Remove optional for veiw_fn during View Tracking by ejguan · Pull Request #50067 · pytorch/pytorch

ejguan · 2021-01-04T21:18:10Z

Stack from ghstack:

Remove optional for veiw_fn during View Tracking #50067 Remove optional for veiw_fn during View Tracking

Using the Callgrind to test the performance.

import torch
import timeit
from torch.utils.benchmark import Timer

timer = Timer("x.view({100, 5, 20});", setup="torch::Tensor x = torch::ones({10, 10, 100});", language="c++", timer=timeit.default_timer)
res = timer.collect_callgrind(number=10)

Nightly

<torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7fdd49d61c70>
x.view({100, 5, 20});
setup: torch::Tensor x = torch::ones({10, 10, 100});
                           All          Noisy symbols removed
    Instructions:        41850                      41850
    Baseline:                0                          0
10 runs per measurement, 1 thread
Warning: PyTorch was not built with debug symbols.
         Source information may be limited. Rebuild with
         REL_WITH_DEB_INFO=1 for more detailed results.

Current

<torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7fdcf322d1c0>
x.view({100, 5, 20});
setup: torch::Tensor x = torch::ones({10, 10, 100});
                           All          Noisy symbols removed
    Instructions:        42020                      42020
    Baseline:                0                          0
10 runs per measurement, 1 thread
Warning: PyTorch was not built with debug symbols.
         Source information may be limited. Rebuild with
         REL_WITH_DEB_INFO=1 for more detailed results.

Compare

There are 170 instructions added

<torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.FunctionCounts object at 0x7fdd427df7f0>
    970  ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, std::function<at::Tensor (at::Tensor const&)>, torch::autograd::CreationMeta, bool)
    240  ???:torch::autograd::ViewInfo::~ViewInfo()
    180  ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, std::function<at::Tensor (at::Tensor const&)>)
    130  ???:torch::autograd::make_variable_differentiable_view(at::Tensor const&, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta, bool)
    105  /tmp/benchmark_utils_jit_build_66d13e5035484028ab02edcb72545e37/timer_cpp_2208139565547260839/timer_src.cpp:main
    100  ???:std::function<at::Tensor (at::Tensor const&)>::function(std::function<at::Tensor (at::Tensor const&)> const&)
     70  ???:torch::autograd::DifferentiableViewMeta::~DifferentiableViewMeta()
     70  ???:torch::autograd::DifferentiableViewMeta::DifferentiableViewMeta(c10::TensorImpl*, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta)
   -100  ???:c10::optional_base<torch::autograd::ViewInfo>::optional_base(c10::optional_base<torch::autograd::ViewInfo>&&)
   -105  /tmp/benchmark_utils_jit_build_a62a84fbc4cd4f9e91376c1bb3dcb697/timer_cpp_7842929001219208348/timer_src.cpp:main
   -120  ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, c10::optional<std::function<at::Tensor (at::Tensor const&)> >)
   -210  ???:c10::optional_base<std::function<at::Tensor (at::Tensor const&)> >::~optional_base()
   -240  ???:c10::optional_base<torch::autograd::ViewInfo>::~optional_base()
   -920  ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, c10::optional<std::function<at::Tensor (at::Tensor const&)> >, torch::autograd::CreationMeta, bool)

Total: 170

Differential Revision: D25900495

[ghstack-poisoned]

ghstack-source-id: 88cc152 Pull Request resolved: #50067

facebook-github-bot · 2021-01-04T21:18:27Z

💊 CI failures summary and remediations

As of commit 31b2a5f (more details on the Dr. CI page):

💚 💚 Looks good so far! There are no failures yet. 💚 💚

This comment was automatically generated by Dr. CI (expand for details).

Follow this link to opt-out of these comments for your Pull Requests.

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Fixes #49257 Using the following script to test the performance. ```python import torch x = torch.randn(100, 100, requires_grad=True) with torch.autograd.profiler.profile() as prof: for _ in range(100): y = torch.sum(torch.diagonal(x)) y.backward() print(prof.key_averages().table(sort_by="self_cpu_time_total")) ``` ### Nightly ![image](https://user-images.githubusercontent.com/68879799/103581107-53f5f680-4ea9-11eb-800a-d6570d1c8b6b.png) ### Current ![image](https://user-images.githubusercontent.com/68879799/103581114-5bb59b00-4ea9-11eb-9aa8-038741bc8407.png) I don't find any significant regression over backward function (using the `view_fn`) Differential Revision: [D25768847](https://our.internmc.facebook.com/intern/diff/D25768847) [ghstack-poisoned]

ghstack-source-id: 72c9679 Pull Request resolved: #50067

codecov · 2021-01-05T19:35:38Z

Codecov Report

Merging #50067 (31b2a5f) into gh/ejguan/18/base (5834438) will decrease coverage by 0.00%.
The diff coverage is 86.66%.

