本文整理汇总了Python中tvm.relay.add函数的典型用法代码示例。如果您正苦于以下问题：Python add函数的具体用法？Python add怎么用？Python add使用的例子？那么恭喜您, 这里精选的函数代码示例或许可以为您提供帮助。
在下文中一共展示了add函数的15个代码示例，这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞，您的评价将有助于系统推荐出更棒的Python代码示例。
示例1: gen_intermediate_tuple
def gen_intermediate_tuple(x):
y1 = relay.add(x, relay.const(1, "float32"))
y2 = relay.add(x, relay.const(1, "float32"))
y3 = relay.add(x, relay.const(1, "float32"))
concat = relay.concatenate((y1, y2, y3), axis=1)
out = relay.add(concat, relay.const(1, "float32"))
return out示例2: before
def before():
c = relay.const(c_data)
x = relay.var("x")
y = relay.Tuple([x, c])
z = relay.add(y[1], c)
z = relay.add(z, y[0])
return relay.Function([x], z)示例3: before
def before(x):
concat = gen_consecutive_tuple(x)
pooled = relay.nn.max_pool2d(concat, pool_size=(2, 2), strides=(2, 2), padding=(0, 0))
out = relay.add(pooled, relay.const(1, "float32"))
out2 = relay.add(out, relay.const(1, "float32"))
out_tup = relay.Tuple((out, out2))
return relay.Function(relay.ir_pass.free_vars(out_tup), out_tup)示例4: expected
def expected():
x = relay.var("x", shape=(1, 16))
y = relay.nn.relu(x)
y = relay.add(y, relay.const(1.0, "float32"))
y = relay.add(y, y)
f = relay.Function([x], y)
return f示例5: convnet
def convnet():
"""Alternating layout of simple convnet (from image super-resolution).
"""
bias1 = relay.var('bias1', shape=(64,))
bias2 = relay.var('bias2', shape=(64,))
bias3 = relay.var('bias3', shape=(64,))
bias4 = relay.var('bias4', shape=(64,))
weight1 = relay.var('weight1', shape=(64, 1, 5, 5))
weight2 = relay.var('weight2', shape=(64, 64, 3, 3))
weight3 = relay.var('weight3', shape=(64, 64, 3, 3))
weight4 = relay.var('weight4', shape=(64, 64, 3, 3))
data = relay.var("x", shape=(1, 1, 224, 224))
n00 = relay.nn.conv2d(data, weight1, padding=[2, 2], kernel_size=[5, 5])
n01 = relay.expand_dims(bias1, axis=1, num_newaxis=2)
n02 = relay.add(n00, n01)
n03 = relay.nn.relu(n02)
n04 = relay.nn.conv2d(n03, weight2, padding=[1, 1], kernel_size=[3, 3])
n05 = relay.expand_dims(bias2, axis=1, num_newaxis=2)
n06 = relay.add(n04, n05)
n07 = relay.nn.relu(n06)
n08 = relay.nn.conv2d(n07, weight3, padding=[1, 1], kernel_size=[3, 3])
n09 = relay.expand_dims(bias3, axis=1, num_newaxis=2)
n10 = relay.add(n08, n09)
n11 = relay.nn.relu(n10)
n12 = relay.nn.conv2d(n11, weight4, padding=[1, 1], kernel_size=[3, 3])
n13 = relay.expand_dims(bias4, axis=1, num_newaxis=2)
n14 = relay.add(n12, n13)
n15 = relay.reshape(n14, newshape=[1, 1, 3, 3, 224, 224])
n16 = relay.transpose(n15, axes=[0, 1, 4, 2, 5, 3])
net = relay.reshape(n16, newshape=[1, 1, 672, 672])
args = relay.ir_pass.free_vars(net)
return relay.Function(args, net)示例6: before
def before():
x = relay.var("x", shape=(1, 16))
y1 = relay.nn.relu(x)
y2 = relay.nn.relu(x)
y1 = relay.add(y1, relay.const(1.0, "float32"))
y2 = relay.add(y2, relay.const(1.0, "float32"))
y = relay.add(y1, y2)
f = relay.Function([x], y)
return f示例7: test_func
def test_func():
x = relay.var("x", shape=(3, 2))
y = relay.var("y")
one = relay.const(10e10, dtype="float32")
z = relay.add(x, one)
z = relay.add(z, z)
f = relay.Function([x, y], z)
show(z.astext())
show(f.astext())示例8: test_env
def test_env():
x = relay.var("x", "float32")
y = relay.var("y", "float32")
z = relay.add(x, y)
z = relay.add(z, z)
f = relay.Function([x, y], z)
env = relay.Module()
env["myf"] = f
text = env.astext()
assert "def @myf" in text
assert "%1 = add(%0, %0) # ty=float32" in text
show(env.astext(annotate=lambda x: str(x.checked_type.dtype)))
show(text)示例9: test_bind_params
def test_bind_params():
x = relay.var("x")
y = relay.var("y")
z = relay.add(x, y)
f = relay.Function([x, y], z)
fbinded = relay.bind(f, {x : relay.const(1, "float32")})
fexpected =relay.Function(
[y],
relay.add(relay.const(1, "float32"), y))
assert relay.ir_pass.alpha_equal(fbinded, fexpected)
