl_saved[i] = other_[i].getIntrusivePtr();
#endif
{
at::AutoDispatchBelowAutograd guard;
at::redispatch::_foreach_add_(ks & c10::after_autograd_keyset, self_, other_, alpha);
}
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
for (size_t i=0; i<other_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (other__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(other_))
TORCH_INTERNAL_ASSERT(other__storage_saved[i].value().is_alias_of(other_[i].storage()));
}
for (size_t i=0; i<other_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (other__impl_saved[i] && !at::impl::tensorlist_has_dispatch(other_))
TORCH_INTERNAL_ASSERT(other__impl_saved[i] == other_[i].getIntrusivePtr());
}
#endif
if (!grad_fns.empty()) {
auto differentiable_outputs = flatten_tensor_args( self );
TORCH_INTERNAL_ASSERT(differentiable_outputs.size() == grad_fns.size());
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
rebase_history(differentiable_outputs[i], grad_fns[i]);
}
}
}
}
...
void _foreach_exp_(c10::DispatchKeySet ks, at::TensorList self) {
auto self_ = unpack(self, "self", 0);
auto _any_requires_grad = compute_requires_grad( self );
(void)_any_requires_grad;
std::vector<c10::optional<at::Tensor>> original_selfs(self.size());
std::vector<std::shared_ptr<ExpBackward0>> grad_fns;
if (_any_requires_grad) {
for (const auto& i : c10::irange( self.size() )) {
const auto ith_requires_grad = compute_requires_grad(self[i]);
check_inplace(self[i], ith_requires_grad);
grad_fns.push_back([&]() -> std::shared_ptr<ExpBackward0> {
if (!ith_requires_grad) {
return nullptr;
} else {
auto grad_fn = std::shared_ptr<ExpBackward0>(new ExpBackward0(), deleteNode);
grad_fn->set_next_edges(collect_next_edges( self[i] ));
return grad_fn;
}
}());
}
}
#ifndef NDEBUG
std::vector<c10::optional<Storage>> self__storage_saved(self_.size());
for (const Tensor& tensor : self_)
self__storage_saved.push_back(
tensor.has_storage() ? c10::optional<Storage>(tensor.storage()) : c10::nullopt);
std::vector<c10::intrusive_ptr<TensorImpl>> self__impl_saved(self_.size());
for (size_t i=0; i<self_.size(); i++)
if (self_[i].defined()) self__impl_saved[i] = self_[i].getIntrusivePtr();
#endif
{
at::AutoDispatchBelowAutograd guard;
at::redispatch::_foreach_exp_(ks & c10::after_autograd_keyset, self_);
}
#ifndef NDEBUG
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__storage_saved[i].has_value() && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__storage_saved[i].value().is_alias_of(self_[i].storage()));
}
for (size_t i=0; i<self_.size() && !at::impl::dispatch_mode_enabled(); i++) {
if (self__impl_saved[i] && !at::impl::tensorlist_has_dispatch(self_))
TORCH_INTERNAL_ASSERT(self__impl_saved[i] == self_[i].getIntrusivePtr());
}
#endif
if (!grad_fns.empty()) {
auto differentiable_outputs = flatten_tensor_args( self );
TORCH_INTERNAL_ASSERT(differentiable_outputs.size() == grad_fns.size());
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
rebase_history(differentiable_outputs[i], grad_fns[i]);
}
}
}
if (!grad_fns.empty()) {
for (const auto& i : c10::irange(grad_fns.size())) {
auto grad_fn = grad_fns[i];
if (grad_fn != nullptr) {
grad_fn->result_ = SavedVariable(self[i], true, self[i].is_view());
}
}
}
}
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/96405
Approved by: https://github.com/soulitzer
2023-04-28 00:55:04 +00:00
Masaki Kozuki
13ca08435c
[test_foreach] add cases of zero size tensors ( #95028 )
...
supply zero-size tensors only if multi_tensor_apply_kernel would be called w.h.p, i.e. device is cuda and dtype is float32
rel:
- https://github.com/pytorch/pytorch/pull/94655
- https://github.com/pytorch/pytorch/issues/94865
Pull Request resolved: https://github.com/pytorch/pytorch/pull/95028
Approved by: https://github.com/ngimel
2023-03-23 00:12:13 +00:00
Masaki Kozuki
a48d518e45
test_foreach: remove skipMeta ( #96599 )
...
