@ -10,7 +10,8 @@ We propose a novel multi-stream architecture and training methodology that explo
## Prerequisites:
1. Linux
2. Python 2 or 3
3. CPU or NVIDIA GPU + CUDA CuDNN (CUDA 8.0)
3. Pytorch version 0.4.1
4. CPU or NVIDIA GPU + CUDA CuDNN (CUDA 8.0)
## To test UMRL: