Train Your Model
/cpuTrain/bin/DLtrain -m train -s NewNetwork.dat -c network_prop.txt -n 2000 -e 3 -d /Images
-m train or infer
-s save model in given file
-c Model configuration in txt file
-e number of epochs
-n number of images
DLtrain is our AI tool set which is developed in C and C++. Tested with IBM Power 9 machine by using a tool set for "ppc64el " , In similar lines X86 and ARM also handled and tested DLtrain in respective machines with x86 and ARM core. Effort went into extending the GPU by using the NVIDIA toolset called "nvcc" .
40For example, DLtrain version for Power 9 CPU is tested and details on the same is given in the following link https://www.jkuse.com/home/jkevents/baranovichi/train-dl-model/build-dltrain-in-power-9
Link https://www.jkuse.com/home/jkevents/baranovichi/train-dl-model/build-dltrain-for-jetson-nano provides information on DLtrain which is working in CPU+GPU platform ( Jetson Nano used to test DLtrain) .
DLtrain is used handling AI training workflow such as
Training NN model by using a large data set.
Training CNN model by using a large data set. handling AI inference workload
The MNIST data set is also used in the above and also a custom data set used in the above training process.
DLtrain with CPU support
DLtrain with CPU and GPU support
DLtrain with CPU, GPU and FPGA support
DLtrain deployment in IoT Node and using Inference Service ( DLtrain in Node)
DLtrain, early version of source code and associated information is given in the following