Validation dataset

Dockerfile-build

@@ -171,4 +171,4 @@ ENV GIT_AUTHOR_NAME="anonymous"
 ENV GIT_AUTHOR_EMAIL="[email protected]"
 ENV GIT_COMMITTER_NAME="anonymous"
 ENV GIT_COMMITTER_EMAIL="[email protected]"
-ENV EMAIL="[email protected]"
+ENV EMAIL="[email protected]"

Makefile

@@ -289,6 +289,7 @@ run_in_docker-cpu:
 #########################################
 # TARGETS INSTALLING LIBRARIES
 #########################################
+
 # http://developer2.download.nvidia.com/compute/cuda/9.0/secure/rc/docs/sidebar/CUDA_Quick_Start_Guide.pdf?_ZyOB0PlGZzBUluXp3FtoWC-LMsTsc5H6SxIaU0i9pGNyWzZCgE-mhnAg2m66Nc3WMDvxWvvQWsXGMqr1hUliGOZvoothMTVnDe12dQQgxwS4Asjoz8XiOvPYOjV6yVQtkFhvDztUlJbNSD4srPWUU2-XegCRFII8_FIpxXERaWV
 libcuda9:
 	# wget https://developer.nvidia.com/compute/cuda/9.0/rc/local_installers/cuda-repo-ubuntu1604-9-0-local-rc_9.0.103-1_amd64-deb

ci/Jenkinsfile.template

@@ -1,3 +1,19 @@
+<<<<<<< Updated upstream:ci/Jenkinsfile.template
+=======
+// Just Notes:
+//
+//def jobnums       = [0 , 1 , 2  , 3]
+//def tags          = ["nccl" , "nonccl" , "nccl"  , "nonccl"]
+//def cudatags      = ["cuda8", "cuda8"  , "cuda9" , "cuda9"]
+//def dobuilds      = [1, 0, 0, 0]
+//def dofulltests   = [1, 0, 0, 0]
+//def dopytests     = [1, 0, 0, 0]
+//def doruntimes    = [1, 1, 1, 1]
+//def dockerimagesbuild    = ["nvidia/cuda:8.0-cudnn5-devel-centos7", "nvidia/cuda:8.0-cudnn5-devel-centos7", "nvidia/cuda:9.0-cudnn7-devel-centos7", "nvidia/cuda:9.0-cudnn7-devel-centos7"]
+//def dockerimagesruntime  = ["nvidia/cuda:8.0-cudnn5-runtime-centos7", "nvidia/cuda:8.0-cudnn5-runtime-centos7", "nvidia/cuda:9.0-cudnn7-runtime-centos7", "nvidia/cuda:9.0-cudnn7-runtime-centos7"]
+//def dists         = ["dist1","dist2","dist3","dist4"]
+
+>>>>>>> Stashed changes:Jenkinsfile.utils2
 def benchmark_commit_trigger
 
 pipeline {

src/base/ffm/ffm.cpp

@@ -17,7 +17,7 @@ FFM<T>::FFM(Params & params, T *weights) : params(params), trainer(weights, para
 
 template<typename T>
 void FFM<T>::fit(const Dataset<T> &dataset, const Dataset<T> &valid_dataset) {
-  this->trainer.setDataset(dataset);
+  this->trainer.setTrainingDataset(dataset);
   this->trainer.setValidationDataset(valid_dataset);
 
   Timer timer;

src/base/ffm/trainer.h

@@ -18,7 +18,8 @@ class Trainer {
   Trainer(const T *weights, Params &params);
   ~Trainer();
 
-  void setDataset(const Dataset<T> &dataset);
+  void setTrainingDataset(const Dataset<T> &dataset);
+  void setValidationDataset(const Dataset<T> &dataset);
 
   T validationLoss();
 

src/cpu/ffm/trainer.cpp

@@ -17,7 +17,12 @@ Trainer<T>::Trainer(const T* weights, Params &params) : params(params) {
 }
 
 template<typename T>
-void Trainer<T>::setDataset(const Dataset<T> &dataset) {
+void Trainer<T>::setTrainingDataset(const Dataset<T> &dataset) {
+  // TODO implement
+}
+
+template<typename T>
+void Trainer<T>::setValidationDataset(const Dataset<T> &dataset) {
   // TODO implement
 }
 

src/gpu/ffm/trainer.cu

@@ -69,14 +69,21 @@ Trainer<T>::Trainer(const T* weights, Params &params) : params(params), trainDat
 }
 
 template<typename T>
-void Trainer<T>::setDataset(const Dataset<T> &dataset) {
+void Trainer<T>::setTrainingDataset(const Dataset<T> &dataset) {
   DatasetBatcherGPU<T> *batcher = new DatasetBatcherGPU<T>(dataset, params);
   trainDataBatcher[0] = batcher;
 }
 
+template<typename T>
+void Trainer<T>::setValidationDataset(const Dataset<T> &dataset) {
+  DatasetBatcherGPU<T> *batcher = new DatasetBatcherGPU<T>(dataset, params);
+  validationDataBatcher[0] = batcher;
+}
+
 template<typename T>
 Trainer<T>::~Trainer() {
   delete trainDataBatcher[0];
+  delete validationDataBatcher[0];
   delete model;
   CUDA_CHECK(cudaDeviceReset());
 }

src/interface_r/vignettes/getting_started.Rmd

@@ -24,7 +24,11 @@ The R package makes use of RStudio's [reticulate](https://rstudio.github.io/reti
 
 ## Installation
 
+<<<<<<< Updated upstream
 There are a few [system requirements](https://github.com/h2oai/h2o4gpu#requirements), including Linux with glibc 2.17+, Python >=3.6, R >=3.1, CUDA 8 or 9, and a machine with Nvidia GPUs.  The code should still run if you have CPUs, but it will fall back to scikit-learn CPU based versions of the algorithms.
+=======
+There are a few [system requirements](https://github.com/h2oai/h2o4gpu#requirements), including Linux with glibc TODO MATEUSZ what glibc is required??, Python >=3.6, R >=3.1, CUDA 8 or 9, and a machine with Nvidia GPUs.  The code should still run if you have CPUs, but it will fall back to scikit-learn CPU based versions of the algorithms.
+>>>>>>> Stashed changes
 
 The **h2o4gpu** Python module is a prerequisite for the R package. So first, follow the instructions [here](https://github.com/h2oai/h2o4gpu#user-installation) to install the **h2o4gpu** Python package (either at the system level or in a Python virtual envivonment). The easiest thing to do is to `pip install` the stable release `whl` file. To ensure compatibility, the Python package version number should match the R package version number. 
 

src/swig/ch2o4gpu_cpu.i

@@ -15,4 +15,4 @@
 %include "solver/pogs.i"
 %include "solver/ffm.i"
 %include "matrix/matrix_dense.i"
-%include "metrics.i"
+%include "metrics.i"