Nothing Special   »   [go: up one dir, main page]
Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Port leaky_relu_backward to structured #58483

Closed
wants to merge 4 commits into from
Closed

Port leaky_relu_backward to structured #58483

wants to merge 4 commits into from

Conversation

Freey0
Copy link
Contributor
@Freey0 Freey0 commented May 18, 2021 •
edited
Loading

Stack from ghstack:

Differential Revision: D28572526

@Freey0 Freey0 requested a review from ezyang as a code owner May 18, 2021 15:49
@facebook-github-bot
Copy link
Contributor
facebook-github-bot commented May 18, 2021 •
edited
Loading

💊 CI failures summary and remediations

As of commit bff1e80 (more details on the Dr. CI page):

  • 3/3 failures introduced in this PR

🕵️ 3 new failures recognized by patterns

The following CI failures do not appear to be due to upstream breakages:

See CircleCI build pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_test2 (1/3)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

May 22 11:53:34 ERROR [4.728s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithSpawn) 
May 22 11:53:25   test_sparse_all_reduce_sum (__main__.TestDistBackendWithSpawn) ... ok (1.224s)
May 22 11:53:28   test_sparse_all_reduce_sum_cuda (__main__.TestDistBackendWithSpawn) ... ok (3.128s)
May 22 11:53:29   test_static_graph_api_cpu (__main__.TestDistBackendWithSpawn) ... ok (1.024s)
May 22 11:53:34   test_undefined_grad_parity_unused_parameters (__main__.TestDistBackendWithSpawn) ... /opt/conda/lib/python3.9/site-packages/torch/autograd/__init__.py:147: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 11:53:34   Variable._execution_engine.run_backward(
May 22 11:53:34 /opt/conda/lib/python3.9/site-packages/torch/autograd/__init__.py:147: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 11:53:34   Variable._execution_engine.run_backward(
May 22 11:53:34 ok (5.031s)
May 22 11:53:34 
May 22 11:53:34 ======================================================================
May 22 11:53:34 ERROR [4.728s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithSpawn)
May 22 11:53:34 ----------------------------------------------------------------------
May 22 11:53:34 Traceback (most recent call last):
May 22 11:53:34   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 391, in wrapper
May 22 11:53:34     self._join_processes(fn)
May 22 11:53:34   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 583, in _join_processes
May 22 11:53:34     self._check_return_codes(elapsed_time)
May 22 11:53:34   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 626, in _check_return_codes
May 22 11:53:34     raise RuntimeError(error)
May 22 11:53:34 RuntimeError: Process 0 exited with error code 10 and exception:
May 22 11:53:34 Traceback (most recent call last):

See CircleCI build pytorch_linux_xenial_cuda11_1_cudnn8_py3_gcc7_test2 (2/3)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

May 22 12:36:12 ERROR [5.430s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithSpawn) 
May 22 12:36:01   test_sparse_all_reduce_sum (__main__.TestDistBackendWithSpawn) ... ok (1.325s)
May 22 12:36:04   test_sparse_all_reduce_sum_cuda (__main__.TestDistBackendWithSpawn) ... ok (3.530s)
May 22 12:36:06   test_static_graph_api_cpu (__main__.TestDistBackendWithSpawn) ... ok (1.327s)
May 22 12:36:11   test_undefined_grad_parity_unused_parameters (__main__.TestDistBackendWithSpawn) ... /opt/conda/lib/python3.6/site-packages/torch/autograd/__init__.py:149: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 12:36:11   allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag
May 22 12:36:11 /opt/conda/lib/python3.6/site-packages/torch/autograd/__init__.py:149: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 12:36:11   allow_unreachable=True, accumulate_grad=True)  # allow_unreachable flag
May 22 12:36:12 ok (5.734s)
May 22 12:36:12 
May 22 12:36:12 ======================================================================
May 22 12:36:12 ERROR [5.430s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithSpawn)
May 22 12:36:12 ----------------------------------------------------------------------
May 22 12:36:12 Traceback (most recent call last):
May 22 12:36:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 391, in wrapper
May 22 12:36:12     self._join_processes(fn)
May 22 12:36:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 583, in _join_processes
May 22 12:36:12     self._check_return_codes(elapsed_time)
May 22 12:36:12   File "/opt/conda/lib/python3.6/site-packages/torch/testing/_internal/common_distributed.py", line 626, in _check_return_codes
May 22 12:36:12     raise RuntimeError(error)
May 22 12:36:12 RuntimeError: Process 0 exited with error code 10 and exception:
May 22 12:36:12 Traceback (most recent call last):

