def PartialFcJob(labels: oft.Numpy.Placeholder(
(batch_size, ), dtype=type_name_to_flow_type[label_type])):
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x-weight",
shape=(num_classes, 128),
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
with flow.scope.placement(device_type, "0:0-3"):
lebels_distribute = flow.distribute.broadcast()
weight_distribute = flow.distribute.split(0)
(
maped_label,
sampled_label,
sampled_weight,
) = flow.distributed_partial_fc_sample(
weight=x.with_distribute(weight_distribute),
label=labels.with_distribute(lebels_distribute),
num_sample=num_sample,
)
with flow.scope.placement(device_type, "0:0"):
sampled_weight = flow.identity(sampled_weight)
loss = flow.math.square(sampled_weight)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch_diff(sampled_weight,
test_global_storage.Setter("sampled_weight_diff"))
return x, maped_label, sampled_label, sampled_weight
def test_job(x: oft.Numpy.Placeholder(input_shape, dtype=flow.float32),):
v = flow.get_variable(
name="v",
shape=(1,),
dtype=flow.float32,
initializer=flow.zeros_initializer(),
)
x = x + v
x1 = flow.identity(x)
x2 = flow.identity(x)
flow.watch_diff(x1, test_global_storage.Setter("x1_diff"))
flow.watch_diff(x2, test_global_storage.Setter("x2_diff"))
x1 = flow.cast(x1, data_type)
x2 = flow.cast(x2, data_type)
y1 = flow.layers.batch_normalization_relu(x1, axis=axis, name="BN1")
y2 = flow.math.relu(flow.layers.batch_normalization(x2, axis=axis, name="BN2"))
y1 = flow.cast(y1, flow.float32)
y2 = flow.cast(y2, flow.float32)
flow.watch(y1, test_global_storage.Setter("y1"))
flow.watch(y2, test_global_storage.Setter("y2"))
loss = flow.math.reduce_mean(y1 + y2)
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [0.001]), momentum=0
).minimize(flow.math.reduce_sum(loss))
return loss
def UpsampleJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"input",
shape=input_shape,
dtype=type_name_to_flow_type[dtype],
initializer=flow.random_uniform_initializer(minval=2, maxval=5),
trainable=True,
)
loss = flow.layers.upsample_2d(
x,
size=size,
data_format=data_format,
interpolation=interpolation,
align_corners=align_corners,
)
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [1e-4]), momentum=0
).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def SoftmaxJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=x_shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=-0.1,
maxval=0.1),
trainable=True,
)
if data_type == "float16":
loss = flow.cast(
flow.nn.softmax(flow.cast(x, dtype=flow.float16),
axis=axis),
dtype=flow.float,
)
else:
loss = flow.nn.softmax(x, axis=axis)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def SparseSoftmaxCrossEntropyWithLogitsJob(labels: oft.Numpy.Placeholder(
(batch_size, ), dtype=type_name_to_flow_type[label_type])):
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=(batch_size, num_classes),
dtype=type_name_to_flow_type[data_type],
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
with flow.scope.placement(device_type, "0:0-3"):
lebels_distribute = flow.distribute.broadcast()
logits_distribute = flow.distribute.split(len(x.shape) - 1)
loss = flow.nn.sparse_softmax_cross_entropy_with_logits(
labels=labels.with_distribute(lebels_distribute),
logits=x.with_distribute(logits_distribute),
)
loss = flow.math.square(loss)
with flow.scope.placement(device_type, "0:0"):
loss = flow.identity(loss)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def TestMultiInputJob():
with flow.scope.placement("gpu", "0:0"):
x1 = flow.get_variable(
"x1",
shape=shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
x2 = flow.get_variable(
"x2",
shape=shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
loss = TestMultiInput(x1, x2)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x1, test_global_storage.Setter("x1"))
flow.watch_diff(x1, test_global_storage.Setter("x1_diff"))
flow.watch(x2, test_global_storage.Setter("x2"))
flow.watch_diff(x2, test_global_storage.Setter("x2_diff"))
return loss
def test_masked_fill_fw_bw_job(
x: oft.Numpy.Placeholder(x_shape, dtype=flow_type),
mask: oft.Numpy.Placeholder(mask_shape, dtype=flow_type),
