MCPcopy Index your code
hub / github.com/pytorch/pytorch / setUpModule

Function setUpModule

caffe2/python/caffe_translator_test.py:27–55  ·  view source on GitHub ↗
()

Source from the content-addressed store, hash-verified

25import unittest
26
27def setUpModule():
28 # Do nothing if caffe and test data is not found
29 if not (CAFFE_FOUND and os.path.exists('data/testdata/caffe_translator')):
30 return
31 # We will do all the computation stuff in the global space.
32 caffenet = caffe_pb2.NetParameter()
33 caffenet_pretrained = caffe_pb2.NetParameter()
34 with open('data/testdata/caffe_translator/deploy.prototxt') as f:
35 text_format.Merge(f.read(), caffenet)
36 with open('data/testdata/caffe_translator/'
37 'bvlc_reference_caffenet.caffemodel') as f:
38 caffenet_pretrained.ParseFromString(f.read())
39 for remove_legacy_pad in [True, False]:
40 net, pretrained_params = caffe_translator.TranslateModel(
41 caffenet, caffenet_pretrained, is_test=True,
42 remove_legacy_pad=remove_legacy_pad
43 )
44 with open('data/testdata/caffe_translator/'
45 'bvlc_reference_caffenet.translatedmodel',
46 'w') as fid:
47 fid.write(str(net))
48 for param in pretrained_params.protos:
49 workspace.FeedBlob(param.name, utils.Caffe2TensorToNumpyArray(param))
50 # Let's also feed in the data from the Caffe test code.
51 data = np.load('data/testdata/caffe_translator/data_dump.npy').astype(
52 np.float32)
53 workspace.FeedBlob('data', data)
54 # Actually running the test.
55 workspace.RunNetOnce(net.SerializeToString())
56
57
58@unittest.skipIf(not CAFFE_FOUND,

Callers

nothing calls this directly

Calls 5

TranslateModelMethod · 0.80
astypeMethod · 0.80
readMethod · 0.45
writeMethod · 0.45
loadMethod · 0.45

Tested by

no test coverage detected

Used in the wild real call sites across dependent graphs

searching dependent graphs…