-
Davis King authored
batch normalization. The running mean and variance need to be initialized when calling cuDNN, even if the averaging factor is 1. I don't think this was the case prior to cuDNN v5, but it certainly is the case now. This patch fixes this.
ceb6a119
| Name |
Last commit
|
Last update |
|---|---|---|
| dlib | ||
| docs | ||
| examples | ||
| python_examples | ||
| tools | ||
| .gitignore | ||
| .hgignore | ||
| .hgtags | ||
| CMakeLists.txt | ||
| MANIFEST.in | ||
| README.txt | ||
| setup.py |