提交 e31a469e 编写于 作者: Q Qiao Longfei 提交者: GitHub

add gradient test framework (#3226)

* init grad op checker

* can run

* add GradeChecker class

* use get_numeric_gradient

* refine code

* add softmax and cross entropy auto grad test

* use close to judge op_grad and numeric_grad

* add cpu and gpu compare

* add comments

* add support_gpu

* fix allclose

* fix name error and symplify code

* optimize gradient checker

* add test_cross_entropy_op

* update gradient_checker.py

* optimize code

* use random.uniform instead of random.random

* fix type bug

* optimize check_grad

* put SupportGPU into OperatorBase

* typo
上级 6540701f
develop 2.0.1-rocm-post Ligoml-patch-1 OliverLPH-patch-1 OliverLPH-patch-2 PaddlePM-patch-1 PaddlePM-patch-2 ZHUI-patch-1 add_default_att add_model_benchmark_ci add_some_yaml_config addfile all_new_design_exec ascendrc ascendrelease cherry_undefined_var compile_windows cp_2.4_fix_numpy delete_2.0.1-rocm-post delete_add_default_att delete_all_new_design_exec delete_ascendrc delete_compile_windows delete_delete_addfile delete_disable_iterable_dataset_unittest delete_fix_dataloader_memory_leak delete_fix_imperative_dygraph_error delete_fix_retry_ci delete_fix_undefined_var delete_improve_sccache delete_incubate/lite delete_paddle_tiny_install delete_paralleltest delete_prv-disable-more-cache delete_revert-31068-fix_conv3d_windows delete_revert-31562-mean delete_revert-33630-bug-fix delete_revert-34159-add_npu_bce_logical_dev delete_revert-34910-spinlocks_for_allocator delete_revert-35069-revert-34910-spinlocks_for_allocator delete_revert-36057-dev/read_flags_in_ut dingjiaweiww-patch-1 disable_iterable_dataset_unittest dy2static enable_eager_model_test final_state_gen_python_c final_state_intermediate fix-numpy-issue fix_concat_slice fix_dataloader_memory_leak fix_dlpack_for fix_imperative_dygraph_error fix_npu_ci fix_op_flops fix_retry_ci fix_rnn_docs fix_tensor_type fix_undefined_var fix_var_stop_gradient_error fixiscan fixiscan1 fixiscan2 fixiscan3 github/fork/123malin/netifaces github/fork/123malin/tdm_abacus github/fork/AshburnLee/dev_unique github/fork/ForFishes/fix_memory_matmul github/fork/ForFishes/rm_fluid github/fork/LielinJiang/move-2.0-api github/fork/LielinJiang/visual-dl-cb github/fork/LiuChiachi/add-transformer-generate-square-subsequent-mask-api github/fork/LiuChiachi/fix-example-code-for-hapi-Model github/fork/LiuChiachi/remove-input-requirment-in-dygraph-Model github/fork/MrChengmo/fix_ps_profiler github/fork/MrChengmo/update_ps_heter github/fork/PWhiddy/patch-1 github/fork/Shixiaowei02/dev/save_load_upgrade github/fork/TCChenlong/fix_hapi github/fork/TCChenlong/fix_inden github/fork/Thunderbrook/xpu_slice github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_2 github/fork/XieYunshen/disable_ut_test_parallel_executor_fetch_isolated_var_3 github/fork/XieYunshen/timeout_20S_ut github/fork/ZeyuChen/remove-nltk github/fork/arlesniak/arlesniak/selective__mkldnn_flags github/fork/baiyfbupt/code_doc_mig github/fork/chalsliu/set_timeout github/fork/chen-zhiyu/develop github/fork/chenwhql/ci/try_to_find_test_buffer_shared_memory_reuse_pass_error github/fork/chenwhql/dygraph/remove_scale_loss_and_apply_collective_grads github/fork/chenwhql/saveload/add_get_inference_program github/fork/chenwhql/saveload/remove_save_load_config github/fork/cryoco/pass-compatibility-trt github/fork/danleifeng/isempty_api2.0 github/fork/frankwhzhang/api_transfer github/fork/hbwx24/error_msg/cuda_kernel_error_msg github/fork/heavengate/cherry_yolo_box github/fork/heavengate/update_yolo_box