Quantco / slim-trees

Pickle your ML models more efficiently for deployment 🚀
MIT License
17 stars 1 forks source link

Use [!WARNING] #78

Closed jonashaag closed 1 year ago

pavelzw commented 1 year ago

Ah wait just saw that I already did this in #77

github-actions[bot] commented 1 year ago
(benchmark 5772942000 / attempt 1) Base results / Our results / Change Model Size Dump Time Load Time
sklearn rf 20M 23.4 MiB / 3.0 MiB / 7.67 x 0.03 s / 0.05 s / 1.36 x 0.03 s / 0.05 s / 1.46 x
sklearn rf 20M lzma 7.6 MiB / 2.0 MiB / 3.78 x 14.89 s / 1.21 s / 0.08 x 0.64 s / 0.21 s / 0.33 x
sklearn rf 200M 238.8 MiB / 30.8 MiB / 7.75 x 0.26 s / 0.37 s / 1.42 x 0.32 s / 0.53 s / 1.69 x
sklearn rf 200M lzma 49.8 MiB / 14.8 MiB / 3.36 x 122.76 s / 20.20 s / 0.16 x 4.77 s / 1.64 s / 0.34 x
sklearn rf 1G 1302.2 MiB / 168.0 MiB / 7.75 x 1.79 s / 2.06 s / 1.15 x 1.93 s / 2.75 s / 1.42 x
sklearn rf 1G lzma 295.6 MiB / 99.2 MiB / 2.98 x 706.45 s / 110.66 s / 0.16 x 26.47 s / 9.65 s / 0.36 x
sklearn gb 2M 2.5 MiB / 1.1 MiB / 2.16 x 0.04 s / 0.31 s / 8.29 x 0.04 s / 0.31 s / 7.34 x
sklearn gb 2M lzma 0.7 MiB / 0.2 MiB / 4.44 x 1.11 s / 0.45 s / 0.41 x 0.10 s / 0.27 s / 2.68 x
lgbm gbdt 2M 2.6 MiB / 1.0 MiB / 2.78 x 0.08 s / 0.27 s / 3.18 x 0.01 s / 0.12 s / 10.17 x
lgbm gbdt 2M lzma 0.9 MiB / 0.5 MiB / 1.90 x 1.28 s / 0.50 s / 0.39 x 0.08 s / 0.16 s / 1.96 x
lgbm gbdt 5M 5.3 MiB / 1.9 MiB / 2.81 x 0.15 s / 0.53 s / 3.49 x 0.03 s / 0.25 s / 9.88 x
lgbm gbdt 5M lzma 1.7 MiB / 0.8 MiB / 1.96 x 3.24 s / 1.07 s / 0.33 x 0.15 s / 0.32 s / 2.07 x
lgbm gbdt 20M 22.7 MiB / 7.6 MiB / 3.00 x 0.63 s / 2.06 s / 3.29 x 0.11 s / 1.05 s / 9.29 x
lgbm gbdt 20M lzma 6.3 MiB / 3.0 MiB / 2.09 x 18.39 s / 5.11 s / 0.28 x 0.62 s / 1.27 s / 2.05 x
lgbm gbdt 100M 101.1 MiB / 33.0 MiB / 3.06 x 2.70 s / 8.97 s / 3.33 x 0.50 s / 49.00 s / 98.39 x
lgbm gbdt 100M lzma 25.6 MiB / 10.6 MiB / 2.41 x 86.08 s / 24.25 s / 0.28 x 2.44 s / 5.30 s / 2.17 x
lgbm rf 10M 10.9 MiB / 3.2 MiB / 3.46 x 0.31 s / 0.65 s / 2.09 x 0.05 s / 0.41 s / 8.97 x
lgbm rf 10M lzma 0.7 MiB / 0.4 MiB / 1.86 x 1.78 s / 0.90 s / 0.51 x 0.12 s / 0.45 s / 3.65 x