-
Notifications
You must be signed in to change notification settings - Fork 81
/
Copy pathmodel_persistence.html
1206 lines (977 loc) · 76.7 KB
/
model_persistence.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!DOCTYPE html>
<html lang="en" data-content_root="./" >
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1.0" /><meta name="viewport" content="width=device-width, initial-scale=1" />
<meta property="og:title" content="9. Model persistence" />
<meta property="og:type" content="website" />
<meta property="og:url" content="https://scikit-learn/stable/model_persistence.html" />
<meta property="og:site_name" content="scikit-learn" />
<meta property="og:description" content="Summary of model persistence methods:,,, Persistence method, Pros, Risks / Cons,,, ONNX, Serve models without a Python environment, Serving and training environments independent of one another, Mos..." />
<meta property="og:image" content="https://scikit-learn.org/stable/_static/scikit-learn-logo-small.png" />
<meta property="og:image:alt" content="scikit-learn" />
<meta name="description" content="Summary of model persistence methods:,,, Persistence method, Pros, Risks / Cons,,, ONNX, Serve models without a Python environment, Serving and training environments independent of one another, Mos..." />
<title>9. Model persistence — scikit-learn 1.6.1 documentation</title>
<script data-cfasync="false">
document.documentElement.dataset.mode = localStorage.getItem("mode") || "";
document.documentElement.dataset.theme = localStorage.getItem("theme") || "";
</script>
<!--
this give us a css class that will be invisible only if js is disabled
-->
<noscript>
<style>
.pst-js-only { display: none !important; }
</style>
</noscript>
<!-- Loaded before other Sphinx assets -->
<link href="_static/styles/theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link href="_static/styles/pydata-sphinx-theme.css?digest=8878045cc6db502f8baf" rel="stylesheet" />
<link rel="stylesheet" type="text/css" href="_static/pygments.css?v=a746c00c" />
<link rel="stylesheet" type="text/css" href="_static/copybutton.css?v=76b2166b" />
<link rel="stylesheet" type="text/css" href="_static/plot_directive.css" />
<link rel="stylesheet" type="text/css" href="https://fonts.googleapis.com/css?family=Vibur" />
<link rel="stylesheet" type="text/css" href="_static/jupyterlite_sphinx.css?v=e3ca86de" />
<link rel="stylesheet" type="text/css" href="_static/sg_gallery.css?v=d2d258e8" />
<link rel="stylesheet" type="text/css" href="_static/sg_gallery-binder.css?v=f4aeca0c" />
<link rel="stylesheet" type="text/css" href="_static/sg_gallery-dataframe.css?v=2082cf3c" />
<link rel="stylesheet" type="text/css" href="_static/sg_gallery-rendered-html.css?v=1277b6f3" />
<link rel="stylesheet" type="text/css" href="_static/sphinx-design.min.css?v=95c83b7e" />
<link rel="stylesheet" type="text/css" href="_static/styles/colors.css?v=cc94ab7d" />
<link rel="stylesheet" type="text/css" href="_static/styles/custom.css?v=d67e4bb0" />
<!-- So that users can add custom icons -->
<script src="_static/scripts/fontawesome.js?digest=8878045cc6db502f8baf"></script>
<!-- Pre-loaded scripts that we'll load fully later -->
<link rel="preload" as="script" href="_static/scripts/bootstrap.js?digest=8878045cc6db502f8baf" />
<link rel="preload" as="script" href="_static/scripts/pydata-sphinx-theme.js?digest=8878045cc6db502f8baf" />
<script src="_static/documentation_options.js?v=d6a008b6"></script>
<script src="_static/doctools.js?v=9a2dae69"></script>
<script src="_static/sphinx_highlight.js?v=dc90522c"></script>
<script src="_static/clipboard.min.js?v=a7894cd8"></script>
<script src="_static/copybutton.js?v=97f0b27d"></script>
<script src="_static/jupyterlite_sphinx.js?v=d6bdf5f8"></script>
<script src="_static/design-tabs.js?v=f930bc37"></script>
<script data-domain="scikit-learn.org" defer="defer" src="https://views.scientific-python.org/js/script.js"></script>
<script>DOCUMENTATION_OPTIONS.pagename = 'model_persistence';</script>
<script>
DOCUMENTATION_OPTIONS.theme_version = '0.16.1';
DOCUMENTATION_OPTIONS.theme_switcher_json_url = 'https://scikit-learn.org/dev/_static/versions.json';
DOCUMENTATION_OPTIONS.theme_switcher_version_match = '1.6.1';
DOCUMENTATION_OPTIONS.show_version_warning_banner =
true;
</script>
<script src="_static/scripts/dropdown.js?v=e2048168"></script>
<script src="_static/scripts/version-switcher.js?v=a6dd8357"></script>
<script src="_static/scripts/sg_plotly_resize.js?v=eeb41cab"></script>
<link rel="canonical" href="https://scikit-learn.org/stable/model_persistence.html" />
<link rel="icon" href="_static/favicon.ico"/>
<link rel="author" title="About these documents" href="about.html" />
<link rel="search" title="Search" href="search.html" />
<link rel="next" title="10. Common pitfalls and recommended practices" href="common_pitfalls.html" />
<link rel="prev" title="8.3. Parallelism, resource management, and configuration" href="computing/parallelism.html" />
<meta name="viewport" content="width=device-width, initial-scale=1"/>
<meta name="docsearch:language" content="en"/>
<meta name="docsearch:version" content="1.6" />
</head>
<body data-bs-spy="scroll" data-bs-target=".bd-toc-nav" data-offset="180" data-bs-root-margin="0px 0px -60%" data-default-mode="">
<div id="pst-skip-link" class="skip-link d-print-none"><a href="#main-content">Skip to main content</a></div>
<div id="pst-scroll-pixel-helper"></div>
<button type="button" class="btn rounded-pill" id="pst-back-to-top">
<i class="fa-solid fa-arrow-up"></i>Back to top</button>
<dialog id="pst-search-dialog">
<form class="bd-search d-flex align-items-center"
action="search.html"
method="get">
<i class="fa-solid fa-magnifying-glass"></i>
<input type="search"
class="form-control"
name="q"
placeholder="Search the docs ..."
