Frequently Asked Questions (FAQs)¶
This page is dedicated to summarizing some frequently asked questions about ThunderSVM.
FAQs of users¶
- Why does ThunderSVM have large errors? My data set is not normalized.
To reduce the errors, you need to add
-DUSE_DOUBLE=ONin cmake (e.g.,
cmake -DUSE_DOUBLE=ON ..). The reason is that ThunderSVM uses
floatto store kernel values for better efficiency. For problems require high precision, you need to tell ThunderSVM to use
- How can I use the source code? Please refere to How To page.
- What is the data format of the input file? ThunderSVM uses the LibSVM format.
- Can ThunderSVM run on CPUs? Yes. Please see Working without GPUs.
- How can I do grid search? Please refer to How To page.
FAQs of developers¶
- Why not use shrinking? Shrinking is used in ThunderSVM and is implemented by the working set size. We don’t provide the shrinking option to users, because the traditional way of shrinking is inefficient on GPUs.