):
with flow.scope.placement(device, "0:0"):
y = flow.get_variable(
name="vx",
shape=(1, ),
dtype=flow.float,
initializer=flow.zeros_initializer(),
)
x += flow.cast(y, flow_type)
mask = flow.cast(mask, dtype=flow.int8)
if type_name == "float16":
out = flow.cast(
flow.masked_fill(flow.cast(x, flow.float16), mask, value),
flow.float,
)
else:
out = flow.masked_fill(x, mask, value)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(out)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(out, test_global_storage.Setter("out"))
flow.watch_diff(out, test_global_storage.Setter("out_diff"))
return out
def diag_job(
input_tensor: tp.Numpy.Placeholder(shape=(input_shape),
dtype=flow.float),
) -> tp.Numpy:
input_var = flow.get_variable(
"input_tensor",
shape=(input_shape),
dtype=flow.float,
initializer=flow.zeros_initializer(),
trainable=True,
)
input_tensor = input_tensor + input_var
input_tensor = flow.cast_to_current_logical_view(input_tensor)
input_tensor = flow.cast(input_tensor, type_name_to_flow_type[dtype])
output = flow.diag(input_tensor, dim)
if (output.dtype == flow.int64 or output.dtype == flow.int8
or output.dtype == flow.int32):
output = flow.cast(output, flow.float)
flow.optimizer.Adam(
flow.optimizer.PiecewiseConstantScheduler([],
[1e-4])).minimize(output)
flow.watch(input_tensor, test_global_storage.Setter("x"))
flow.watch_diff(input_tensor, test_global_storage.Setter("x_diff"))
flow.watch(output, test_global_storage.Setter("output"))
flow.watch_diff(output, test_global_storage.Setter("output_diff"))
return output
def SparseSoftmaxCrossEntropyWithLogitsJob(labels: oft.Numpy.Placeholder(
(batch_size, ), dtype=type_name_to_flow_type[label_type])):
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=(batch_size, num_classes),
dtype=type_name_to_flow_type[data_type],
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
prediction = flow.nn.softmax(logits=x)
with flow.scope.placement(device_type, "0:0-3"):
lebels_distribute = flow.distribute.broadcast()
prediction_distribute = flow.distribute.split(
len(prediction.shape) - 1)
loss = flow.nn.sparse_cross_entropy(
labels=labels.with_distribute(lebels_distribute),
prediction=prediction.with_distribute(prediction_distribute),
)
with flow.scope.placement(device_type, "0:0"):
loss = flow.math.square(loss)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def DropoutJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=x_shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=-1,
maxval=1),
trainable=True,
)
if data_type == "float16":
x = flow.cast(flow.cast(x, flow.float16), dtype)
of_out = flow.cast(
flow.nn.dropout(flow.cast(x, flow.float16),
rate=rate,
seed=seed,
name="dropout"),
dtype,
)
else:
of_out = flow.nn.dropout(x,
rate=rate,
seed=seed,
name="dropout")
loss = flow.math.square(of_out)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(of_out, test_global_storage.Setter("out"))
flow.watch_diff(of_out, test_global_storage.Setter("out_diff"))
return loss
def test_fused_scale_tril_fw_bw_job(
x: oft.Numpy.Placeholder(shape, dtype=flow_type), ):
with flow.scope.placement(device, "0:0"):
x_var = flow.get_variable(
name="xv",
shape=(1, ),
dtype=flow.float,
initializer=flow.zeros_initializer(),
)
x += flow.cast(x_var, dtype=flow_type)
if type_name == "float16":
out = flow.cast(
flow.math.fused_scale_tril(flow.cast(x, flow.float16),
diagonal,
scale=scale),
flow.float,
)
else:
out = flow.math.fused_scale_tril(x, diagonal, scale=scale)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(out)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(out, test_global_storage.Setter("out"))
flow.watch_diff(out, test_global_storage.Setter("out_diff"))
return out
def SplitLikeJob(x: oft.Numpy.Placeholder(x_shape, dtype=flow.float)):
v = flow.get_variable(
"x",
shape=x_shape,
dtype=flow.float,
initializer=flow.constant_initializer(0),
trainable=True,
)
x += v
like0 = flow.constant(0, dtype=flow.float, shape=like0_shape)
like1 = flow.constant(0, dtype=flow.float, shape=like1_shape)
with flow.scope.placement("gpu", "0:0"):
y0, y1 = split_like(x, [like0, like1], "split_like")
