h2o4gpu

h2o4gpu copied to clipboard

h2o4gpu :Genetic algorithm along with Random Forest Regression produces error: terminate called after throwing an instance of 'thrust::system::system_error' what(): parallel_for failed: out of memory

1

I am working on feature selection using Genetic Algorithm (GA) with Random forest regression model (h2o4gpu.RandomForest Regressor). The number of estimators is 100, rest of the parameters are default. Here,...

Geerthy11

hemenkapadia

Tests for model (un)pickle

7

We should add tests for saving and loading pickled models (something like what we already have for XGBoost) for all algorithms (pogs based, kmeans, tsvd, pca) to see if we...

mdymczyk


pseudotensor

navdeep-G

cpp DAAL implementation - wrapper for raw pointer input

4

cpp part of DAAL implementation

stem-lab-tech

e.g. Few ways to get different errors for svd: 1) sudo mv /usr/local/cuda-9.0 /usr/local/asdf # temporarily, and then run make testquick 2) CUDA_VISIBLE_DEVICES= make testquick 3) Have cuda9.1 or some...

pseudotensor

Large Transfer time in k-means with big number of clusters

9

Hi, I run benchmark from https://github.com/h2oai/h2o4gpu/blob/master/presentations/benchmarks.pdf I use 100 iterations, and different number of clusters. I see following picture Timetransfer | Fit on GPU time, H2O, seconds 1.38588 | 9.23817...

PlayMen

For big data use cases we will need to figure out a parallelizable algo for tSVD. Ideally we want to be able to chunk the data per gpu/node. We can...

mdymczyk

make svd and kmeans like rest of tests (for make test and make testperf)

1

E.g. Currently test_tsvd_bench.py is done during all tests and (e.g.) kills my home system with only 32GB of system memory because it also uses sklearn. For tests we should only...

pseudotensor