See CircleCI build pytorch_linux_bionic_cuda10_2_cudnn7_py3_9_gcc7_test1 (3/3)

Step: "Run tests" (full log | diagnosis details | 🔁 rerun)

May 22 11:41:08 ERROR [3.824s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithFork) 
May 22 11:41:01   test_sparse_all_reduce_sum (__main__.TestDistBackendWithFork) ... ok (0.220s)
May 22 11:41:04   test_sparse_all_reduce_sum_cuda (__main__.TestDistBackendWithFork) ... ok (2.223s)
May 22 11:41:04   test_static_graph_api_cpu (__main__.TestDistBackendWithFork) ... ok (0.118s)
May 22 11:41:07   test_undefined_grad_parity_unused_parameters (__main__.TestDistBackendWithFork) ... /opt/conda/lib/python3.9/site-packages/torch/autograd/__init__.py:147: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 11:41:07   Variable._execution_engine.run_backward(
May 22 11:41:07 /opt/conda/lib/python3.9/site-packages/torch/autograd/__init__.py:147: UserWarning: find_unused_parameters=True was specified in DDP constructor, but did not find any unused parameters in the forward pass. This flag results in an extra traversal of the autograd graph every iteration,  which can adversely affect performance. If your model indeed never has any unused parameters in the forward pass, consider turning this flag off. Note that this warning may be a false positive if your model has flow control causing later iterations to have unused parameters. (Triggered internally at  /var/lib/jenkins/workspace/torch/lib/c10d/reducer.cpp:1168.)
May 22 11:41:07   Variable._execution_engine.run_backward(
May 22 11:41:08 ok (3.926s)
May 22 11:41:08 
May 22 11:41:08 ======================================================================
May 22 11:41:08 ERROR [3.824s]: test_ddp_hook_parity_powerSGD (__main__.TestDistBackendWithFork)
May 22 11:41:08 ----------------------------------------------------------------------
May 22 11:41:08 Traceback (most recent call last):
May 22 11:41:08   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 391, in wrapper
May 22 11:41:08     self._join_processes(fn)
May 22 11:41:08   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 583, in _join_processes
May 22 11:41:08     self._check_return_codes(elapsed_time)
May 22 11:41:08   File "/opt/conda/lib/python3.9/site-packages/torch/testing/_internal/common_distributed.py", line 626, in _check_return_codes
May 22 11:41:08     raise RuntimeError(error)
May 22 11:41:08 RuntimeError: Process 0 exited with error code 10 and exception:
May 22 11:41:08 Traceback (most recent call last):
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.

Click here to manually regenerate this comment.

This was referenced May 18, 2021
Closed
Closed
This was referenced May 20, 2021
Closed
Closed
Closed
Closed
Closed
Closed
@ezyang
Copy link
Contributor
ezyang commented May 20, 2021

@ezyang has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@ezyang
Copy link
Contributor
ezyang commented May 24, 2021

@ezyang has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

@facebook-github-bot
Copy link
Contributor

@ezyang merged this pull request in 3f0b7e0.

@facebook-github-bot facebook-github-bot deleted the gh/Feey0/18/head branch May 28, 2021 14:18
deniskokarev pushed a commit to deniskokarev/pytorch that referenced this pull request Jun 9, 2021
@Freey0 @deniskokarev
Port leaky_relu_backward to structured (pytorch#58483)
1f116eb 
Summary: Pull Request resolved: pytorch#58483

Test Plan: Imported from OSS

Reviewed By: soulitzer

Differential Revision: D28572526

Pulled By: ezyang

fbshipit-source-id: a73bdf06967687dbb1d4fbb0f2ca80115db57a07
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@ezyang ezyang ezyang approved these changes

Assignees
No one assigned
Labels
cla signed Merged open source
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@Freey0 @facebook-github-bot @ezyang @pytorchbot