leezu opened a new issue #19081:
URL: https://github.com/apache/incubator-mxnet/issues/19081
```
[2020-09-02T00:02:22.521Z] ================================== FAILURES
===================================
[2020-09-02T00:02:22.521Z] ________________________________ test_np_round
________________________________
[2020-09-02T00:02:22.521Z] [gw3] win32 -- Python 3.7.3 C:\Python37\python.exe
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] @with_seed()
[2020-09-02T00:02:22.521Z] @use_np
[2020-09-02T00:02:22.521Z] def test_np_round():
[2020-09-02T00:02:22.521Z] class TestRound(HybridBlock):
[2020-09-02T00:02:22.521Z] def __init__(self, func, decimals):
[2020-09-02T00:02:22.521Z] super(TestRound, self).__init__()
[2020-09-02T00:02:22.521Z] self.func = func
[2020-09-02T00:02:22.521Z] self.decimals = decimals
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] def hybrid_forward(self, F, x):
[2020-09-02T00:02:22.521Z] return getattr(F.np,
self.func)(x, self.decimals)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] shapes = [(), (1, 2, 3), (1, 0)]
[2020-09-02T00:02:22.521Z] types = ['int32', 'int64', 'float32',
'float64']
[2020-09-02T00:02:22.521Z] funcs = ['round', 'round_']
[2020-09-02T00:02:22.521Z] for hybridize, oneType, func in
itertools.product([True, False], types, funcs):
[2020-09-02T00:02:22.521Z] rtol, atol = 1e-3, 1e-5
[2020-09-02T00:02:22.521Z] for shape in shapes:
[2020-09-02T00:02:22.521Z] for d in range(-5, 6):
[2020-09-02T00:02:22.521Z] test_round = TestRound(func,
d)
[2020-09-02T00:02:22.521Z] if hybridize:
[2020-09-02T00:02:22.521Z] test_round.hybridize()
[2020-09-02T00:02:22.521Z] x = rand_ndarray(shape,
dtype=oneType).as_np_ndarray()
[2020-09-02T00:02:22.521Z] np_out = getattr(_np,
func)(x.asnumpy(), d)
[2020-09-02T00:02:22.521Z] mx_out = test_round(x)
[2020-09-02T00:02:22.521Z] assert mx_out.shape ==
np_out.shape
[2020-09-02T00:02:22.521Z] >
assert_almost_equal(mx_out.asnumpy(), np_out, rtol=rtol, atol=atol)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] tests\python\unittest\test_numpy_op.py:7691:
[2020-09-02T00:02:22.521Z] _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] a = array([[[ 1.029 , -0.371 , 1.11 ],
[2020-09-02T00:02:22.521Z] [ 1.4289999, -0.636 , -0.306 ]]],
dtype=float32)
[2020-09-02T00:02:22.521Z] b = array([[[ 1.029, -0.371, 1.11 ],
[2020-09-02T00:02:22.521Z] [ 1.429, -0.636, -0.307]]], dtype=float32)
[2020-09-02T00:02:22.521Z] rtol = 0.001, atol = 1e-05, names = ('a', 'b'),
equal_nan = False
[2020-09-02T00:02:22.521Z] use_broadcast = True, mismatches = (10, 10)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] def assert_almost_equal(a, b, rtol=None,
atol=None, names=('a', 'b'), equal_nan=False,
[2020-09-02T00:02:22.521Z] use_broadcast=True,
mismatches=(10, 10)):
[2020-09-02T00:02:22.521Z] """Test that two numpy arrays are almost
equal. Raise exception message if not.
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] Parameters
[2020-09-02T00:02:22.521Z] ----------
[2020-09-02T00:02:22.521Z] a : np.ndarray or mx.nd.array
[2020-09-02T00:02:22.521Z] b : np.ndarray or mx.nd.array
[2020-09-02T00:02:22.521Z] rtol : None or float or dict of dtype ->
float
[2020-09-02T00:02:22.521Z] The relative threshold. Default
threshold will be used if set to ``None``.
[2020-09-02T00:02:22.521Z] atol : None or float or dict of dtype ->
float
[2020-09-02T00:02:22.521Z] The absolute threshold. Default
threshold will be used if set to ``None``.
