@@ -31,7 +31,7 @@ class Function: | |||
self.y = y | |||
return y | |||
def backward(self. output_grads): | |||
def backward(self, output_grads): | |||
y = self.y | |||
return output_grads * y * (1-y) | |||
@@ -194,9 +194,9 @@ class Compose(VisionTransform): | |||
will be random shuffled, the 2nd and 4th transform will also be shuffled. | |||
:param order: The same with :class:`VisionTransform` | |||
Example: | |||
Examples: | |||
..testcode:: | |||
.. testcode:: | |||
from megengine.data.transform import RandomHorizontalFlip, RandomVerticalFlip, CenterCrop, ToMode, Compose | |||
@@ -197,8 +197,8 @@ def sqrt(inp: Tensor) -> Tensor: | |||
.. testoutput:: | |||
[[0. 1. 1.4142] | |||
[1.7321 2. 2.2361 ]] | |||
[[0. 1. 1.4142] | |||
[1.7321 2. 2.2361]] | |||
""" | |||
return inp ** 0.5 | |||
@@ -227,8 +227,8 @@ def square(inp: Tensor) -> Tensor: | |||
.. testoutput:: | |||
[[0. 1. 4.] | |||
[9. 16. 25.]] | |||
[[ 0. 1. 4.] | |||
[ 9. 16. 25.]] | |||
""" | |||
return inp ** 2 | |||
@@ -437,7 +437,7 @@ def clamp(inp: Tensor, lower=None, upper=None) -> Tensor: | |||
:param lower: lower-bound of the range to be clamped to | |||
:param upper: upper-bound of the range to be clamped to | |||
Example: | |||
Examples: | |||
.. testcode:: | |||
@@ -452,6 +452,8 @@ def clamp(inp: Tensor, lower=None, upper=None) -> Tensor: | |||
print(F.clamp(a, upper=3).numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[2 2 2 3 4] | |||
@@ -58,6 +58,8 @@ def isnan(inp: Tensor) -> Tensor: | |||
print(F.isnan(x).numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[False True False] | |||
@@ -83,6 +85,8 @@ def isinf(inp: Tensor) -> Tensor: | |||
print(F.isinf(x).numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[False True False] | |||
@@ -141,7 +145,9 @@ def sum( | |||
data = tensor(np.arange(1, 7, dtype=np.int32).reshape(2, 3)) | |||
out = F.sum(data) | |||
print(out.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[21] | |||
@@ -208,6 +214,8 @@ def mean( | |||
out = F.mean(data) | |||
print(out.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[3.5] | |||
@@ -250,9 +258,11 @@ def var( | |||
out = F.var(data) | |||
print(out.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[2.9166667] | |||
[2.9167] | |||
""" | |||
if axis is None: | |||
m = mean(inp, axis=axis, keepdims=False) | |||
@@ -288,9 +298,11 @@ def std( | |||
out = F.std(data, axis=1) | |||
print(out.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[0.8164966 0.8164966] | |||
[0.8165 0.8165] | |||
""" | |||
return var(inp, axis=axis, keepdims=keepdims) ** 0.5 | |||
@@ -354,6 +366,8 @@ def max( | |||
y = F.max(x) | |||
print(y.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[6] | |||
@@ -388,9 +402,11 @@ def norm( | |||
y = F.norm(x) | |||
print(y.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[4.358899] | |||
[4.3589] | |||
""" | |||
if p == 0: | |||
@@ -426,6 +442,8 @@ def argmin( | |||
y = F.argmin(x) | |||
print(y.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[0] | |||
@@ -479,6 +497,8 @@ def argmax( | |||
x = tensor(np.arange(1, 7, dtype=np.int32).reshape(2,3)) | |||
y = F.argmax(x) | |||
print(y.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
@@ -372,10 +372,12 @@ def softplus(inp: Tensor) -> Tensor: | |||
x = tensor(np.arange(-3, 3, dtype=np.float32)) | |||
y = F.softplus(x) | |||
print(y.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
.. output:: | |||
[0.04858735 0.126928 0.3132617 0.6931472 1.3132617 2.126928 ] | |||
[0.0486 0.1269 0.3133 0.6931 1.3133 2.1269] | |||
""" | |||
return log1p(exp(-abs(inp))) + relu(inp) | |||
@@ -411,10 +413,12 @@ def log_softmax(inp: Tensor, axis: Union[int, Sequence[int]]) -> Tensor: | |||
y = F.log_softmax(x, axis=1) | |||
print(y.numpy()) | |||
.. output:: | |||
Outputs: | |||
.. testoutput:: | |||
[[-4.4519143 -3.4519143 -2.4519143 -1.4519144 -0.4519144] | |||