Happened to notice that the test doesn't seem to require the guard (at least on my local environment)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/96599
Approved by: https://github.com/bdhirsh
2023-03-13 22:14:36 +00:00
Masaki Kozuki
f54233e273
[foreach] bump tensor's version and define backward via torchgen (as possible) ( #93901 )
...
## summary
- increment tensor versions in inplace foreach functions
- add a logic to take care of `ArrayRef<Scalar>`
rel: https://github.com/pytorch/pytorch/issues/58833 , https://github.com/pytorch/pytorch/pull/89591
Pull Request resolved: https://github.com/pytorch/pytorch/pull/93901
Approved by: https://github.com/albanD
2023-02-20 23:18:07 +00:00
Masaki Kozuki
3e9df622fb
[mta] implement _foreach_pow ( #92303 )
...
Mainly for foreach path of `Adam` and `AdamW`
rel: https://github.com/pytorch/pytorch/issues/58833
Pull Request resolved: https://github.com/pytorch/pytorch/pull/92303
Approved by: https://github.com/albanD
2023-02-16 02:28:26 +00:00
Aaron Gokaslan
67d9790985
[BE] Apply almost all remaining flake8-comprehension checks ( #94676 )
...
Applies the remaining flake8-comprehension fixes and checks. This changes replace all remaining unnecessary generator expressions with list/dict/set comprehensions which are more succinct, performant, and better supported by our torch.jit compiler. It also removes useless generators such as 'set(a for a in b)`, resolving it into just the set call.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/94676
Approved by: https://github.com/ezyang
2023-02-12 01:01:25 +00:00
Masaki Kozuki
30876229a7
[mta] Backward of unary foreach functions ( #89591 )
...
as per title, this PR defines backward of those.
This doesn't implement forward-mode automatic differentiation as [the current codegen](a747326423/tools/autograd/gen_variable_type.py (L1513)https://github.com/pytorch/pytorch/issues/53796
- https://github.com/pytorch/pytorch/issues/58833
Pull Request resolved: https://github.com/pytorch/pytorch/pull/89591
Approved by: https://github.com/albanD
2023-01-23 08:28:06 +00:00
Masaki Kozuki
32b2d8009a
check if multi_tensor_apply_kernel was called ( #92077 )
...
Replacing all the hard coded number of cuda kernel launches with `multi_tensor_apply_kernel` call check, keeping the dependency on kineto profiler there
Rel: https://github.com/pytorch/pytorch/pull/91844#issuecomment-1379844523
Pull Request resolved: https://github.com/pytorch/pytorch/pull/92077
Approved by: https://github.com/ngimel
2023-01-23 06:46:36 +00:00
milesial
a76bc410df
Fix _foreach_norm on some tensor sizes ( #91844 )
...
This PR fixes 2 bugs with CUDA `_foreach_norm`:
1. Wrong norm when tensors are larger than kChunkSize = 65536
```
>>> torch._foreach_norm([torch.ones(60000, device="cuda") for _ in range(1)])
(tensor(244.9490, device='cuda:0', grad_fn=<NotImplemented>),)
>>> torch._foreach_norm([torch.ones(70000, device="cuda") for _ in range(1)])
(tensor(256., device='cuda:0', grad_fn=<NotImplemented>),)
>>> torch.ones(60000, device="cuda").norm()
tensor(244.9490, device='cuda:0', grad_fn=<LinalgVectorNormBackward0>)
>>> torch.ones(70000, device="cuda").norm()
tensor(264.5751, device='cuda:0', grad_fn=<LinalgVectorNormBackward0>)
```
2. Error when a tensor numel is smaller than the number of tensors
```
>> torch._foreach_norm([torch.ones(9, device="cuda") for _ in range(10)])
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
IndexError: select(): index 9 out of range for tensor of size [9] at dimension 0
```
This bug could have been caught by tests if `PYTORCH_TEST_WITH_SLOW` was 1, because it would have tested tensors of size 300*300=90000. It's not enabled by default, does someone know if it's ever enabled?
Pull Request resolved: https://github.com/pytorch/pytorch/pull/91844
Approved by: https://github.com/ngimel
2023-01-12 05:48:01 +00:00
Masaki Kozuki
554a796aef
Implement torch._foreach_lerp ( #87562 )
...
As per title.
