How to speed up the evaluation process

[ll0ruc]

Hi, I noticed that you are evaluating one user at each time. While my test data has hundreds of thousands users, and the evaluation will take about a few minutes. While the train on GPU is very fast. So can you provide some fast evaluation method? Thank you!

[Coder-Yu]

I have use Numba to speedup the ranking, which reduces the time cost from 20 minutes (using numpy) to one minute for datasets of 50,000 users * 100,000 items. I guess the ranking for hundreds of thousands users can be finished within 10 minutes. Currently, I have no plan for faster ranking. You can try multi-thread programming to further speedup the ranking.

[ll0ruc]

I have use Numba to speedup the ranking, which reduces the time cost from 20 minutes (using numpy) to one minute for datasets of 50,000 users * 100,000 items. I guess the ranking for hundreds of thousands users can be finished within 10 minutes. Currently, I have no plan for faster ranking. You can try multi-thread programming to further speedup the ranking.

Can you provide a code reference about Numba and numpy to speedup the ranking? Thank you

[Coder-Yu]

QRec/util/qmath.py

Line 133 in 2a13e3d

@jit(nopython=True)

You can also refer to this page.
https://switowski.com/blog/easy-speedup-wins-with-numba

[ll0ruc]

QRec/util/qmath.py

Line 133 in 2a13e3d

@jit(nopython=True)

You can also refer to this page. https://switowski.com/blog/easy-speedup-wins-with-numba

Thank you