loss = y0
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [1e-4]), momentum=0
).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return y0, y1
def Matmul(
x: tp.Numpy.Placeholder((4, 4), dtype=flow.float32, batch_axis=None),
y: tp.Numpy.Placeholder((4, 4), dtype=flow.float32, batch_axis=1),
) -> tp.Numpy:
s = flow.matmul(x, y) # model parallel
flow.watch(s, Watch)
z = flow.matmul(s, x) # data parallel
return z
def slice_fn():
with flow.scope.placement(device_tag, "0:0-{}".format(device_num - 1)):
var = flow.get_variable(
name="var",
shape=var_shape,
dtype=flow_dtype,
initializer=flow.random_uniform_initializer(-10, 10, dtype=flow_dtype),
distribute=flow.distribute.split(split_axis),
)
flow.watch(var, test_global_storage.Setter("var"))
ret = flow.experimental.logical_slice(var, slice_tuples)
return ret
def MatmulJob():
with flow.scope.placement(device_type, "0:0"):
a = flow.get_variable(
"a",
shape=a_shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=0, maxval=1),
trainable=True,
)
b = flow.get_variable(
"b",
shape=b_shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=0, maxval=1),
trainable=True,
)
if data_type == "float16":
out = flow.matmul(
flow.cast(a, dtype=flow.float16),
flow.cast(b, dtype=flow.float16),
transpose_a,
transpose_b,
alpha,
)
c = flow.get_variable(
"c",
shape=out.shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=-1, maxval=1),
trainable=True,
)
loss = flow.cast(
out + flow.cast(c, dtype=flow.float16), dtype=flow.float
)
else:
out = flow.matmul(a, b, transpose_a, transpose_b, alpha)
c = flow.get_variable(
"c",
shape=out.shape,
dtype=dtype,
initializer=flow.random_uniform_initializer(minval=-1, maxval=1),
trainable=True,
)
loss = out + c
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [1e-4]), momentum=0
).minimize(loss)
flow.watch(a, test_global_storage.Setter("a"))
flow.watch_diff(a, test_global_storage.Setter("a_diff"))
flow.watch(b, test_global_storage.Setter("b"))
flow.watch_diff(b, test_global_storage.Setter("b_diff"))
flow.watch(c, test_global_storage.Setter("c"))
flow.watch_diff(c, test_global_storage.Setter("c_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def ReduceSumJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"in",
shape=input_shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=2,
maxval=5),
trainable=True,
)
loss = flow.math.reduce_sum(x, axis=axis, keepdims=keepdims)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def CastJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"in",
shape=input_shape,
dtype=type_name_to_flow_type[dtype],
initializer=flow.random_uniform_initializer(),
trainable=True,
)
loss = flow.cast(x, dtype=flow.float)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def ReduceMaxJob(x: oft.Numpy.Placeholder(input_shape, dtype=flow.float)):
with flow.scope.placement(device_type, "0:0"):
x += flow.get_variable(
name="v1",
shape=input_shape,
dtype=flow.float,
initializer=flow.zeros_initializer(),
)
loss = flow.math.reduce_max(x, axis=axis, keepdims=keepdims)
loss = flow.identity(loss)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def ReduceSumJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"in",
shape=input_shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=2, maxval=5),
trainable=True,
)
loss = flow.math.reduce_sum(x, axis=axis, keepdims=keepdims)
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [1e-4]), momentum=0
).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def LeakyReluJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=x_shape,
dtype=type_name_to_flow_type[data_type],
initializer=flow.random_uniform_initializer(minval=-10, maxval=10),
trainable=True,
)
loss = flow.nn.leaky_relu(x, alpha=alpha)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def L2NormalizeJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=x_shape,
dtype=type_name_to_flow_type[data_type],
initializer=flow.random_uniform_initializer(minval=-10, maxval=10),
trainable=True,
)
loss = flow.math.l2_normalize(x, axis=axis, epsilon=epsilon)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def SquareJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=x_shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=-10, maxval=10),