- [ ] ~~Q: Do we want `torch._foreach_lerp.ScalarList` as well?~~
- [ ] ~~we might want to have `ATen/native/cuda/lerp.cuh` and include it in `ATen/native/cuda/Lerp.cu` and `ATen/native/cuda/ForeachTernaryOp.cu`~~
Related:
- https://github.com/pytorch/pytorch/issues/58833
- https://github.com/pytorch/pytorch/issues/71683
Pull Request resolved: https://github.com/pytorch/pytorch/pull/87562
Approved by: https://github.com/ngimel
2023-01-11 02:52:04 +00:00
milesial
9d20d6d5ec
Foreach clamp_min clamp_max ( #91384 )
...
Adds `_foreach_clamp_min` and `_foreach_clamp_max` as binary ops, with scalar, scalarlist and tensorlist support.
Timing example for `_foreach_clamp_min_` on a GTX3070Ti across a list of tensors with varying count and item size (times are in microseconds (us)):
CUDA:
```
[------------------ (tensors, scalar) -------------------]
| for loop | foreach
10 tensors of size 4 | 29.0 | 10.2
100 tensors of size 4 | 234.4 | 18.3
1000 tensors of size 4 | 2194.1 | 113.5
10000 tensors of size 4 | 21745.6 | 1144.5
10 tensors of size 16 | 29.5 | 12.0
100 tensors of size 16 | 256.9 | 19.9
1000 tensors of size 16 | 2499.7 | 123.6
10000 tensors of size 16 | 25022.2 | 1295.6
10 tensors of size 256 | 32.8 | 11.2
100 tensors of size 256 | 258.8 | 19.7
1000 tensors of size 256 | 2509.2 | 123.7
10000 tensors of size 256 | 25016.2 | 1295.4
10 tensors of size 65536 | 32.9 | 18.7
100 tensors of size 65536 | 327.1 | 150.3
1000 tensors of size 65536 | 3051.3 | 1388.0
10000 tensors of size 65536 | 30476.9 | 14021.5
[------------------ (tensors, tensors) ------------------]
| for loop | foreach
10 tensors of size 4 | 26.8 | 17.3
100 tensors of size 4 | 206.8 | 90.5
1000 tensors of size 4 | 1993.0 | 828.9
10000 tensors of size 4 | 19851.0 | 9063.3
10 tensors of size 16 | 34.7 | 20.0
100 tensors of size 16 | 232.2 | 102.1
1000 tensors of size 16 | 2220.9 | 977.3
10000 tensors of size 16 | 22644.5 | 10361.4
10 tensors of size 256 | 30.5 | 19.7
100 tensors of size 256 | 231.6 | 102.4
1000 tensors of size 256 | 2251.9 | 978.7
10000 tensors of size 256 | 22680.3 | 10405.8
10 tensors of size 65536 | 30.6 | 34.4
100 tensors of size 65536 | 315.1 | 223.6
1000 tensors of size 65536 | 3252.1 | 2114.4
10000 tensors of size 65536 | 30578.0 | 22826.3
```
CPU:
```
[------------------- (tensors, scalar) -------------------]
| for loop | foreach
10 tensors of size 4 | 13.0 | 9.6
100 tensors of size 4 | 62.4 | 31.6
1000 tensors of size 4 | 562.2 | 245.6
10000 tensors of size 4 | 5552.2 | 2517.7
10 tensors of size 16 | 14.9 | 11.3
100 tensors of size 16 | 74.1 | 36.9
1000 tensors of size 16 | 663.7 | 285.5
10000 tensors of size 16 | 6765.2 | 2947.5
10 tensors of size 256 | 15.2 | 11.8
100 tensors of size 256 | 76.0 | 37.7
1000 tensors of size 256 | 728.8 | 323.9
10000 tensors of size 256 | 7274.4 | 3800.3
10 tensors of size 65536 | 105.6 | 124.5
100 tensors of size 65536 | 982.8 | 939.7
1000 tensors of size 65536 | 14993.1 | 14579.2
10000 tensors of size 65536 | 163091.0 | 151555.8
[------------------- (tensors, tensors) ------------------]
| for loop | foreach
10 tensors of size 4 | 11.8 | 10.5
100 tensors of size 4 | 53.1 | 38.2
1000 tensors of size 4 | 465.1 | 316.1
10000 tensors of size 4 | 4616.9 | 3625.9
10 tensors of size 16 | 13.5 | 12.3
100 tensors of size 16 | 63.0 | 46.5
1000 tensors of size 16 | 560.1 | 359.9
10000 tensors of size 16 | 5586.8 | 3765.9
10 tensors of size 256 | 15.2 | 13.7
100 tensors of size 256 | 64.4 | 48.3
1000 tensors of size 256 | 653.7 | 410.0
10000 tensors of size 256 | 5916.6 | 3901.3
10 tensors of size 65536 | 109.1 | 106.8
100 tensors of size 65536 | 1128.9 | 1105.0
1000 tensors of size 65536 | 16245.0 | 15950.8
10000 tensors of size 65536 | 171111.3 | 163540.2
```
Example use:
```
tensors = [torch.randn(16, device='cuda') for _ in range(10)]
out = torch._foreach_clamp_min(tensors, 0.1)
out = torch._foreach_clamp_min(tensors, [0.1] * len(tensors))
out = torch._foreach_clamp_min(tensors, tensors)
torch._foreach_clamp_min_(tensors, 0.1)
torch._foreach_clamp_min_(tensors, [0.1] * len(tensors))
torch._foreach_clamp_min_(tensors, tensors)
```
Does not support complex types.