@@                  Coverage Diff                  @@
##           gh/ejguan/18/base   #50067      +/-   ##
=====================================================
- Coverage              80.70%   80.70%   -0.01%     
=====================================================
  Files                   1905     1905              
  Lines                 206813   206811       -2     
=====================================================
- Hits                  166916   166912       -4     
- Misses                 39897    39899       +2

albanD

This change looks good.

Can you check the perf implication with the new Timer class (https://pytorch.org/docs/master/benchmark_utils.html?highlight=timer#torch.utils.benchmark.Timer) with the instruction count mode that will be much more precise!
You can do somthing like this to save a trace (here for a basic view op, but maybe something else if it is more relevant to your PR):

import torch
from torch.utils.benchmark import Timer

timer = Timer("x.view(-1);", setup="torch::Tensor x = torch::ones({1,2,3});", language="c++")
res = timer.collect_callgrind(number=10)

torch.save(res, "save.pth")

And once you do that for master and your branch, you can compare with:

import torch
torch.set_printoptions(linewidth=260)

master_instr = torch.load("master.pth")
fast_view_instr = torch.load("fast_view.pth")

master_instr = master_instr.as_standardized()
fast_view_instr = fast_view_instr.as_standardized()


print(master_instr)
print(fast_view_instr)

delta = fast_view_instr.delta(master_instr).denoise()

print(delta)

Fixes #49257 Using the following script to test the performance. ```python import torch x = torch.randn(100, 100, requires_grad=True) with torch.autograd.profiler.profile() as prof: for _ in range(100): y = torch.sum(torch.diagonal(x)) y.backward() print(prof.key_averages().table(sort_by="self_cpu_time_total")) ``` ### Nightly ![image](https://user-images.githubusercontent.com/68879799/103666773-0d54da80-4f43-11eb-9f04-d7b94276f82f.png) ### Current ![image](https://user-images.githubusercontent.com/68879799/103665874-eba72380-4f41-11eb-9843-61fcd801464f.png) I don't find any significant regression over backward function (using the `view_fn`) Differential Revision: [D25768847](https://our.internmc.facebook.com/intern/diff/D25768847) [ghstack-poisoned]

ghstack-source-id: 63e21c1 Pull Request resolved: #50067

Fixes #49257 Using the following script to test the performance. ```python import torch x = torch.randn(100, 100, requires_grad=True) with torch.autograd.profiler.profile() as prof: for _ in range(100): y = torch.sum(torch.diagonal(x)) y.backward() print(prof.key_averages().table(sort_by="self_cpu_time_total")) ``` ### Nightly ![image](https://user-images.githubusercontent.com/68879799/103666773-0d54da80-4f43-11eb-9f04-d7b94276f82f.png) ### Current ![image](https://user-images.githubusercontent.com/68879799/103665874-eba72380-4f41-11eb-9843-61fcd801464f.png) I don't find any significant regression over backward function (using the `view_fn`) Differential Revision: [D25768847](https://our.internmc.facebook.com/intern/diff/D25768847) [ghstack-poisoned]

