mindspore-Ecosystem
/
mindspore
21
0
Fork 27
Code Issues Pull Requests 6 Projects Releases Wiki Activity

!9857 [MS][GPU] SparseApplyFtrl_int64_Support 

From: @danishnxt
Reviewed-by: @tom__chen,@robingrosman
Signed-off-by: @robingrosman
This commit is contained in:
mindspore-ci-bot 2020-12-18 03:05:54 +08:00 committed by Gitee
parent 7697492fb2 55c455df9e
commit 904f61b7fd
3 changed files with 37 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/sparse_ftrl_impl.cu
View File

@ -96,8 +96,19 @@ template void CalSparseApplyFtrl<float, int>(const float *gradient, const int *i
const float l1_regularization, const float l2_regularization,
const float learning_rate_power, const bool use_locking, float *variable,
float *accumulation, float *linear, cudaStream_t cuda_stream);
template void CalSparseApplyFtrl<float, int64_t>(const float *gradient, const int64_t *indices, const int num_index,
const size_t n_stride, const float learning_rate,
const float l1_regularization, const float l2_regularization,
const float learning_rate_power, const bool use_locking, float *variable,
float *accumulation, float *linear, cudaStream_t cuda_stream);
template void CalSparseApplyFtrl<half, int>(const half *gradient, const int *indices, const int num_index,
const size_t n_stride, const float learning_rate,
const float l1_regularization, const float l2_regularization,
const float learning_rate_power, const bool use_locking, half *variable,
half *accumulation, half *linear, cudaStream_t cuda_stream);
template void CalSparseApplyFtrl<half, int64_t>(const half *gradient, const int64_t *indices, const int num_index,
const size_t n_stride, const float learning_rate,
const float l1_regularization, const float l2_regularization,
const float learning_rate_power, const bool use_locking, half *variable,
half *accumulation, half *linear, cudaStream_t cuda_stream);

22
mindspore/ccsrc/backend/kernel_compiler/gpu/nn/sparse_ftrl_gpu_kernel.cc
View File

@ -29,6 +29,17 @@ MS_REG_GPU_KERNEL_TWO(SparseApplyFtrl,
.AddOutputAttr(kNumberTypeFloat32)
.AddOutputAttr(kNumberTypeFloat32),
SparseFtrlGpuKernel, float, int)
MS_REG_GPU_KERNEL_TWO(SparseApplyFtrl,
KernelAttr()
.AddInputAttr(kNumberTypeFloat32)
.AddInputAttr(kNumberTypeFloat32)
.AddInputAttr(kNumberTypeFloat32)
.AddInputAttr(kNumberTypeFloat32)
.AddInputAttr(kNumberTypeInt64)
.AddOutputAttr(kNumberTypeFloat32)
.AddOutputAttr(kNumberTypeFloat32)
.AddOutputAttr(kNumberTypeFloat32),
SparseFtrlGpuKernel, float, int64_t)
MS_REG_GPU_KERNEL_TWO(SparseApplyFtrl,
KernelAttr()
.AddInputAttr(kNumberTypeFloat16)
@ -40,5 +51,16 @@ MS_REG_GPU_KERNEL_TWO(SparseApplyFtrl,
.AddOutputAttr(kNumberTypeFloat16)
.AddOutputAttr(kNumberTypeFloat16),
SparseFtrlGpuKernel, half, int)
MS_REG_GPU_KERNEL_TWO(SparseApplyFtrl,
KernelAttr()
.AddInputAttr(kNumberTypeFloat16)
.AddInputAttr(kNumberTypeFloat16)
.AddInputAttr(kNumberTypeFloat16)
.AddInputAttr(kNumberTypeFloat16)
.AddInputAttr(kNumberTypeInt64)
.AddOutputAttr(kNumberTypeFloat16)
.AddOutputAttr(kNumberTypeFloat16)
.AddOutputAttr(kNumberTypeFloat16),
SparseFtrlGpuKernel, half, int64_t)
} // namespace kernel
} // namespace mindspore

8
tests/st/ops/gpu/test_sparse_apply_ftrl_op.py
View File

@ -77,9 +77,9 @@ def test_ftrl():
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_ftrl_sparse():
def test_ftrl_sparse_int64_ind():
gradient = Tensor(np.ones([2, 3, 3]).astype(np.float32))
indices = Tensor([0, 2], mstype.int32)
indices = Tensor([0, 2], mstype.int64)
expect_var = np.array([[[0.291479, 0.291479, 0.291479],
[0.291479, 0.291479, 0.291479],
[0.291479, 0.291479, 0.291479]],
@ -127,9 +127,9 @@ def test_ftrl_half():
@pytest.mark.level0
@pytest.mark.platform_x86_gpu_training
@pytest.mark.env_onecard
def test_ftrl_sparse_half():
def test_ftrl_sparse_half_int64_ind():
gradient = Tensor(np.ones([2, 3, 3]).astype(np.float16))
indices = Tensor([0, 2], mstype.int32)
indices = Tensor([0, 2], mstype.int64)
expect_var = np.array([[[0.291479, 0.291479, 0.291479],
[0.291479, 0.291479, 0.291479],
[0.291479, 0.291479, 0.291479]],