Changes the existing `foreach_minimum/maximum` to use this new implementation.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/91384
Approved by: https://github.com/ngimel
2023-01-09 19:28:47 +00:00
Christian Puhrsch
6fd416650a
Add _foreach_addc(div/mul)(_).Tensor ( #88157 )
...
Support passing value scalars as a flat 1D Tensor.
Currently we can only pass either an individual scalar or a ScalarList.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/88157
Approved by: https://github.com/ngimel , https://github.com/albanD
2022-11-02 23:24:35 +00:00
Elias Ellison
f701cb04fb
Test Dynamo CI w Fake Tensors ( #84282 )
...
Pull Request resolved: https://github.com/pytorch/pytorch/pull/84282
Approved by: https://github.com/anijain2305
2022-09-01 00:15:05 +00:00
Masaki Kozuki
3139722679
[foreach][mta] Inplace maximum and minimum ( #82523 )
...
### Description
<!-- What did you change and why was it needed? -->
Implement `torch._foreach_maximum_` and `torch._foreach_minimum_` mainly for `_multi_tensor_adam` and `_multi_tensor_adamw` with `amsgrad=True` to correctly update their `max_exp_avg_sqs`.
### Issue
<!-- Link to Issue ticket or RFP -->
- https://github.com/pytorch/pytorch/issues/78807
- https://github.com/pytorch/pytorch/pull/81894
- https://github.com/pytorch/pytorch/pull/81348
- https://github.com/pytorch/pytorch/pull/81705
- https://github.com/pytorch/pytorch/issues/58833
- https://github.com/pytorch/pytorch/issues/68041
### Testing
<!-- How did you test your change? -->
Updated `test_foreach.py::TestForeach::_minmax_test` to compare the outputs of `_foreach_maximum_` (and `_foreach_minimum_`) against those of `[torch.maximum(a, b) for a, b in zip(tensors1, tensors2)]`
cc @ngimel @albanD @mikaylagawarecki
Pull Request resolved: https://github.com/pytorch/pytorch/pull/82523
Approved by: https://github.com/albanD
2022-08-03 03:40:42 +00:00
Nikita Shulga
bfac65dfe5
[testing] Update dispatch macros ( #74977 )
...
This PR is reland of #74289
Co-authored-by: Khushi Agrawal <khushiagrawal411@gmail.com>
2022-03-30 14:13:21 -07:00
PyTorch MergeBot
2e4152b118
Revert "[testing] Update dispatch macros"
...
This reverts commit eed19a0f38https://github.com/pytorch/pytorch/pull/74289 on behalf of https://github.com/malfet
2022-03-30 19:52:37 +00:00
Khushi Agrawal
eed19a0f38
[testing] Update dispatch macros
...
Hi,
This PR is the follow-up PR of #71561 . (the previous PR had a couple of merge conflicts and was reverted, this PR resolves that).
Please take a look. Thanks!
cc: @pmeier @mruberry @kshitij12345
Pull Request resolved: https://github.com/pytorch/pytorch/pull/74289
Approved by: https://github.com/pmeier , https://github.com/mruberry
2022-03-30 16:10:16 +00:00