ghstack-source-id: 9a65e14 Pull Request resolved: #50067

Fixes #49257 Using the `Callgrind` to test the performance. ```python import torch import timeit from torch.utils.benchmark import Timer timer = Timer("x.view({100, 5, 20});", setup="torch::Tensor x = torch::ones({10, 10, 100});", language="c++", timer=timeit.default_timer) res = timer.collect_callgrind(number=10) ``` ### Nightly ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f7949138c40> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42310 42310 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Current ```python <torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f78f271a580> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42480 42480 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Compare There are 170 instructions reduced ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.FunctionCounts object at 0x7f7941b7a7c0> 970 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, std::function<at::Tensor (at::Tensor const&)>, torch::autograd::CreationMeta, bool) 240 ???:torch::autograd::ViewInfo::~ViewInfo() 180 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, std::function<at::Tensor (at::Tensor const&)>) 130 ???:torch::autograd::make_variable_differentiable_view(at::Tensor const&, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta, bool) 105 /tmp/benchmark_utils_jit_build_69e2f1710544485588feeca0719a3a57/timer_cpp_4435526292782672407/timer_src.cpp:main 100 ???:std::function<at::Tensor (at::Tensor const&)>::function(std::function<at::Tensor (at::Tensor const&)> const&) 70 ???:torch::autograd::DifferentiableViewMeta::~DifferentiableViewMeta() 70 ???:torch::autograd::DifferentiableViewMeta::DifferentiableViewMeta(c10::TensorImpl*, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta) -100 ???:c10::optional_base<torch::autograd::ViewInfo>::optional_base(c10::optional_base<torch::autograd::ViewInfo>&&) -105 /tmp/benchmark_utils_jit_build_2e75f38b553e42eba00523a86ad9aa05/timer_cpp_3360771523810516633/timer_src.cpp:main -120 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, c10::optional<std::function<at::Tensor (at::Tensor const&)> >) -210 ???:c10::optional_base<std::function<at::Tensor (at::Tensor const&)> >::~optional_base() -240 ???:c10::optional_base<torch::autograd::ViewInfo>::~optional_base() -920 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, c10::optional<std::function<at::Tensor (at::Tensor const&)> >, torch::autograd::CreationMeta, bool) ``` Differential Revision: [D25768847](https://our.internmc.facebook.com/intern/diff/D25768847) [ghstack-poisoned]

ghstack-source-id: cf2c646 Pull Request resolved: #50067

Fixes #49257 Using the `Callgrind` to test the performance. ```python import torch import timeit from torch.utils.benchmark import Timer timer = Timer("x.view({100, 5, 20});", setup="torch::Tensor x = torch::ones({10, 10, 100});", language="c++", timer=timeit.default_timer) res = timer.collect_callgrind(number=10) ``` ### Nightly ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f7949138c40> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42310 42310 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Current ```python <torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f78f271a580> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42480 42480 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Compare There are 170 instructions reduced ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.FunctionCounts object at 0x7f7941b7a7c0> 970 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, std::function<at::Tensor (at::Tensor const&)>, torch::autograd::CreationMeta, bool) 240 ???:torch::autograd::ViewInfo::~ViewInfo() 180 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, std::function<at::Tensor (at::Tensor const&)>) 130 ???:torch::autograd::make_variable_differentiable_view(at::Tensor const&, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta, bool) 105 /tmp/benchmark_utils_jit_build_69e2f1710544485588feeca0719a3a57/timer_cpp_4435526292782672407/timer_src.cpp:main 100 ???:std::function<at::Tensor (at::Tensor const&)>::function(std::function<at::Tensor (at::Tensor const&)> const&) 70 ???:torch::autograd::DifferentiableViewMeta::~DifferentiableViewMeta() 70 ???:torch::autograd::DifferentiableViewMeta::DifferentiableViewMeta(c10::TensorImpl*, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta) -100 ???:c10::optional_base<torch::autograd::ViewInfo>::optional_base(c10::optional_base<torch::autograd::ViewInfo>&&) -105 /tmp/benchmark_utils_jit_build_2e75f38b553e42eba00523a86ad9aa05/timer_cpp_3360771523810516633/timer_src.cpp:main -120 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, c10::optional<std::function<at::Tensor (at::Tensor const&)> >) -210 ???:c10::optional_base<std::function<at::Tensor (at::Tensor const&)> >::~optional_base() -240 ???:c10::optional_base<torch::autograd::ViewInfo>::~optional_base() -920 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, c10::optional<std::function<at::Tensor (at::Tensor const&)> >, torch::autograd::CreationMeta, bool) ``` [ghstack-poisoned]

