Skip to content

P
Paddle

PaddlePaddle / Paddle
大约 1 年 前同步成功

通知 2298
Star 20931
Fork 5422
P
Paddle
提交 0fca1684 编写于 作者: J JiabinYang
浏览文件
操作

temp

上级 e9be3366
隐藏空白更改
内联 并排
Showing with 4 addition and 0 deletion
paddle/fluid/operators/math/matrix_bit_code.cc
浏览文件 @ 0fca1684
...@@ -102,6 +102,8 @@ void MatrixBitCodeFunctor<T>::MulGradWeight(const framework::LoDTensor& tmat, ...@@ -102,6 +102,8 @@ void MatrixBitCodeFunctor<T>::MulGradWeight(const framework::LoDTensor& tmat,
size_t input_width = input.dims()[1]; size_t input_width = input.dims()[1];
size_t tmat_width = tmat.dims()[1]; size_t tmat_width = tmat.dims()[1];
size_t weight_width = weight->dims()[1]; size_t weight_width = weight->dims()[1];
VLOG(30) << "sparse w_grad dims is [" << weight->dims()[0] << " ,"
<< weight->dims()[1] << " ]";
auto tmat_value = tmat.data<T>(); auto tmat_value = tmat.data<T>();
auto weight_value = weight->data<T>(); auto weight_value = weight->data<T>();
auto input_value = input.data<T>(); auto input_value = input.data<T>();
...@@ -127,6 +129,8 @@ void MatrixBitCodeFunctor<T>::MulGradWeight(const framework::LoDTensor& tmat, ...@@ -127,6 +129,8 @@ void MatrixBitCodeFunctor<T>::MulGradWeight(const framework::LoDTensor& tmat,
size_t input_width = input.dims()[1]; size_t input_width = input.dims()[1];
size_t tmat_width = tmat.dims()[1]; size_t tmat_width = tmat.dims()[1];
size_t weight_width = weight->value().dims()[1]; size_t weight_width = weight->value().dims()[1];
VLOG(30) << "sparse w_grad dims is: [" << weight->value().dims()[0] << " ,"
<< weight->value().dims()[1] << " ]";
auto tmat_value = tmat.data<T>(); auto tmat_value = tmat.data<T>();
auto weight_value = weight->mutable_value()->data<T>(); auto weight_value = weight->mutable_value()->data<T>();
auto input_value = input.data<T>(); auto input_value = input.data<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