search

Xtra-Computing / thundersvm

ThunderSVM: A Fast SVM Library on GPUs and CPUs
Apache License 2.0
1.55k stars 215 forks source link

Problem with "max_mem_size" on Colab #247

Open fcrdossantos opened 2 years ago

fcrdossantos commented 2 years ago

I'm using the following code:

clf = SVR(kernel= 'linear',        
          max_mem_size=6000
           ) 

clf.fit(X_train,y_train)

Using Google Colab with GPU

It's always reaching the limit (12GB RAM)

There's any fix or work-around?

I'm using the last master branch (builded on Colab)