github/fork/iclementine/rnn_fix github/fork/iducn/testestse github/fork/jczaja/prv-25537-fix github/fork/jeff41404/release/1.8 github/fork/jiweibo/api_2.0 github/fork/jiweibo/fix_lite_resnet50_test github/fork/juncaipeng/fix_doc_1 github/fork/lfchener/sample_code github/fork/littletomatodonkey/fix_reg_doc github/fork/liym27/dy2stat_update_assign_to_rc20 github/fork/luotao1/profiler_ut github/fork/mapingshuo/add_wait github/fork/mapingshuo/doc_2.0 github/fork/mapingshuo/zero-0.5 github/fork/miraiwk/dev github/fork/pangyoki/add-Categorical-class-branch github/fork/pangyoki/add-multinomial-op-branch github/fork/pangyoki/fix-test_distritbution-CI github/fork/qjing666/doublegrad github/fork/qjing666/fix_hdfs_download github/fork/sandyhouse/add_gather_etc github/fork/sandyhouse/add_send_recv_alltoall_etc github/fork/sandyhouse/pipeline_exe_run github/fork/seiriosPlus/feature/large_scale_kv_save_delta github/fork/seiriosPlus/fix/paddle_errors_fix github/fork/seiriosPlus/fix/paddle_op_errors github/fork/shangzhizhou/fix_test_activation_op_random_bug github/fork/smallv0221/yxp0924 github/fork/smallv0221/yxp0925 github/fork/swtkiwi/del-matplotlib github/fork/tianshuo78520a/kunlun_test github/fork/tianshuo78520a/update_dockerfile github/fork/wanghaoshuang/bert_fuse github/fork/wanghaoshuang/label_smooth github/fork/wanghuancoder/develop_CUDASynchronize github/fork/wanghuancoder/develop_Layer_doc github/fork/wanghuancoder/develop_ParameterList_doc github/fork/wanghuancoder/develop_Sequential_doc github/fork/wanghuancoder/develop_bilinear_tensor_product github/fork/wanghuancoder/develop_coverage_build_sh github/fork/wanghuancoder/develop_in_dynamic_mode_doc github/fork/wanghuancoder/develop_unique_name_doc github/fork/wangxicoding/fleet_meta_combine github/fork/wawltor/error_message_fix_5 github/fork/willthefrog/remove_l2_norm github/fork/windstamp/momentum_op github/fork/windstamp/mv_op_5 github/fork/windstamp/normal_api github/fork/wojtuss/wojtuss/fusion_gru_quantization github/fork/wojtuss/wojtuss/quantization-with-shift github/fork/wzzju/fix_err_info github/fork/wzzju/pure_fp16 github/fork/xiemoyuan/op_error_message github/fork/xiemoyuan/optimize_error_message github/fork/yaoxuefeng6/fix_doc github/fork/yaoxuefeng6/mod_dataset_v2 github/fork/yongqiangma/lod github/fork/ysh329/fix-clip-by-norm-error github/fork/ysh329/fix-error-clip-by-value github/fork/yukavio/error_info github/fork/zhangting2020/conv_filter_grad github/fork/zhangting2020/is_compile_with_cuda github/fork/zhangting2020/place_doc github/fork/zhangting2020/program github/fork/zhhsplendid/fix_any github/fork/zhhsplendid/refine_api2 github/fork/zhhsplendid/refine_api2_test github/fork/zhhsplendid/refine_api_test_ptb_lm github/fork/zhhsplendid/refine_api_test_resnet github/fork/zhhsplendid/refine_api_test_simnet github/fork/zhiqiu/dev/refine_initializer github/fork/zhiqiu/dev/remove_inplace_argument github/fork/zlsh80826/nvinfer_plugin_var_len_cuda11 improve_sccache incubate/frl_train_eval incubate/infrt incubate/lite inplace_addto layer_norm make_flag_adding_easier master matmul_double_grad move_embedding_to_phi move_histogram_to_pten move_sgd_to_phi move_slice_to_pten move_temporal_shift_to_phi move_yolo_box_to_phi npu_fix_alloc numel paddle_tiny_install paralleltest preln_ernie prv-disable-more-cache prv-md-even-more prv-onednn-2.5 prv-reshape-mkldnn-ut2 pten_tensor_refactor release/0.11.0 release/0.12.0 release/0.13.0 release/0.14.0 release/0.15.0 release/1.0.0 release/1.1 release/1.2 release/1.3 release/1.4 release/1.5 release/1.6 release/1.7 release/1.8 release/2.0 release/2.0-alpha