legalforce-research / easy-segmenter

Fast and customizable rule-based sentence segmenter (WIP)
Apache License 2.0
2 stars 0 forks source link

Set bits with chunking #11

Closed kampersanda closed 2 years ago

kampersanda commented 2 years ago
waganeko/only-periods   time:   [1.9039 ms 1.9080 ms 1.9123 ms]
                        change: [-1.4656% -1.0993% -0.7241%] (p = 0.00 < 0.05)
                        Change within noise threshold.
waganeko/with-parentheses
                        time:   [3.5668 ms 3.5776 ms 3.5899 ms]
                        change: [-7.2192% -6.8281% -6.4069%] (p = 0.00 < 0.05)
                        Performance has improved.
waganeko/with-regex     time:   [2.0292 ms 2.0316 ms 2.0339 ms]
                        change: [-2.0195% -1.7824% -1.5594%] (p = 0.00 < 0.05)
                        Performance has improved.
waganeko/with-template-ja
                        time:   [5.2469 ms 5.2610 ms 5.2765 ms]
                        change: [-5.0669% -4.6997% -4.2934%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) high mild

gakumon/only-periods    time:   [439.60 µs 440.62 µs 441.50 µs]
                        change: [-2.6220% -2.1845% -1.7850%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) low severe
gakumon/with-parentheses
                        time:   [831.31 µs 833.24 µs 835.32 µs]
                        change: [-2.5493% -2.1835% -1.8049%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) high mild
gakumon/with-regex      time:   [487.22 µs 488.68 µs 490.22 µs]
                        change: [-2.4981% -2.0086% -1.5773%] (p = 0.00 < 0.05)
                        Performance has improved.
gakumon/with-template-ja
                        time:   [1.3633 ms 1.3667 ms 1.3708 ms]
                        change: [-2.1908% -1.7709% -1.3229%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 10 measurements (10.00%)
  1 (10.00%) high mild