zbinded = relay.bind(z, {y: x})
zexpected = relay.add(x, x)
assert relay.ir_pass.alpha_equal(zbinded, zexpected)示例10: test_env
def test_env():
x = relay.var("x", "float32")
y = relay.var("y", "float32")
z = relay.add(x, y)
z = relay.add(z, z)
f = relay.Function([x, y], z)
env = relay.Module()
env["myf"] = f
text = env.astext()
assert "def @myf" in text
assert "def @myf" in str(env)
assert "add(%0, %0) /* ty=float32 */" in text
assert "add(%0, %0) /* ty=float32 */" in str(env)
show(env.astext(annotate=lambda x: str(x.checked_type.dtype) if type(x) == relay.Call else ""))
show(text)示例11: annotated
def annotated():
conv2d_1 = relay.nn.conv2d(
data1,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
_conv2d_1 = relay.annotation.on_device(conv2d_1, dev2)
conv2d_2 = relay.nn.conv2d(
data2,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
_conv2d_2 = relay.annotation.on_device(conv2d_2, dev2)
add = relay.add(conv2d_1, conv2d_2)
_add = relay.annotation.on_device(add, dev1)
conv2d_3 = relay.nn.conv2d(
add,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
_conv2d_3 = relay.annotation.on_device(conv2d_3, dev2)
func = relay.Function([data1, data2, weight],
relay.Tuple(tvm.convert([_conv2d_1, _conv2d_2,
_conv2d_3, _add,
conv2d_3])))
func = relay.ir_pass.infer_type(func)
func = relay.ir_pass.rewrite_annotated_ops(func,
tvm.context(3).device_type)
func = relay.ir_pass.infer_type(func)
return relay.Function(relay.ir_pass.free_vars(func.body[4]),
func.body[4])示例12: before
def before(x, conv_weight, out_bias, out_scale, channels):
args = [x, conv_weight, out_bias]
y0 = relay.nn.conv2d(x, conv_weight,
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y0 = relay.multiply(y0, out_scale)
y0 = relay.nn.relu(y0)
y1 = relay.nn.conv2d(y0, conv_weight,
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y1 = relay.multiply(y1, out_scale)
y1 = relay.nn.relu(y1)
y2 = relay.nn.conv2d(y0, conv_weight,
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y2 = relay.multiply(y2, out_scale)
y2 = relay.nn.relu(y2)
y = relay.add(y1, y2)
return relay.Function(args, y)示例13: expected
def expected():
sb = relay.ScopeBuilder()
x = relay.var("x")
c_folded = (c_data + c_data)
t3 = sb.let("t3", relay.add(relay.const(c_folded), x))
sb.ret(t3)
return relay.Function([x], sb.get())示例14: expected
def expected(x, conv_weight, out_bias, out_scale, channels):
# use a fixed order of args so alpha equal check can pass
args = [x, conv_weight, out_bias]
def fold_conv_weight():
squeezed_scale = relay.squeeze(out_scale, axis=[1,2])
return relay.multiply(
conv_weight ,
relay.expand_dims(squeezed_scale, axis=1, num_newaxis=3))
y0 = relay.nn.conv2d(x, fold_conv_weight(),
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y0 = relay.nn.relu(y0)
y1 = relay.nn.conv2d(y0, fold_conv_weight(),
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y1 = relay.nn.relu(y1)
y2 = relay.nn.conv2d(y0, fold_conv_weight(),
channels=channels,
kernel_size=(3, 3),
padding=(1, 1))
y2 = relay.nn.relu(y2)
y = relay.add(y1, y2)
return relay.Function(args, y)示例15: test_depthwise_conv2d
def test_depthwise_conv2d():
batch_size = 1
dshape = (batch_size, 64, 56, 56)
weight_conv = relay.var("weight_depthwiseconv", shape=(64, 1, 3, 3))
data1 = relay.var("data1", shape=dshape)
data2 = relay.var("data2", shape=dshape)
depthwise_conv2d_1 = relay.nn.conv2d(
data1,
weight_conv,
kernel_size=(3, 3),
padding=(1, 1),
groups=64)
depthwise_conv2d_2 = relay.nn.conv2d(
data2,
weight_conv,
kernel_size=(3, 3),
padding=(1, 1),
groups=64)
add = relay.add(depthwise_conv2d_1, depthwise_conv2d_2)
func = relay.Function([data1, data2, weight_conv],
relay.Tuple(tvm.convert([depthwise_conv2d_1,
depthwise_conv2d_2,
add])))
func = relay.ir_pass.infer_type(func)
compute_count = relay.ir_pass.get_total_mac_number(func)
assert compute_count == 2 * np.prod(dshape) * 3*3