release/2.0-beta release/2.0-rc release/2.0-rc1 release/2.1 release/2.2 release/2.3 release/2.3-fc-ernie-fix release/2.4 release/lite-0.1 revert-24981-add_device_attr_for_regulization revert-26856-strategy_example2 revert-27520-disable_pr revert-31068-fix_conv3d_windows revert-31562-mean revert-32290-develop-hardlabel revert-33037-forci revert-33475-fix_cifar_label_dimension revert-33630-bug-fix revert-34159-add_npu_bce_logical_dev revert-34406-add_copy_from_tensor revert-34910-spinlocks_for_allocator revert-35069-revert-34910-spinlocks_for_allocator revert-36057-dev/read_flags_in_ut revert-36201-refine_fast_threaded_ssa_graph_executor revert-36985-add_license revert-37318-refactor_dygraph_to_eager revert-37926-eager_coreops_500 revert-37956-revert-37727-pylayer_support_tuple revert-38100-mingdong revert-38301-allocation_rearrange_pr revert-38703-numpy_bf16_package_reupload revert-38732-remove_useless_header_in_elementwise_mul_grad revert-38959-Reduce_Grad revert-39143-adjust_empty revert-39227-move_trace_op_to_pten revert-39268-dev/remove_concat_fluid_kernel revert-40170-support_partial_grad revert-41056-revert-40727-move_some_activaion_to_phi revert-41065-revert-40993-mv_ele_floordiv_pow revert-41068-revert-40790-phi_new revert-41944-smaller_inference_api_test revert-42149-do-not-reset-default-stream-for-stream-safe-cuda-allocator revert-43155-fix_ut_tempfile revert-43882-revert-41944-smaller_inference_api_test revert-45808-phi/simplify_size_op revert-46827-deform_comment revert-47325-remove_cudnn_hardcode revert-47645-add_npu_storage_dims revert-48815-set_free_when_no_cache_hit_default_value_true revert-49654-prim_api_gen revert-49763-fix_static_composite_gen rocm_dev_0217 support-0D-sort support_weight_transpose test_benchmark_ci test_feature_precision_test_c test_for_Filtetfiles test_model_benchmark test_model_benchmark_ci zhiqiu-patch-1 v2.4.1 v2.4.0 v2.4.0-rc0 v2.3.2 v2.3.1 v2.3.0 v2.3.0-rc0 v2.2.2 v2.2.1 v2.2.0 v2.2.0-rc0 v2.2.0-bak0 v2.1.3 v2.1.2 v2.1.1 v2.1.0 v2.1.0-rc0 v2.0.2 v2.0.1 v2.0.0 v2.0.0-rc1 v2.0.0-rc0 v2.0.0-beta0 v2.0.0-alpha0 v1.8.5 v1.8.4 v1.8.3 v1.8.2 v1.8.1 v1.8.0 v1.7.2 v1.7.1 v1.7.0 v1.6.3 v1.6.2 v1.6.1 v1.6.0 v1.6.0-rc0 v1.5.2 v1.5.1 v1.5.0 v1.4.1 v1.4.0 v1.3.2 v1.3.1 v1.3.0 v1.2.1 v1.2.0 v1.1.0 v1.0.2 v1.0.1 v1.0.0 v1.0.0-rc0 v0.15.0 v0.15.0-rc0 v0.14.0 v0.13.0 v0.12.0 v0.11.1a2 v0.11.1a1 v0.11.0 lite-v0.1
无相关合并请求
......@@ -260,12 +260,6 @@ class OpRegistry {
return CreateOp(op_desc.type(), inputs, outputs, attrs);
}
static bool SupportGPU(const std::string& op_type) {
OperatorWithKernel::OpKernelKey key;
key.place_ = platform::GPUPlace();
return OperatorWithKernel::AllOpKernels().at(op_type).count(key) != 0;
}
static std::shared_ptr<OperatorBase> CreateGradOp(const OperatorBase& op) {
PADDLE_ENFORCE(!op.IsNetOp(),
"Use framework::Backward to get backward ops");
......
......@@ -88,6 +88,8 @@ class OperatorBase {
virtual bool IsNetOp() const { return false; }
virtual bool SupportGPU() const { return false; }
/// rename inputs outputs name
void Rename(const std::string& old_name, const std::string& new_name);
......@@ -308,7 +310,7 @@ class OperatorWithKernel : public OperatorBase {
using OpKernelMap =
std::unordered_map<OpKernelKey, std::unique_ptr<OpKernel>, OpKernelHash>;
void InferShape(const Scope& scope) const {
void InferShape(const Scope& scope) const override {
InferShape(InferShapeContext(this, scope));
}
......@@ -324,6 +326,12 @@ class OperatorWithKernel : public OperatorBase {
return g_all_op_kernels;
}
bool SupportGPU() const override {
OperatorWithKernel::OpKernelKey key;
