Workaround for PyTorch issue 2575 · HadokenCode/docker-python@20adb40 · GitHub

HadokenCode / docker-python Public

forked from Kaggle/docker-python

• Notifications You must be signed in to change notification settings
• Fork 0
• Star 0

• Code
• Pull requests
• Actions
• Projects
• Wiki
• Security and quality
• Insights

Additional navigation options

Commit 20adb40

Browse filesBrowse the repository at this point in the historyBrowse files

nerdcha

committed

Workaround for PyTorch issue 2575

1 parent 5677cbe commit 20adb40
Copy full SHA for 20adb40

2 files changed

+18Lines changed: 18 additions & 0 deletions

File tree

Expand file treeCollapse file tree

Open diff view settings

Filter options

• patches
  • sitecustomize.py
• test_build.py

Expand file treeCollapse file tree

Open diff view settings

Collapse file

‎patches/sitecustomize.py‎

Copy file name to clipboardExpand all lines: patches/sitecustomize.py

+4Lines changed: 4 additions & 0 deletions

Original file line number	Diff line number	Diff line change
@@ -1,4 +1,8 @@
1	1	# Monkey patches BigQuery client creation to use proxy.
	2	+
	3	+ # Import torch before anything else. This is a hacky workaround to an error on dlopen
	4	+ # reporting a limit on static TLS, tracked in https://github.com/pytorch/pytorch/issues/2575
	5	+ import torch
2	6	import os
3	7
4	8	kaggle_proxy_token = os.getenv("KAGGLE_DATA_PROXY_TOKEN")

Collapse file

‎test_build.py‎

Copy file name to clipboardExpand all lines: test_build.py

+14Lines changed: 14 additions & 0 deletions

Original file line number	Diff line number	Diff line change
@@ -96,6 +96,20 @@
96	96	import seaborn
97	97	print("seaborn ok")
98	98
	99	+ # PyTorch smoke test based on http://pytorch.org/tutorials/beginner/nlp/deep_learning_tutorial.html
	100	+ import torch
	101	+ import torch.nn as tnn
	102	+ import torch.autograd as autograd
	103	+ torch.manual_seed(31337)
	104	+ linear_torch = tnn.Linear(5,3)
	105	+ data_torch = autograd.Variable(torch.randn(2, 5))
	106	+ print(linear_torch(data_torch))
	107	+ print("PyTorch ok")
	108	+
	109	+ import fastai
	110	+ from fastai.io import get_data
	111	+ print("fast.ai ok")
	112	+
99	113	import os
100	114	import threading
101	115	from http.server import BaseHTTPRequestHandler, HTTPServer

0 commit comments

Comments

0 (0)