TileLayer #2083
TileLayer #2083
Conversation
|
Oh yeah, as noted by @petergehler in #966, this can already be done by a combination of |
src/caffe/layers/tile_layer.cpp
Outdated
There was a problem hiding this comment.
INSTANTIATE_LAYER_GPU_FUNCS(TileLayer); and INSTANTIATE_CLASS(TileLayer); seem to cause duplicate explicit instantiation of Forward_gpu and Backward_gpu. I put the else block in a new tile_layer.cu and solved the problem.
TileLayer * jeffdonahue/tile-layer: Add TileLayer
77f3cf5 to
22751c5
Compare
|
This is rebased and ready for review. I also added CUDA kernels for the GPU implementation in the latter commit. |
|
@jeffdonahue this looks ready to me once it's rebased for the layer param ID since #2032 was merged. Feel free to update and merge. |
33735c4 to
cbff225
Compare
|
Rebased. Thanks for reviewing @shelhamer! |
|
Hi, would you please take a look at TileLayer Usage #4422 |
This adds a layer to duplicate an input Blob
Ntimes along any axis. AKArepmat. (This is less general than eithernp.tileorrepmatas those allow arbitrary tiling along all axes, but this satisfies a lot of use cases and could be generalized later, or even now, chained together with otherTilelayers to get the more general behavior in a less convenient and probably less efficient way.)