ghstack-source-id: 302d8da Pull Request resolved: #50067

Fixes #49257 Using the `Callgrind` to test the performance. ```python import torch import timeit from torch.utils.benchmark import Timer timer = Timer("x.view({100, 5, 20});", setup="torch::Tensor x = torch::ones({10, 10, 100});", language="c++", timer=timeit.default_timer) res = timer.collect_callgrind(number=10) ``` ### Nightly ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f7949138c40> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42310 42310 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Current ```python <torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f78f271a580> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42480 42480 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Compare There are 170 instructions reduced ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.FunctionCounts object at 0x7f7941b7a7c0> 970 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, std::function<at::Tensor (at::Tensor const&)>, torch::autograd::CreationMeta, bool) 240 ???:torch::autograd::ViewInfo::~ViewInfo() 180 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, std::function<at::Tensor (at::Tensor const&)>) 130 ???:torch::autograd::make_variable_differentiable_view(at::Tensor const&, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta, bool) 105 /tmp/benchmark_utils_jit_build_69e2f1710544485588feeca0719a3a57/timer_cpp_4435526292782672407/timer_src.cpp:main 100 ???:std::function<at::Tensor (at::Tensor const&)>::function(std::function<at::Tensor (at::Tensor const&)> const&) 70 ???:torch::autograd::DifferentiableViewMeta::~DifferentiableViewMeta() 70 ???:torch::autograd::DifferentiableViewMeta::DifferentiableViewMeta(c10::TensorImpl*, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta) -100 ???:c10::optional_base<torch::autograd::ViewInfo>::optional_base(c10::optional_base<torch::autograd::ViewInfo>&&) -105 /tmp/benchmark_utils_jit_build_2e75f38b553e42eba00523a86ad9aa05/timer_cpp_3360771523810516633/timer_src.cpp:main -120 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, c10::optional<std::function<at::Tensor (at::Tensor const&)> >) -210 ???:c10::optional_base<std::function<at::Tensor (at::Tensor const&)> >::~optional_base() -240 ???:c10::optional_base<torch::autograd::ViewInfo>::~optional_base() -920 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, c10::optional<std::function<at::Tensor (at::Tensor const&)> >, torch::autograd::CreationMeta, bool) ``` [ghstack-poisoned]

ghstack-source-id: c002b77 Pull Request resolved: #50067

ejguan · 2021-01-13T18:02:09Z

@albanD
It's quite interesting that It doesn't seem there is benefit in terms of instruction counts by removing optional for function, as shown in the above comment.

albanD · 2021-01-14T12:42:54Z

Hi,

Thanks for the benchmark, after discussing with other people, it is expected that the runtime is the same (the compiler is too smart for us): https://godbolt.org/z/W16xos (thanks @swolchok for the godbolt link).
One benefit that is not measured here though is the smaller memory footprint which is always nice!

albanD

The updated code looks good to me, thanks!

facebook-github-bot · 2021-01-15T16:33:48Z

@ejguan merged this pull request in 00d432a.

Summary: Pull Request resolved: pytorch#50067 Fixes pytorch#49257 Using the `Callgrind` to test the performance. ```python import torch import timeit from torch.utils.benchmark import Timer timer = Timer("x.view({100, 5, 20});", setup="torch::Tensor x = torch::ones({10, 10, 100});", language="c++", timer=timeit.default_timer) res = timer.collect_callgrind(number=10) ``` ### Nightly ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f7949138c40> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42310 42310 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Current ```python <torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.CallgrindStats object at 0x7f78f271a580> x.view({100, 5, 20}); setup: torch::Tensor x = torch::ones({10, 10, 100}); All Noisy symbols removed Instructions: 42480 42480 Baseline: 0 0 10 runs per measurement, 1 thread Warning: PyTorch was not built with debug symbols. Source information may be limited. Rebuild with REL_WITH_DEB_INFO=1 for more detailed results. ``` ### Compare There are 170 instructions reduced ```python torch.utils.benchmark.utils.valgrind_wrapper.timer_interface.FunctionCounts object at 0x7f7941b7a7c0> 970 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, std::function<at::Tensor (at::Tensor const&)>, torch::autograd::CreationMeta, bool) 240 ???:torch::autograd::ViewInfo::~ViewInfo() 180 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, std::function<at::Tensor (at::Tensor const&)>) 130 ???:torch::autograd::make_variable_differentiable_view(at::Tensor const&, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta, bool) 105 /tmp/benchmark_utils_jit_build_69e2f1710544485588feeca0719a3a57/timer_cpp_4435526292782672407/timer_src.cpp:main 100 ???:std::function<at::Tensor (at::Tensor const&)>::function(std::function<at::Tensor (at::Tensor const&)> const&) 70 ???:torch::autograd::DifferentiableViewMeta::~DifferentiableViewMeta() 70 ???:torch::autograd::DifferentiableViewMeta::DifferentiableViewMeta(c10::TensorImpl*, c10::optional<torch::autograd::ViewInfo>, c10::optional<torch::autograd::ViewInfo>, torch::autograd::CreationMeta) -100 ???:c10::optional_base<torch::autograd::ViewInfo>::optional_base(c10::optional_base<torch::autograd::ViewInfo>&&) -105 /tmp/benchmark_utils_jit_build_2e75f38b553e42eba00523a86ad9aa05/timer_cpp_3360771523810516633/timer_src.cpp:main -120 ???:torch::autograd::ViewInfo::ViewInfo(at::Tensor, c10::optional<std::function<at::Tensor (at::Tensor const&)> >) -210 ???:c10::optional_base<std::function<at::Tensor (at::Tensor const&)> >::~optional_base() -240 ???:c10::optional_base<torch::autograd::ViewInfo>::~optional_base() -920 ???:torch::autograd::as_view(at::Tensor const&, at::Tensor const&, bool, bool, c10::optional<std::function<at::Tensor (at::Tensor const&)> >, torch::autograd::CreationMeta, bool) ``` Test Plan: Imported from OSS Reviewed By: albanD Differential Revision: D25900495 Pulled By: ejguan fbshipit-source-id: dedd30e69db6b48601a18ae98d6b28faeae30d90

