提交 1d56d4a7 编写于 作者: E eclipsess

some bug h->w

上级 b4a5994f
...@@ -1396,7 +1396,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1396,7 +1396,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter,
input_const[in_w] * w21 + input_const[in_w] * w21 +
input_const[in_w + 1] * w22; input_const[in_w + 1] * w22;
out2in_mid = (out_h - 1) * 2; out2in_mid = (out_w - 1) * 2;
output_data_tmp[out_w - 1] = output_data_tmp[out_w - 1] =
w10 * input_const[out2in_mid - 1] + w11 * input_const[out2in_mid] + w10 * input_const[out2in_mid - 1] + w11 * input_const[out2in_mid] +
w20 * input_const[out2in_mid + in_w - 1] + w20 * input_const[out2in_mid + in_w - 1] +
...@@ -1412,7 +1412,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1412,7 +1412,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter,
w11 * input_const[out2in_mid] + w12 * input_const[out2in_mid + 1] + w11 * input_const[out2in_mid] + w12 * input_const[out2in_mid + 1] +
(1 - if_pad) * (w21 * input_const[out2in_mid + in_w] + (1 - if_pad) * (w21 * input_const[out2in_mid + in_w] +
w22 * input_const[out2in_mid + in_w + 1]); w22 * input_const[out2in_mid + in_w + 1]);
out2in_mid = (out_h - 1) * 2 * in_w + (out_h - 1) * 2; out2in_mid = (out_h - 1) * 2 * in_w + (out_w - 1) * 2;
output_data_tmp[out_h * out_w - 1] = output_data_tmp[out_h * out_w - 1] =
w00 * input_const[out2in_mid - in_w - 1] + w00 * input_const[out2in_mid - in_w - 1] +
...@@ -1438,7 +1438,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1438,7 +1438,7 @@ void DepthwiseConv3x3s2p1v2(const Tensor *input, const Tensor *filter,
w21 * input_const[out2in_mid + in_w] + w21 * input_const[out2in_mid + in_w] +
w22 * input_const[out2in_mid + in_w + 1]; w22 * input_const[out2in_mid + in_w + 1];
out2in_mid = i * 2 * in_w + (out_h - 1) * 2; out2in_mid = i * 2 * in_w + (out_w - 1) * 2;
output_data_tmp[i * out_w + out_w - 1] = output_data_tmp[i * out_w + out_w - 1] =
w00 * input_const[out2in_mid - in_w - 1] + w00 * input_const[out2in_mid - in_w - 1] +
w01 * input_const[out2in_mid - in_w] + w01 * input_const[out2in_mid - in_w] +
...@@ -1791,7 +1791,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1791,7 +1791,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter,
input_const[in_w] * w21 + input_const[in_w] * w21 +
input_const[in_w + 1] * w22; input_const[in_w + 1] * w22;
out2in_mid = (out_h - 1) * 2; out2in_mid = (out_w - 1) * 2;
output_data_tmp[out_w - 1] = output_data_tmp[out_w - 1] =
w10 * input_const[out2in_mid - 1] + w11 * input_const[out2in_mid] + w10 * input_const[out2in_mid - 1] + w11 * input_const[out2in_mid] +
w20 * input_const[out2in_mid + in_w - 1] + w20 * input_const[out2in_mid + in_w - 1] +
...@@ -1807,7 +1807,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1807,7 +1807,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter,
w11 * input_const[out2in_mid] + w12 * input_const[out2in_mid + 1] + w11 * input_const[out2in_mid] + w12 * input_const[out2in_mid + 1] +
(1 - if_pad) * (w21 * input_const[out2in_mid + in_w] + (1 - if_pad) * (w21 * input_const[out2in_mid + in_w] +
w22 * input_const[out2in_mid + in_w + 1]); w22 * input_const[out2in_mid + in_w + 1]);
out2in_mid = (out_h - 1) * 2 * in_w + (out_h - 1) * 2; out2in_mid = (out_h - 1) * 2 * in_w + (out_w - 1) * 2;
output_data_tmp[out_h * out_w - 1] = output_data_tmp[out_h * out_w - 1] =
w00 * input_const[out2in_mid - in_w - 1] + w00 * input_const[out2in_mid - in_w - 1] +
...@@ -1850,7 +1850,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter, ...@@ -1850,7 +1850,7 @@ void DepthwiseConvAddBNRelu3x3s2p1v2(const Tensor *input, const Tensor *filter,
w21 * input_const[out2in_mid + in_w] + w21 * input_const[out2in_mid + in_w] +
w22 * input_const[out2in_mid + in_w + 1]; w22 * input_const[out2in_mid + in_w + 1];
out2in_mid = i * 2 * in_w + (out_h - 1) * 2; out2in_mid = i * 2 * in_w + (out_w - 1) * 2;
output_data_tmp[i * out_w + out_w - 1] = output_data_tmp[i * out_w + out_w - 1] =
w00 * input_const[out2in_mid - in_w - 1] + w00 * input_const[out2in_mid - in_w - 1] +
w01 * input_const[out2in_mid - in_w] + w01 * input_const[out2in_mid - in_w] +
......
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