Here we show a rough benchmark of the different ways to run libspot. With a similar setup (gaussian data) we compare how many inputs the library can treat every second. The absolute values are not so relevant since they may depend on the parameters.
| Parameters | Value |
|---|---|
q |
1e-4 |
level |
0.99 |
max_excess |
200 |
#train_set |
20_000 |
#test_set |
500_000 |