diff options
author | Evan Shelhamer <shelhamer@imaginarynumber.net> | 2015-11-10 22:47:27 -0800 |
---|---|---|
committer | Evan Shelhamer <shelhamer@imaginarynumber.net> | 2015-11-10 22:47:27 -0800 |
commit | 692295e39270eec0b7e10d2de09f777617d234f1 (patch) | |
tree | 50ed5668ce44d9b0097457543b9a420396a2263b /include | |
parent | da4d39af5b94988b0cb55653e66c43071c766168 (diff) | |
parent | 987b3d8794e3fe27b4402d52fb3921555104b451 (diff) | |
download | caffeonacl-692295e39270eec0b7e10d2de09f777617d234f1.tar.gz caffeonacl-692295e39270eec0b7e10d2de09f777617d234f1.tar.bz2 caffeonacl-692295e39270eec0b7e10d2de09f777617d234f1.zip |
Merge pull request #3295 from timmeinhardt/fix_issue_3274
[bug] fix issue #3274 -- shape argmax top carefully
Diffstat (limited to 'include')
-rw-r--r-- | include/caffe/common_layers.hpp | 14 |
1 files changed, 7 insertions, 7 deletions
diff --git a/include/caffe/common_layers.hpp b/include/caffe/common_layers.hpp index 72f39ee0..d42d15c4 100644 --- a/include/caffe/common_layers.hpp +++ b/include/caffe/common_layers.hpp @@ -53,8 +53,8 @@ class ArgMaxLayer : public Layer<Dtype> { * -# @f$ (N \times C \times H \times W) @f$ * the inputs @f$ x @f$ * @param top output Blob vector (length 1) - * -# @f$ (N \times 1 \times K \times 1) @f$ or, if out_max_val - * @f$ (N \times 2 \times K \times 1) @f$ unless axis set than e.g. + * -# @f$ (N \times 1 \times K) @f$ or, if out_max_val + * @f$ (N \times 2 \times K) @f$ unless axis set than e.g. * @f$ (N \times K \times H \times W) @f$ if axis == 1 * the computed outputs @f$ * y_n = \arg\max\limits_i x_{ni} @@ -81,13 +81,13 @@ class ArgMaxLayer : public Layer<Dtype> { * each channel in the data (i.e. axis 1), it subtracts the mean and divides * by the variance, where both statistics are computed across both spatial * dimensions and across the different examples in the batch. - * + * * By default, during training time, the network is computing global mean/ * variance statistics via a running average, which is then used at test * time to allow deterministic outputs for each input. You can manually * toggle whether the network is accumulating or using the statistics via the * use_global_stats option. IMPORTANT: for this feature to work, you MUST - * set the learning rate to zero for all three parameter blobs, i.e., + * set the learning rate to zero for all three parameter blobs, i.e., * param {lr_mult: 0} three times in the layer definition. * * Note that the original paper also included a per-channel learned bias and @@ -96,10 +96,10 @@ class ArgMaxLayer : public Layer<Dtype> { * followed by a Convolution layer with output the same size as the current. * This produces a channel-specific value that can be added or multiplied by * the BatchNorm layer's output. - * + * * [1] S. Ioffe and C. Szegedy, "Batch Normalization: Accelerating Deep Network - * Training by Reducing Internal Covariate Shift." arXiv preprint - * arXiv:1502.03167 (2015). + * Training by Reducing Internal Covariate Shift." arXiv preprint + * arXiv:1502.03167 (2015). * * TODO(dox): thorough documentation for Forward, Backward, and proto params. */ |