I ran into similar issues as @biederfrau described in #115 when running multiple calls to fast_tsne(): Independent calls incorrectly got the same input or returned incorrect output data. This happens because data.dat and results.dat are shared between calls.
I briefly discussed with @dkobak, and now appended a timestamp and some random digits to the IO-file names to make them unique. I also delete these files after retrieving the results.
linqiaozhi
commented
3 years ago
I ran into similar issues as @biederfrau described in #115 when running multiple calls to
fast_tsne(): Independent calls incorrectly got the same input or returned incorrect output data. This happens becausedata.datandresults.datare shared between calls.I briefly discussed with @dkobak, and now appended a timestamp and some random digits to the IO-file names to make them unique. I also delete these files after retrieving the results.
Let me know what you think!