2

I used annoy to build an index for 2 million vectors of size 1024. Here is the code that I used.

f = 1024
t = AnnoyIndex(f, 'euclidean')
t.on_disk_build('test.ann')

'''code for adding 2 million vectors'''

t.build(25)

After adding vectors, t.build(25) runs forever. ( I let it to run about six hours and it doesn't finish) However, the code works fine with 1 million vectors.

Braiam
  • 1
  • 11
  • 47
  • 78
Janaka
  • 481
  • 4
  • 14
  • I'm in a similar position and need to use Annoy for a large dataset (~1 million vectors). could you please share your experience with me and let me know how much memory it uses for such large dataset and how long it takes for the index to be created? – Soheil Mar 25 '21 at 12:56
  • https://github.com/spotify/annoy/issues/526 – Janaka Mar 28 '21 at 13:32

0 Answers0