key.place_ = platform::GPUPlace();
return OperatorWithKernel::AllOpKernels().at(type_).count(key) != 0;
}
protected:
virtual void InferShape(const InferShapeContext& ctx) const = 0;
};
......
......@@ -57,6 +57,26 @@ void ExposeOperator(ClassType &m) {
[](const typename ClassType::type &op) -> std::vector<std::string> {
return op.outputs_;
})
.def("inputs",
[](const typename ClassType::type &op) -> std::vector<std::string> {
return op.inputs_;
})
.def("support_gpu", &ClassType::type::SupportGPU)
.def("temp_outputs",
[](const typename ClassType::type &op) -> std::vector<std::string> {
auto iter = op.attrs_.find("temporary_index");
std::vector<std::string> ret;
if (iter == op.attrs_.end()) {
return ret;
} else {
auto tmp_idx = boost::get<std::vector<int>>(iter->second);
ret.reserve(tmp_idx.size());
for (auto &index : tmp_idx) {
ret.push_back(op.outputs_.at(index));
}
return ret;
}
})
.def("__str__", &ClassType::type::DebugString);
}
......@@ -202,8 +222,6 @@ All parameter, weight, gradient are variables in Paddle.
return OpRegistry::CreateOp(desc);
});
operator_base.def_static("support_gpu", &OpRegistry::SupportGPU);
operator_base.def("backward",
[](const OperatorBase &forwardOp,
const std::unordered_set<std::string> &no_grad_vars) {
......
......@@ -70,7 +70,8 @@ REGISTER_OP(onehot_cross_entropy, ops::OnehotCrossEntropyOp,
ops::OnehotCrossEntropyOpMaker);
REGISTER_OP_CPU_KERNEL(onehot_cross_entropy,
ops::OnehotCrossEntropyOpKernel<ops::CPUPlace, float>);
REGISTER_GRADIENT_OP(onehot_cross_entropy, onehot_cross_entropy_grad,
ops::OnehotCrossEntropyGradientOp);
REGISTER_OP_CPU_KERNEL(
onehot_cross_entropy_grad,
ops::OnehotCrossEntropyGradientOpKernel<ops::CPUPlace, float>);
......@@ -65,6 +65,15 @@ class NetOp : public framework::OperatorBase {
}
}
bool SupportGPU() const override {
for (auto& op : ops_) {
if (!op->SupportGPU()) {
return false;
}
}
return true;
}
/**
* @brief Add an operator by ptr
*/
......
......@@ -13,6 +13,7 @@ py_test(test_protobuf SRCS test_protobuf.py)
py_test(test_add_two_op SRCS test_add_two_op.py)
py_test(test_sigmoid_op SRCS test_sigmoid_op.py)
py_test(test_softmax_op SRCS test_softmax_op.py)
py_test(test_cross_entropy_op SRCS test_cross_entropy_op.py)
py_test(test_fill_zeros_like_op SRCS test_fill_zeros_like_op.py)
py_test(gradient_checker SRCS gradient_checker.py)
......
import unittest
import numpy
import paddle.v2.framework.core as core
from paddle.v2.framework.op import Operator
import numpy
import unittest
__all__ = ['get_numeric_gradient']
def create_op(op_type):
kwargs = dict()
for in_name in Operator.get_op_input_names(op_type):
kwargs[in_name] = in_name
for out_name in Operator.get_op_output_names(op_type):
kwargs[out_name] = out_name
return Operator(op_type, **kwargs)
def grad_var_name(var_name):
return var_name + "@GRAD"
def get_numeric_gradient(op,
input_values,
output_name,
input_to_check,
delta=1e-2,
delta=0.005,
local_scope=None):
"""
Get Numeric Gradient for an operator's input.
......@@ -76,6 +91,113 @@ def get_numeric_gradient(op,
return gradient_flat.reshape(tensor_to_check.get_dims())
class GradientChecker(unittest.TestCase):
def __is_close(self, numeric_grads, scope, max_relative_error):
for name in numeric_grads:
op_grad = numpy.array(
scope.find_var(grad_var_name(name)).get_tensor())
is_close = numpy.allclose(
numeric_grads[name], op_grad, rtol=max_relative_error, atol=100)
if not is_close:
return False
return True
def check_grad(self,
forward_op,
input_vars,
inputs_to_check,
output_name,
no_grad_set=None,
only_cpu=False,
max_relative_error=0.005):
"""
:param forward_op: used to create backward_op
