Skip to content

P
Paddle

Crayon鑫 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
提交 bf99396a 编写于 作者: F fengjiayi
浏览文件
操作

fix errors in sequence_slice_op

上级 baa9f50d
隐藏空白更改
内联 并排
Showing with 4 addition and 8 deletion
paddle/fluid/operators/sequence_slice_op.h
浏览文件 @ bf99396a
......@@ -66,13 +66,11 @@ class SequenceSliceOpKernel : public framework::OpKernel<T> {
if (platform::is_gpu_place(ctx.GetPlace())) {
offset_cpu.mutable_data<T>(offset->dims(), platform::CPUPlace());
framework::TensorCopy(*offset, platform::CPUPlace(), ctx.device_context(),
&offset_cpu);
framework::TensorCopySync(*offset, platform::CPUPlace(), &offset_cpu);
offset_data = offset_cpu.data<int64_t>();
length_cpu.mutable_data<T>(length->dims(), platform::CPUPlace());
framework::TensorCopy(*length, platform::CPUPlace(), ctx.device_context(),
&length_cpu);
framework::TensorCopySync(*length, platform::CPUPlace(), &length_cpu);
length_data = length_cpu.data<int64_t>();
}
......@@ -127,13 +125,11 @@ class SequenceSliceGradOpKernel : public framework::OpKernel<T> {
if (platform::is_gpu_place(ctx.GetPlace())) {
offset_cpu.mutable_data<T>(offset->dims(), platform::CPUPlace());
framework::TensorCopy(*offset, platform::CPUPlace(), ctx.device_context(),
&offset_cpu);
framework::TensorCopySync(*offset, platform::CPUPlace(), &offset_cpu);
offset_data = offset_cpu.data<int64_t>();
length_cpu.mutable_data<T>(length->dims(), platform::CPUPlace());
framework::TensorCopy(*length, platform::CPUPlace(), ctx.device_context(),
&length_cpu);
framework::TensorCopySync(*length, platform::CPUPlace(), &length_cpu);
length_data = length_cpu.data<int64_t>();
}
......
Markdown is supported
0% 或 .
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 或