claesenm / optunity

optimization routines for hyperparameter tuning
http://www.optunity.net
Other
416 stars 78 forks source link

cv.setup broken pipe error in R #83

Open alanchalk opened 7 years ago

alanchalk commented 7 years ago

I am running the following code: cv <- cv.setup(train_design, train_y, score = mean_se, num_folds = 2, num_iter = 1)

(train_design is a sparse matrix with around 250 columns My total dataset has 9m rows and I am experimenting with how the chosen hyper-parameters change as I allow more data to be used. The code works fine when there are 100,000, or 200,000 or 400,000 rows. However when I increase the number of rows to 800,000 rows I get the following error Error in readpipe(py2r) : Optunity error: broken pipe. It is not my machine running out of memory ( I checked that). I did notice even for smaller runs that cv.setup was taking a minute or two to return. I wonder if it is just timing out due to the delay? Is there anything I can do to avoid the broken pipe error? Many thanks Alan Chalk