DmitryUlyanov/Multicore-TSNE

segmentation fault (core dumped) sometime

leonexu opened this issue · 2 comments

Hello

I get "segmentation fault (core dumped)" when run multicore-tsne. One of the cases that the crash occurs is when the input data contains lots of zeros. Is there any fix for this problem?
Thanks

from MulticoreTSNE import MulticoreTSNE as TSNE
import numpy as np
tsne = TSNE(n_jobs=40, perplexity=30)
tsne.fit(np.zeros([5,3]))

I have the same problem when I run with 12 cores. I have about 2 million x 2000 sparse matrix. Any resolution available?

Have the same problem. I've been using sparse vectors as well.