[-4.4519143 -3.4519143 -2.4519143 -1.4519144 -0.4519144]] | |||
[[-4.4519 -3.4519 -2.4519 -1.4519 -0.4519] | |||
[-4.4519 -3.4519 -2.4519 -1.4519 -0.4519]] | |||
""" | |||
return inp - logsumexp(inp, axis, keepdims=True) | |||
@@ -432,6 +436,7 @@ def logsigmoid(inp: Tensor) -> Tensor: | |||
:param inp: The input tensor | |||
Examples: | |||
.. testcode:: | |||
import numpy as np | |||
@@ -442,9 +447,12 @@ def logsigmoid(inp: Tensor) -> Tensor: | |||
y = F.logsigmoid(x) | |||
print(y.numpy()) | |||
.. output:: | |||
Outputs: | |||
.. testoutput:: | |||
[-5.0067153 -4.01815 -3.0485873 -2.126928 -1.3132617 -0.6931472 -0.3132617 -0.126928 -0.04858735 -0.01814993] | |||
[-5.0067 -4.0181 -3.0486 -2.1269 -1.3133 -0.6931 -0.3133 -0.1269 -0.0486 | |||
-0.0181] | |||
""" | |||
return -softplus(-inp) | |||
@@ -478,6 +486,7 @@ def logsumexp( | |||
:param keepdims: whether to retain :attr:`axis` or not for the output tensor. | |||
Examples: | |||
.. testcode:: | |||
import numpy as np | |||
@@ -488,9 +497,11 @@ def logsumexp( | |||
y = F.logsumexp(x, axis=1, keepdims=False) | |||
print(y.numpy()) | |||
.. output:: | |||
Outputs: | |||
.. testoutput:: | |||
[-0.5480856 4.4519143] | |||
[-0.5481 4.4519] | |||
""" | |||
max_value = max(inp, axis, keepdims=True) | |||
@@ -577,8 +588,9 @@ def softmax(inp: Tensor, axis: Optional[int] = None) -> Tensor: | |||
Outputs: | |||
.. testoutput:: | |||
[[0.01165623 0.03168492 0.08612854 0.23412167 0.6364086 ] | |||
[0.01165623 0.03168492 0.08612854 0.23412167 0.6364086 ]] | |||
[[0.0117 0.0317 0.0861 0.2341 0.6364] | |||
[0.0117 0.0317 0.0861 0.2341 0.6364]] | |||
""" | |||
if axis is None: | |||
@@ -1026,7 +1038,7 @@ def dot(inp1: Tensor, inp2: Tensor) -> Tensor: | |||
Examples: | |||
.. teestcode:: | |||
.. testcode:: | |||
import numpy as np | |||
from megengine import tensor | |||
@@ -1039,9 +1051,10 @@ def dot(inp1: Tensor, inp2: Tensor) -> Tensor: | |||
Outputs: | |||
.. testoutput:: | |||
[55.] | |||
.. testoutputs:: | |||
""" | |||
op = builtin.Dot() | |||
inp1, inp2 = utils.convert_inputs(inp1, inp2) | |||
@@ -1058,7 +1071,7 @@ def svd(inp: Tensor, full_matrices=False, compute_uv=True) -> Tensor: | |||
Examples: | |||
.. teestcode:: | |||
.. testcode:: | |||
import numpy as np | |||
from megengine import tensor | |||
@@ -1070,7 +1083,9 @@ def svd(inp: Tensor, full_matrices=False, compute_uv=True) -> Tensor: | |||
Outputs: | |||
[7.348, 1.] | |||
.. testoutput:: | |||
[7.3485 1. ] | |||
""" | |||
op = builtin.SVD(full_matrices=full_matrices, compute_uv=compute_uv) | |||
@@ -1445,6 +1460,8 @@ def indexing_one_hot( | |||
val = F.indexing_one_hot(src, index) | |||
print(val.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[1.] | |||
@@ -60,7 +60,7 @@ __all__ = [ | |||
] | |||
def eye(n: int, *, dtype=None, device: Optional[CompNode] = None) -> Tensor: | |||
def eye(n: int, *, dtype="float32", device: Optional[CompNode] = None) -> Tensor: | |||
""" | |||
Returns a 2D tensor with ones on the diagonal and zeros elsewhere. | |||
@@ -80,7 +80,7 @@ def eye(n: int, *, dtype=None, device: Optional[CompNode] = None) -> Tensor: | |||
data_shape = (4, 6) | |||
n, m = data_shape | |||
out = F.eye(n, m, dtype=np.float32) | |||
out = F.eye([n, m], dtype=np.float32) | |||
print(out.numpy()) | |||
Outputs: | |||
@@ -135,6 +135,8 @@ def zeros_like(inp: Tensor) -> Tensor: | |||
out = F.zeros_like(inp) | |||
print(out.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[[0 0 0] | |||
@@ -638,7 +640,7 @@ def cond_take(mask: Tensor, x: Tensor) -> Tensor: | |||
.. testoutput:: | |||
Tensor([1. 4.]) Tensor([0 3], dtype=int32) | |||
[1. 4.] [0 3] | |||
""" | |||
if not isinstance(x, (TensorWrapperBase, TensorBase)): | |||
@@ -888,6 +890,8 @@ def linspace( | |||
a = F.linspace(3,10,5) | |||