:param input_vars: numpy value of input variable. The following
computation will use these variables.
:param inputs_to_check: inputs var names that should check gradient.
:param output_name: output name that used to
:param max_relative_error: The relative tolerance parameter.
:param no_grad_set: used when create backward ops
:param only_cpu: only compute and check gradient on cpu kernel.
:return:
"""
if no_grad_set is None:
no_grad_set = set()
tmp_outs = forward_op.temp_outputs()
no_tmp_out = filter(lambda name: name not in tmp_outs,
forward_op.outputs())
if len(no_tmp_out) != 1:
raise ValueError("non temp out_names should be 1")
in_names = forward_op.inputs()
for no_grad in no_grad_set:
if no_grad not in in_names:
raise ValueError("no_grad should be in in_names")
backward_op = core.Operator.backward(forward_op, no_grad_set)
places = [core.CPUPlace()]
if not only_cpu and core.is_compile_gpu() and backward_op.support_gpu():
places.append(core.GPUPlace(0))
numeric_grad = dict()
# get numeric gradient
for check_name in inputs_to_check:
numeric_grad[check_name] = \
get_numeric_gradient(forward_op, input_vars, output_name, check_name)
# get operator gradient according to different device
for place in places:
scope = core.Scope()
ctx = core.DeviceContext.create(place)
# create input var and set value
for name, value in input_vars.iteritems():
if name not in in_names:
raise ValueError(name + " not in op.inputs_")
var = scope.new_var(name).get_tensor()
var.set_dims(value.shape)
var.set(value, place)
# create output var
for out_name in forward_op.outputs():
scope.new_var(out_name).get_tensor()
# infer the shape of output var and compute/set value of output var
forward_op.infer_shape(scope)
forward_op.run(scope, ctx)
# create output grad var
# set shape as the output var
# set value of this grad to ones
for name in forward_op.outputs():
out_tensor = scope.find_var(name).get_tensor()
grad_tensor = scope.new_var(grad_var_name(name)).get_tensor()
grad_tensor.set_dims(out_tensor.shape())
data = 1.0 * numpy.ones(out_tensor.shape())
grad_tensor.set(data, place)
# create input grad var
for name in backward_op.outputs():
scope.new_var(name).get_tensor()
# infer the shape of input gradient var and compute/set it's value
# with backward op
backward_op.infer_shape(scope)
backward_op.run(scope, ctx)
if isinstance(place, core.CPUPlace):
msg = "CPU kernel gradient is not close to numeric gradient"
else:
if isinstance(place, core.GPUPlace):
msg = "GPU kernel gradient is not close to numeric gradient"
else:
raise ValueError("unknown place " + type(place))
self.assertTrue(
self.__is_close(numeric_grad, scope, max_relative_error), msg)
if __name__ == '__main__':
class GetNumericGradientTest(unittest.TestCase):
......@@ -87,4 +209,28 @@ if __name__ == '__main__':
arr = get_numeric_gradient(add_op, {'X': x, "Y": y}, 'Z', 'X')
self.assertAlmostEqual(arr.mean(), 1.0, delta=1e-2)
def test_softmax_op(self):
def stable_softmax(x):
"""Compute the softmax of vector x in a numerically stable way."""
shiftx = x - numpy.max(x)
exps = numpy.exp(shiftx)
return exps / numpy.sum(exps)
def label_softmax_grad(Y, dY):
dX = Y * 0.0
for i in range(Y.shape[0]):
d = numpy.dot(Y[i, :], dY[i, :])
dX[i, :] = Y[i, :] * (dY[i, :] - d)
return dX
softmax_op = Operator("softmax", X="X", Y="Y")
X = numpy.random.random((2, 2)).astype("float32")
Y = numpy.apply_along_axis(stable_softmax, 1, X)
dY = numpy.ones(Y.shape)
dX = label_softmax_grad(Y, dY)
arr = get_numeric_gradient(softmax_op, {"X": X}, 'Y', 'X')
numpy.testing.assert_almost_equal(arr, dX, decimal=1e-2)
