OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity
09e610e36d
pytorch/torch
History
Ray Peng 09e610e36d [Static Runtime] Out version for softmax (#64243) 
Summary: Pull Request resolved: https://github.com/pytorch/pytorch/pull/64243

Test Plan:
```
> buck run //caffe2/benchmarks/static_runtime:static_runtime_cpptest -- --v=1
...
V0830 16:35:22.524479 613839 impl.cpp:1410] Switch to out variant for node: %5 : Tensor = aten::softmax(%a.1, %dim.1, %dtype.1)
...
[       OK ] StaticRuntime.IndividualOps_Softmax (803 ms)
```

Reviewed By: hlu1

Differential Revision: D30656149

fbshipit-source-id: 115b7b4a75448fd6a5c526808080ca9a4251302c
2021-08-31 18:33:26 -07:00
..
_C Revert D30543236: Add python mode 2021-08-31 15:28:33 -07:00
ao [quant] AO migration of the quantize.py (#64086) 2021-08-29 20:30:01 -07:00
autograd Allow implementing either backward or vjp for Function (#63434) 2021-08-23 07:07:11 -07:00
backends NNAPI: Support const values in binary ops 2021-08-20 21:10:26 -07:00
contrib Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
cpu add operation list for AutocastCPU (#63534) 2021-08-30 19:30:33 -07:00
csrc [Static Runtime] Out version for softmax (#64243) 2021-08-31 18:33:26 -07:00
cuda fix syntax error in bfloat16 PR (#64122) 2021-08-31 14:33:12 -07:00
distributed Adding alltoall_single collective to collective quantization API (#63154) 2021-08-27 12:46:31 -07:00
distributions Poisson zero rate (#61511) 2021-08-19 08:30:28 -07:00
fft Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
for_onnx
futures Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
fx Fix redundant class definition in GraphModule singleton constructor (#64274) 2021-08-31 17:34:14 -07:00
jit More robust check of whether a class is defined in torch (#64083) 2021-08-27 08:55:35 -07:00
legacy
lib Disable avoid-non-const-global-variables lint check (#62008) 2021-07-22 18:04:40 -07:00
linalg Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
multiprocessing
nn Add optional tensor arguments to (#63967) 2021-08-30 19:21:26 -07:00
onnx ENH Adds label_smoothing to cross entropy loss (#63122) 2021-08-29 23:33:04 -07:00
optim [DOC] improve docstring for Optimizer.state_dict (#63153) 2021-08-29 10:20:58 -07:00
package Revert D29639797: [package] error if we try to mock a module in 3.6 2021-07-09 19:31:04 -07:00
profiler [Profiler] Change FLOP/s to Total FLOPs (#62779) 2021-08-16 13:43:32 -07:00
quantization ns for fx: make layer types more readable (#64270) 2021-08-31 12:31:34 -07:00
sparse Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
special [special] use __all__ to hide internal imports (#63135) 2021-08-19 12:45:43 -07:00
testing Add OpInfo for nn.functional.cosine_similarity (#62959) 2021-08-31 10:31:36 -07:00
utils Revert D30543236: Add python mode 2021-08-31 15:28:33 -07:00
__config__.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
__future__.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
__init__.py Updates set_default_dtype documentation (#63233) 2021-08-17 10:41:03 -07:00
_appdirs.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_classes.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_deploy.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_jit_internal.py Fix bug in check_empty_containers (#63492) 2021-08-25 09:05:08 -07:00
_linalg_utils.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_lobpcg.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_lowrank.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_namedtensor_internals.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_ops.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_python_dispatcher.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_six.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_sources.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_storage_docs.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_tensor_docs.py [docs] Remove input parameter from Tensor.flatten docs (#63180) 2021-08-13 12:10:16 -07:00
_tensor_str.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_tensor.py Use stacklevel for floordiv deprecation warnings (#64034) 2021-08-31 11:27:56 -07:00
_torch_docs.py document that torch.triangular_solve has optional out= parameter (#63253) 2021-08-26 17:28:17 -07:00
_utils_internal.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_utils.py Add support for the ONNX Runtime Eager Mode backend (#58248) 2021-08-20 11:17:13 -07:00
_VF.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
_vmap_internals.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
abi-check.cpp
autocast_mode.py bf16 Error message cleanup as well as addition of is_bf16_supported (#63798) 2021-08-25 09:59:59 -07:00
CMakeLists.txt [torch/deploy] add torch.distributed to build (#63918) 2021-08-26 20:58:44 -07:00
custom_class_detail.h
custom_class.h
deploy.h
extension.h
functional.py document why wrappers exist in torch.functional (#62847) 2021-08-18 11:51:21 -07:00
hub.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
library.h Add support for the ONNX Runtime Eager Mode backend (#58248) 2021-08-20 11:17:13 -07:00
overrides.py ENH Adds label_smoothing to cross entropy loss (#63122) 2021-08-29 23:33:04 -07:00
py.typed
quasirandom.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
random.py Adds return type annotation for fork_rng function (#63724) 2021-08-27 09:03:40 -07:00
README.txt
script.h
serialization.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
storage.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
torch_version.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00
types.py Revert D30279364: [codemod][lint][fbcode/c*] Enable BLACK by default 2021-08-12 11:45:01 -07:00

README.txt 

Note [TH abstraction violation]
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

TH/THC provide some hpp headers, which are proper C++ headers rather than
C headers.  These headers serve double duty as *internal implementation
detail* headers, whose contents should largely not be used by external
clients.

Ideally, we would not install these headers at all; instead, you should
use public functions (in headers like `THTensor.h`, NOT `THTensor.hpp`)
to manipulate these structs.  However, there are a few places
in torch/csrc where we violate this abstraction.  They are marked with
a pointer to this note.  Each of those sites will have to be refactored
when we refactor the guts of THTensor and related structures.