trainable=True,
)
loss = flow.math.square(x)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def test_job(
x: oft.Numpy.Placeholder(input_shape, dtype=flow.float32),
labels: oft.Numpy.Placeholder(label_shape, dtype=flow.int32),
):
with flow.scope.placement("gpu", "0:0"):
v = flow.get_variable(
name="v",
shape=(1, ),
dtype=flow.float32,
initializer=flow.zeros_initializer(),
)
x = x + v
x1 = flow.identity(x)
x2 = flow.identity(x)
flow.watch_diff(x1, test_global_storage.Setter("x1_diff"))
flow.watch_diff(x2, test_global_storage.Setter("x2_diff"))
x1 = flow.cast(x1, data_type)
x2 = flow.cast(x2, data_type)
with flow.scope.placement("gpu", "0:0-3"):
y1 = (flow.combined_margin_loss(
x1.with_distribute(flow.distribute.split(1)),
labels.with_distribute(flow.distribute.broadcast()),
m1,
m2,
m3,
) * s)
y2 = margin_loss(m1, m2, m3, s, x2, labels)
with flow.scope.placement("gpu", "0:0"):
y1 = flow.cast(y1, flow.float)
y2 = flow.cast(y2, flow.float)
flow.watch(y1, test_global_storage.Setter("y1"))
flow.watch(y2, test_global_storage.Setter("y2"))
loss = y1 + y2
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [0.001]),
momentum=0).minimize(flow.math.reduce_sum(loss))
return loss
def CastJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"in",
shape=input_shape,
dtype=type_name_to_flow_type[dtype],
initializer=flow.random_uniform_initializer(),
trainable=True,
)
loss = flow.cast(x, dtype=flow.float)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def FlattenJob() -> flow.typing.Numpy:
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"in",
shape=input_shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=2, maxval=5),
trainable=True,
)
loss = flow.flatten(x, start_dim=start_dim, end_dim=end_dim)
flow.optimizer.SGD(
flow.optimizer.PiecewiseConstantScheduler([], [1e-4]), momentum=0
).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
return loss
def TransposeJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"input",
shape=input_shape,
dtype=flow.float,
initializer=flow.random_uniform_initializer(minval=2, maxval=5),
trainable=True,
)
loss = flow.transpose(x, perm)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def ActivationJob():
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=shape,
dtype=data_type,
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
loss = of_activation_map[activation_type](x)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def ReduceMinJob(x: oft.Numpy.Placeholder(input_shape, dtype=flow.float)):
with flow.scope.placement(device_type, "0:0"):
x += flow.get_variable(
name="v1",
shape=input_shape,
dtype=flow.float,
initializer=flow.zeros_initializer(),
)
loss = flow.math.reduce_min(x, axis=axis, keepdims=keepdims)
loss = flow.identity(loss)
flow.optimizer.SGD(flow.optimizer.PiecewiseConstantScheduler(
[], [1e-4]),
momentum=0).minimize(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def SoftmaxCrossEntropyWithLogitsJob(labels: oft.Numpy.Placeholder(
shape, dtype=type_name_to_flow_type[data_type])):
with flow.scope.placement(device_type, "0:0"):
x = flow.get_variable(
"x",
shape=shape,
dtype=type_name_to_flow_type[data_type],
initializer=flow.random_uniform_initializer(minval=-10,
maxval=10),
trainable=True,
)
loss = flow.nn.softmax_cross_entropy_with_logits(labels=labels,
logits=x)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
flow.watch(loss, test_global_storage.Setter("loss"))
flow.watch_diff(loss, test_global_storage.Setter("loss_diff"))
return loss
def two_stage_reduce_job(x: oft.Numpy.Placeholder((4, 20, 20, 20))):
with flow.scope.placement(device_type, "0:0"):
x += flow.get_variable(
name="v1",
shape=(1,),
dtype=flow.float,
initializer=flow.zeros_initializer(),
)
with flow.scope.placement(device_type, "0:0-3"):
loss = flow_func(
x.with_distribute(flow.distribute.split(split_axis)),
axis=axis,
keepdims=True,
)
loss = flow.identity(loss)
flow.losses.add_loss(loss)
flow.watch(x, test_global_storage.Setter("x"))
flow.watch_diff(x, test_global_storage.Setter("x_diff"))
return loss