print(a.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[ 3. 4.75 6.5 8.25 10. ] | |||
@@ -930,6 +934,8 @@ def arange( | |||
a = F.arange(5) | |||
print(a.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
@@ -977,7 +983,9 @@ def param_pack_split(inp: Tensor, offsets: List, shapes: List) -> Tensor: | |||
b, c = F.param_pack_split(a, [0, 1, 1, 10], [(1,), (3, 3)]) | |||
print(b.numpy()) | |||
print(c.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[1] | |||
@@ -1000,7 +1008,7 @@ def param_pack_concat(inps: List, offsets: Tensor, offsets_val: List) -> Tensor: | |||
:param offsets: device value of offsets | |||
:param offsets_val: offsets of inputs, length of 2 * n, | |||
format [begin0, end0, begin1, end1]. | |||
:return: split tensors | |||
:return: concat tensors | |||
Examples: | |||
@@ -1013,10 +1021,12 @@ def param_pack_concat(inps: List, offsets: Tensor, offsets_val: List) -> Tensor: | |||
a = tensor(np.ones((1,), np.int32)) | |||
b = tensor(np.ones((3, 3), np.int32)) | |||
offsets_val = [0, 1, 1, 10] | |||
offsets = tensor(offsets, np.int32) | |||
offsets = tensor(offsets_val, np.int32) | |||
c = F.param_pack_concat([a, b], offsets, offsets_val) | |||
print(c.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
[1 1 1 1 1 1 1 1 1 1] | |||
@@ -63,19 +63,6 @@ def accuracy( | |||
return accs | |||
def zero_grad(inp: Tensor) -> Tensor: | |||
r""" | |||
Returns a tensor which is treated as constant during backward gradient calcuation, | |||
i.e. its gradient is zero. | |||
:param inp: Input tensor. | |||
See implementation of :func:`~.softmax` for example. | |||
""" | |||
print("zero_grad is obsoleted, please use detach instead") | |||
raise NotImplementedError | |||
def copy(inp, cn): | |||
r""" | |||
Copy tensor to another device. | |||
@@ -219,7 +219,7 @@ class LeakyReLU(Module): | |||
.. testoutput:: | |||
[-0.08 -0.12 6. 10. ] | |||
[-0.08 -0.12 6. 10. ] | |||
""" | |||
@@ -267,15 +267,17 @@ class BatchNorm2d(_BatchNorm): | |||
m = M.BatchNorm2d(4) | |||
inp = mge.tensor(np.random.rand(1, 4, 3, 3).astype("float32")) | |||
oup = m(inp) | |||
print(m.weight, m.bias) | |||
print(m.weight.numpy(), m.bias.numpy()) | |||
# Without Learnable Parameters | |||
m = M.BatchNorm2d(4, affine=False) | |||
oup = m(inp) | |||
print(m.weight, m.bias) | |||
Outputs: | |||
.. testoutput:: | |||
Tensor([1. 1. 1. 1.]) Tensor([0. 0. 0. 0.]) | |||
[1. 1. 1. 1.] [0. 0. 0. 0.] | |||
None None | |||
""" | |||
@@ -17,23 +17,25 @@ class Sequential(Module): | |||
Alternatively, an ordered dict of modules can also be passed in. | |||
To make it easier to understand, here is a small example: | |||
Examples: | |||
.. testcode:: | |||
import numpy as np | |||
import megengine.nn as nn | |||
import megengine.nn.functional as F | |||
from megengine import tensor | |||
import megengine.functional as F | |||
batch_size = 64 | |||
data = nn.Input("data", shape=(batch_size, 1, 28, 28), dtype=np.float32, value=np.zeros((batch_size, 1, 28, 28))) | |||
label = nn.Input("label", shape=(batch_size,), dtype=np.int32, value=np.zeros(batch_size,)) | |||
data = tensor(np.zeros((batch_size, 1, 28, 28)), dtype=np.float32) | |||
label = tensor(np.zeros(batch_size,), dtype=np.int32) | |||
data = data.reshape(batch_size, -1) | |||
net = nn.Sequential( | |||
nn.Linear(28 * 28, 320), | |||
nn.Linear(320, 500), | |||
nn.Linear(500, 320), | |||
nn.Linear(320, 10) | |||
net = M.Sequential( | |||
M.Linear(28 * 28, 320), | |||
M.Linear(320, 500), | |||
M.Linear(500, 320), | |||
M.Linear(320, 10) | |||
) | |||
pred = net(data) | |||
@@ -37,7 +37,9 @@ def normal( | |||
x = rand.normal(mean=0, std=1, size=(2, 2)) | |||
print(x.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
:options: +SKIP | |||
@@ -73,7 +75,9 @@ def uniform( | |||
x = rand.uniform(size=(2, 2)) | |||
print(x.numpy()) | |||
Outputs: | |||
.. testoutput:: | |||
:options: +SKIP | |||