def test_multi_loss_graph_gradients():
# input data
shape1 = (1000, 100)
data1 = sym.Variable('data1', shape=(1000, 100), dtype=0)
# fake non-sparse label
label = sym.full(fill_value=3)
# square loss
sub1 = sym.elemwise_sub(data1, label, name="sub1")
square_loss = sym.sum(data=sub1**2, axis=1, name="square_loss")
# fake loss1
shape2 = (1000, )
data2 = sym.Variable('data2', shape=shape2, dtype=0)
loss1 = sym.sqrt(data2, name="loss1")
# fake loss2
loss2 = sym.relu(data1, name='loss2')
# block loss1
total_loss = sym.elemwise_sum(sym.block_grad(loss1),
square_loss,
num_args=2,
name="total_loss")
# grad_g.symbol.list_output_names()
# >> ['loss1_grad_0_output', 'grad_sum_output']
grad_g = graph_util.get_gradient_graph([total_loss, loss2],
total_loss.list_input_variables())
# infer shape
in_shapes, out_shapes = graph_util.infer_shape(grad_g)
assert out_shapes == [list(shape2), list(shape1)]
# grad_data1 is elemwise_sum of grad_loss2, grad_square_loss
grad_data1 = grad_g.symbol[1]
assert grad_data1.list_attr()['num_args'] == '2'
# block grad should return zero grad
grad_data2 = grad_g.symbol[0]
assert 'zeros_like' in grad_g.ir()
# test reverse infer shape for label
assert grad_g.apply('InferShape').json_attr('shape_num_unknown_nodes') == 0
# infer type
in_dtypes, out_dtypes = graph_util.infer_dtype(grad_g)
assert out_dtypes == ['float32', 'float32']
# test reverse infer type for label
assert grad_g.apply('InferType').json_attr('dtype_num_unknown_nodes') == 0
def test_multi_loss_graph_gradients():
# input data
shape1 = (1000, 100)
data1 = sym.Variable('data1', shape=(1000, 100), dtype=0)
# fake non-sparse label
label = sym.full(fill_value=3)
# square loss
sub1 = sym.elemwise_sub(data1, label, name="sub1")
square_loss = sym.sum(data=sub1**2, axis=1, name="square_loss")
# fake loss1
shape2 = (1000, )
data2 = sym.Variable('data2', shape=shape2, dtype=0)
loss1 = sym.sqrt(data2, name="loss1")
# fake loss2
loss2 = sym.relu(data1, name='loss2')
# block loss1
total_loss = sym.elemwise_sum(
sym.block_grad(loss1),
square_loss,
num_args=2,
name="total_loss")
# grad_g.symbol.list_output_names()
# >> ['loss1_grad_0_output', 'grad_sum_output']
grad_g = graph_util.get_gradient_graph([total_loss, loss2], total_loss.list_input_variables())
# infer shape
in_shapes, out_shapes = graph_util.infer_shape(grad_g)
assert out_shapes == [list(shape2), list(shape1)]
# grad_data1 is elemwise_sum of grad_loss2, grad_square_loss
grad_data1 = grad_g.symbol[1]
assert grad_data1.list_attr()['num_args'] == '2'
# block grad should return zero grad
grad_data2 = grad_g.symbol[0]
assert 'zeros_like' in grad_g.ir()
# test reverse infer shape for label
assert grad_g.apply('InferShape').json_attr('shape_num_unknown_nodes') == 0
# infer type
in_dtypes, out_dtypes = graph_util.infer_dtype(grad_g)
assert out_dtypes == ['float32', 'float32']
# test reverse infer type for label
assert grad_g.apply('InferType').json_attr('dtype_num_unknown_nodes') == 0
def test_full():
shape = (3, 4, 5)
value = 7
dtype = "float32"
for target, ctx in ctx_list():
data = sym.Variable("data", dtype=dtype)
# full_like
s = sym.full_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape,
fill_value=value,
dtype=dtype),
atol=1e-5,
rtol=1e-5)
# ones_like
s = sym.ones_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape, fill_value=1, dtype=dtype),
atol=1e-5,
rtol=1e-5)
# zeros_like
s = sym.zeros_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape, fill_value=0, dtype=dtype),
atol=1e-5,
rtol=1e-5)
# full
s = sym.full(shape=shape, dtype=dtype, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape,
fill_value=value,
dtype=dtype),
atol=1e-5,
rtol=1e-5)
# ones
s = sym.ones(shape=shape, dtype=dtype, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape, fill_value=1, dtype=dtype),
atol=1e-5,
rtol=1e-5)
# zeros
s = sym.zeros(shape=shape, dtype=dtype, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
np.testing.assert_allclose(out.asnumpy(),
np.full(shape, fill_value=0, dtype=dtype),
atol=1e-5,
rtol=1e-5)
def test_full():
shape = (3, 4, 5)
value = 7
dtype = "float32"
for target, ctx in ctx_list():
data = sym.Variable("data", dtype=dtype)
# full_like
s = sym.full_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=value, dtype=dtype),
atol=1e-5, rtol=1e-5)
# ones_like
s = sym.ones_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=1, dtype=dtype),
atol=1e-5, rtol=1e-5)
# zeros_like
s = sym.zeros_like(data=data, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target, {"data": shape})
m = graph_runtime.create(graph, lib, ctx)
m.run(data=np.random.uniform(size=shape).astype(dtype))
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=0, dtype=dtype),
atol=1e-5, rtol=1e-5)
# full
s = sym.full(shape=shape, dtype=dtype, fill_value=value, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=value, dtype=dtype),
atol=1e-5, rtol=1e-5)
# ones
s = sym.ones(shape=shape, dtype=dtype, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=1, dtype=dtype),
atol=1e-5, rtol=1e-5)
# zeros
s = sym.zeros(shape=shape, dtype=dtype, name="s")
graph, lib, _ = nnvm.compiler.build(s, target)
m = graph_runtime.create(graph, lib, ctx)
m.run()
out = m.get_output(0, tvm.nd.empty(shape, dtype=dtype))
tvm.testing.assert_allclose(
out.asnumpy(),
np.full(shape, fill_value=0, dtype=dtype),
atol=1e-5, rtol=1e-5)