unittest.main()
import paddle.v2.framework.core as core
import unittest
import numpy
import paddle.v2.framework.core as core
from paddle.v2.framework.op import Operator
......@@ -24,7 +23,7 @@ class OpTestMeta(type):
scope = core.Scope()
kwargs = dict()
places = [core.CPUPlace()]
if core.is_compile_gpu() and core.Operator.support_gpu(self.type):
if core.is_compile_gpu():
places.append(core.GPUPlace(0))
for place in places:
......@@ -53,6 +52,8 @@ class OpTestMeta(type):
kwargs[attr_name] = self.attrs[attr_name]
op = Operator(self.type, **kwargs)
if isinstance(place, core.GPUPlace) and not op.support_gpu():
return
op.infer_shape(scope)
......
import unittest
import numpy
from op_test_util import OpTestMeta
from gradient_checker import GradientChecker, create_op
class TestSGD(unittest.TestCase):
class TestCrossEntropy(unittest.TestCase):
__metaclass__ = OpTestMeta
def setUp(self):
......@@ -20,7 +21,18 @@ class TestSGD(unittest.TestCase):
self.outputs = {'Y': numpy.array(Y).astype("float32")}
# TODO(superjom) add gradient check
class CrossEntropyGradOpTest(GradientChecker):
def test_softmax_grad(self):
op = create_op("onehot_cross_entropy")
batch_size = 100
class_num = 10
inputs = {
"X": numpy.random.uniform(
0.1, 1.0, [batch_size, class_num]).astype("float32"),
"label": (class_num / 2) * numpy.ones(batch_size).astype("int32")
}
self.check_grad(op, inputs, set("X"), "Y")
if __name__ == "__main__":
unittest.main()
import unittest
import numpy as np
import paddle.v2.framework.core as core
from paddle.v2.framework.op import Operator
from gradient_checker import GradientChecker, create_op
from op_test_util import OpTestMeta
......@@ -25,62 +24,11 @@ class TestSoftmaxOp(unittest.TestCase):
}
class TestSoftmaxGradOp(unittest.TestCase):
def test_softmax_grad(self):
op = Operator('softmax', X="X", Y="Y")
backward_op = core.Operator.backward(op, set())
self.assertEqual(backward_op.type(), "softmax_grad")
expected = '''Op(softmax_grad), inputs:(X, Y, Y@GRAD), outputs:(X@GRAD).'''
self.assertEqual(expected, str(backward_op))
batch_size = 3
class_num = 5
# Initialize X and add 1e-2 for numerical stability
Y = np.random.rand(batch_size, class_num).astype(np.float32)
Y = Y + 1e-2
dY = np.random.rand(batch_size, class_num).astype(np.float32)
# Reference implementation of cross entropy with soft labels
def label_softmax_grad(Y, dY):
dX = Y * 0.0
for i in range(batch_size):
d = np.dot(Y[i, :], dY[i, :])
dX[i, :] = Y[i, :] * (dY[i, :] - d)
return dX
expected = label_softmax_grad(Y, dY)
scope = core.Scope()
places = []
places.append(core.CPUPlace())
if core.is_compile_gpu():
places.append(core.GPUPlace(0))
for place in places:
y = scope.new_var("Y")
y_tensor = y.get_tensor()
y_tensor.set_dims([batch_size, class_num])
y_tensor.alloc_float(place)
y_tensor.set(Y, place)
dy = scope.new_var("Y@GRAD")
dy_tensor = dy.get_tensor()
dy_tensor.set_dims([batch_size, class_num])
dy_tensor.alloc_float(place)
dy_tensor.set(dY, place)
x = scope.new_var("X")
dx = scope.new_var("X@GRAD")
tensor = scope.find_var("X@GRAD").get_tensor()
backward_op.infer_shape(scope)
self.assertEqual([batch_size, class_num], tensor.shape())
ctx = core.DeviceContext.create(place)
backward_op.run(scope, ctx)
actual = np.array(tensor)
np.testing.assert_almost_equal(actual, expected, decimal=3)
class SoftmaxGradOpTest(GradientChecker):
def test_softmax(self):
op = create_op("softmax")
inputs = {"X": np.random.uniform(0.1, 1, [10, 10]).astype("float32")}
self.check_grad(op, inputs, set("X"), "Y")
if __name__ == '__main__':
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册
反馈
建议
客服 返回
顶部