Deeplearning framework running on Spark
- Apache Hadoop YARN and Spark
- Caffe prerequisites for Worker nodes (http://caffe.berkeleyvision.org/installation.html)
- CUDA
- cuDNN
- protobuf, glog, gflags
$ mvn package
Sample running scripts are on scripts/
- runCIFAR.sh: CIFAR10 demo
- runImageNet.sh: imagenet demo
- modelConvert.sh: convert DeepSpark weight to caffemodel