[2020-09-02T00:02:22.521Z] names : tuple of names, optional
[2020-09-02T00:02:22.521Z] The names used in error message when
an exception occurs
[2020-09-02T00:02:22.521Z] equal_nan : boolean, optional
[2020-09-02T00:02:22.521Z] The flag determining how to treat NAN
values in comparison
[2020-09-02T00:02:22.521Z] mismatches : tuple of mismatches
[2020-09-02T00:02:22.521Z] Maximum number of mismatches to be
printed (mismatches[0]) and determine (mismatches[1])
[2020-09-02T00:02:22.521Z] """
[2020-09-02T00:02:22.521Z] if not use_broadcast:
[2020-09-02T00:02:22.521Z] checkShapes(a, b)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] rtol, atol = get_tols(a, b, rtol, atol)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] if isinstance(a, mx.numpy.ndarray):
[2020-09-02T00:02:22.521Z] a = a.asnumpy()
[2020-09-02T00:02:22.521Z] if isinstance(b, mx.numpy.ndarray):
[2020-09-02T00:02:22.521Z] b = b.asnumpy()
[2020-09-02T00:02:22.521Z] use_np_allclose = isinstance(a,
np.ndarray) and isinstance(b, np.ndarray)
[2020-09-02T00:02:22.521Z] if not use_np_allclose:
[2020-09-02T00:02:22.521Z] if not (hasattr(a, 'ctx') and
hasattr(b, 'ctx') and a.ctx == b.ctx and a.dtype == b.dtype):
[2020-09-02T00:02:22.521Z] use_np_allclose = True
[2020-09-02T00:02:22.521Z] if isinstance(a, mx.nd.NDArray):
[2020-09-02T00:02:22.521Z] a = a.asnumpy()
[2020-09-02T00:02:22.521Z] if isinstance(b, mx.nd.NDArray):
[2020-09-02T00:02:22.521Z] b = b.asnumpy()
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] if use_np_allclose:
[2020-09-02T00:02:22.521Z] if hasattr(a, 'dtype') and a.dtype ==
np.bool_ and hasattr(b, 'dtype') and b.dtype == np.bool_:
[2020-09-02T00:02:22.521Z] np.testing.assert_equal(a, b)
[2020-09-02T00:02:22.521Z] return
[2020-09-02T00:02:22.521Z] if almost_equal(a, b, rtol, atol,
equal_nan=equal_nan):
[2020-09-02T00:02:22.521Z] return
[2020-09-02T00:02:22.521Z] else:
[2020-09-02T00:02:22.521Z] output = mx.nd.contrib.allclose(a, b,
rtol, atol, equal_nan)
[2020-09-02T00:02:22.521Z] if output.asnumpy() == 1:
[2020-09-02T00:02:22.521Z] return
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] a = a.asnumpy()
[2020-09-02T00:02:22.521Z] b = b.asnumpy()
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] index, rel = _find_max_violation(a, b,
rtol, atol)
[2020-09-02T00:02:22.521Z] if index != ():
[2020-09-02T00:02:22.521Z] # a, b are the numpy arrays
[2020-09-02T00:02:22.521Z] indexErr = index
[2020-09-02T00:02:22.521Z] relErr = rel
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] print('\n*** Maximum errors for
vector of size {}: rtol={}, atol={}\n'.format(a.size, rtol, atol))
[2020-09-02T00:02:22.521Z] aTmp = a.copy()
[2020-09-02T00:02:22.521Z] bTmp = b.copy()
[2020-09-02T00:02:22.521Z] i = 1
[2020-09-02T00:02:22.521Z] while i <= a.size:
[2020-09-02T00:02:22.521Z] if i <= mismatches[0]:
[2020-09-02T00:02:22.521Z] print("%3d: Error %f %s"
%(i, rel, locationError(a, b, index, names)))
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] aTmp[index] = bTmp[index] = 0
[2020-09-02T00:02:22.521Z] if almost_equal(aTmp, bTmp, rtol,
atol, equal_nan=equal_nan):
[2020-09-02T00:02:22.521Z] break
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] i += 1
[2020-09-02T00:02:22.521Z] if i <= mismatches[1] or
mismatches[1] <= 0:
[2020-09-02T00:02:22.521Z] index, rel =
_find_max_violation(aTmp, bTmp, rtol, atol)
[2020-09-02T00:02:22.521Z] else:
[2020-09-02T00:02:22.521Z] break
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] mismatchDegree = "at least " if
mismatches[1] > 0 and i > mismatches[1] else ""
[2020-09-02T00:02:22.521Z] errMsg = "Error %f exceeds tolerance
rtol=%e, atol=%e (mismatch %s%f%%).\n%s" % \
[2020-09-02T00:02:22.521Z] (relErr, rtol, atol,
mismatchDegree, 100*i/a.size, \
[2020-09-02T00:02:22.521Z] locationError(a, b,
indexErr, names, maxError=True))
[2020-09-02T00:02:22.521Z] else:
[2020-09-02T00:02:22.521Z] errMsg = "Error %f exceeds tolerance
rtol=%e, atol=%e.\n" % (rel, rtol, atol)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] np.set_printoptions(threshold=4,
suppress=True)
[2020-09-02T00:02:22.521Z] msg = npt.build_err_msg([a, b],
err_msg=errMsg)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] > raise AssertionError(msg)
[2020-09-02T00:02:22.521Z] E AssertionError:
[2020-09-02T00:02:22.521Z] E Items are not equal:
[2020-09-02T00:02:22.521Z] E Error 3.154627 exceeds tolerance
rtol=1.000000e-03, atol=1.000000e-05 (mismatch 16.666667%).
[2020-09-02T00:02:22.521Z] E Location of maximum error: (0, 1, 2),
a=-0.30599999, b=-0.30700001
[2020-09-02T00:02:22.521Z] E ACTUAL: array([[[ 1.029 , -0.371 ,
1.11 ],
[2020-09-02T00:02:22.521Z] E [ 1.4289999, -0.636 , -0.306
]]], dtype=float32)
[2020-09-02T00:02:22.521Z] E DESIRED: array([[[ 1.029, -0.371, 1.11
],
[2020-09-02T00:02:22.521Z] E [ 1.429, -0.636, -0.307]]],
dtype=float32)
[2020-09-02T00:02:22.521Z]
[2020-09-02T00:02:22.521Z] windows_package\python\mxnet\test_utils.py:735:
AssertionError
```
https://jenkins.mxnet-ci.amazon-ml.com/blue/organizations/jenkins/mxnet-validation%2Fwindows-gpu/detail/PR-19034/5/pipeline/
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
