experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Improve docs for tf.nn.depthwise_conv2d_native

Browse Source
This commit is contained in:
Anish Shah 2017-06-21 21:35:20 +05:30
parent b7acb6abe0
commit 63c1befb89
1 changed files with 2 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
tensorflow/core/ops/nn_ops.cc
View File

@ -831,11 +831,13 @@ a different filter to each input channel (expanding from 1 channel to
`channel_multiplier` channels for each), then concatenates the results `channel_multiplier` channels for each), then concatenates the results
together. Thus, the output has `in_channels * channel_multiplier` channels. together. Thus, the output has `in_channels * channel_multiplier` channels.
```
for k in 0..in_channels-1 for k in 0..in_channels-1
for q in 0..channel_multiplier-1 for q in 0..channel_multiplier-1
output[b, i, j, k * channel_multiplier + q] = output[b, i, j, k * channel_multiplier + q] =
sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] * sum_{di, dj} input[b, strides[1] * i + di, strides[2] * j + dj, k] *
filter[di, dj, k, q] filter[di, dj, k, q]
```
Must have `strides[0] = strides[3] = 1`. For the most common case of the same Must have `strides[0] = strides[3] = 1`. For the most common case of the same
horizontal and vertices strides, `strides = [1, stride, stride, 1]`. horizontal and vertices strides, `strides = [1, stride, stride, 1]`.