Remove optional for veiw_fn during View Tracking

b9802b6

[ghstack-poisoned]

ejguan requested a review from albanD as a code owner January 4, 2021 21:18

ejguan added a commit that referenced this pull request Jan 4, 2021

Remove optional for veiw_fn during View Tracking

b9d2c09

ghstack-source-id: 88cc152 Pull Request resolved: #50067

facebook-github-bot added the cla signed label Jan 4, 2021

ejguan linked an issue Jan 4, 2021 that may be closed by this pull request

View tracking for autograd should not save optional std::function #49257

Closed

ejguan marked this pull request as draft January 5, 2021 14:29

ejguan added a commit that referenced this pull request Jan 5, 2021

Remove optional for veiw_fn during View Tracking

9ee2bbb

ghstack-source-id: 72c9679 Pull Request resolved: #50067

ejguan marked this pull request as ready for review January 5, 2021 16:59

albanD reviewed Jan 5, 2021

View reviewed changes

Comment thread torch/csrc/autograd/functions/tensor.cpp Outdated

Comment thread torch/csrc/autograd/variable.cpp Outdated

ejguan commented Jan 5, 2021

View reviewed changes

Comment thread torch/csrc/autograd/variable.h

ejguan added a commit that referenced this pull request Jan 5, 2021

Remove optional for veiw_fn during View Tracking

612f2f5

ghstack-source-id: 63e21c1 Pull Request resolved: #50067

ejguan added a commit that referenced this pull request Jan 6, 2021

Remove optional for veiw_fn during View Tracking

c1ce8fd

ghstack-source-id: 9a65e14 Pull Request resolved: #50067

ejguan added a commit that referenced this pull request Jan 8, 2021

Remove optional for veiw_fn during View Tracking

599b0e5

ghstack-source-id: cf2c646 Pull Request resolved: #50067

ejguan added a commit that referenced this pull request Jan 11, 2021

Remove optional for veiw_fn during View Tracking

af9b929

ghstack-source-id: 302d8da Pull Request resolved: #50067

ejguan added a commit that referenced this pull request Jan 13, 2021

Remove optional for veiw_fn during View Tracking

15fdc19

ghstack-source-id: c002b77 Pull Request resolved: #50067

ejguan requested a review from albanD January 13, 2021 17:57

albanD approved these changes Jan 14, 2021

View reviewed changes

facebook-github-bot closed this in 00d432a Jan 15, 2021

facebook-github-bot added the Merged label Jan 15, 2021

facebook-github-bot deleted the gh/ejguan/18/head branch January 19, 2021 15:16

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove optional for veiw_fn during View Tracking#50067

Remove optional for veiw_fn during View Tracking#50067
ejguan wants to merge 7 commits intogh/ejguan/18/basefrom
gh/ejguan/18/head

ejguan commented Jan 4, 2021 •

edited

Loading

Uh oh!

facebook-github-bot commented Jan 4, 2021 •

edited

Loading

Uh oh!

codecov Bot commented Jan 5, 2021 •

edited

Loading

Uh oh!

albanD left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ejguan commented Jan 13, 2021

Uh oh!

albanD commented Jan 14, 2021

Uh oh!

albanD left a comment

Uh oh!

facebook-github-bot commented Jan 15, 2021

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

ejguan commented Jan 4, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Nightly

Current

Compare

Uh oh!

facebook-github-bot commented Jan 4, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

💊 CI failures summary and remediations

Uh oh!

codecov Bot commented Jan 5, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

albanD left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

ejguan commented Jan 13, 2021

Uh oh!

albanD commented Jan 14, 2021

Uh oh!

albanD left a comment

Choose a reason for hiding this comment

Uh oh!

facebook-github-bot commented Jan 15, 2021

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

ejguan commented Jan 4, 2021 •

edited

Loading

facebook-github-bot commented Jan 4, 2021 •

edited

Loading

codecov Bot commented Jan 5, 2021 •

edited

Loading