aria-label="Search the docs ..."
autocomplete="off"
autocorrect="off"
autocapitalize="off"
spellcheck="false"/>
<span class="search-button__kbd-shortcut"><kbd class="kbd-shortcut__modifier">Ctrl</kbd>+<kbd>K</kbd></span>
</form>
</dialog>
<div class="pst-async-banner-revealer d-none">
<aside id="bd-header-version-warning" class="d-none d-print-none" aria-label="Version warning"></aside>
</div>
<header class="bd-header navbar navbar-expand-lg bd-navbar d-print-none">
<div class="bd-header__inner bd-page-width">
<button class="pst-navbar-icon sidebar-toggle primary-toggle" aria-label="Site navigation">
<span class="fa-solid fa-bars"></span>
</button>
<div class=" navbar-header-items__start">
<div class="navbar-item">
<a class="navbar-brand logo" href="index.html">
<img src="_static/scikit-learn-logo-small.png" class="logo__image only-light" alt="scikit-learn homepage"/>
<img src="_static/scikit-learn-logo-small.png" class="logo__image only-dark pst-js-only" alt="scikit-learn homepage"/>
</a></div>
</div>
<div class=" navbar-header-items">
<div class="me-auto navbar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="install.html">
Install
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="user_guide.html">
User Guide
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="api/index.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="auto_examples/index.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://blog.scikit-learn.org/">
Community
</a>
</li>
<li class="nav-item dropdown">
<button class="btn dropdown-toggle nav-item" type="button"
data-bs-toggle="dropdown" aria-expanded="false"
aria-controls="pst-nav-more-links">
More
</button>
<ul id="pst-nav-more-links" class="dropdown-menu">
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="getting_started.html">
Getting Started
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="whats_new.html">
Release History
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="glossary.html">
Glossary
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-external" href="https://scikit-learn.org/dev/developers/index.html">
Development
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="faq.html">
FAQ
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="support.html">
Support
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="related_projects.html">
Related Projects
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="roadmap.html">
Roadmap
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="governance.html">
Governance
</a>
</li>
<li class=" ">
<a class="nav-link dropdown-item nav-internal" href="about.html">
About us
</a>
</li>
</ul>
</li>
</ul>
</nav></div>
</div>
<div class="navbar-header-items__end">
<div class="navbar-item navbar-persistent--container">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/scikit-learn/scikit-learn" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-2"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-2"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-2"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-2">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
</div>
<div class="navbar-persistent--mobile">
<button class="btn btn-sm pst-navbar-icon search-button search-button__button pst-js-only" title="Search" aria-label="Search" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="fa-solid fa-magnifying-glass fa-lg"></i>
</button>
</div>
<button class="pst-navbar-icon sidebar-toggle secondary-toggle" aria-label="On this page">
<span class="fa-solid fa-outdent"></span>
</button>
</div>
</header>
<div class="bd-container">
<div class="bd-container__inner bd-page-width">
<dialog id="pst-primary-sidebar-modal"></dialog>
<div id="pst-primary-sidebar" class="bd-sidebar-primary bd-sidebar">
<div class="sidebar-header-items sidebar-primary__section">
<div class="sidebar-header-items__center">
<div class="navbar-item">
<nav>
<ul class="bd-navbar-elements navbar-nav">
<li class="nav-item ">
<a class="nav-link nav-internal" href="install.html">
Install
</a>
</li>
<li class="nav-item current active">
<a class="nav-link nav-internal" href="user_guide.html">
User Guide
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="api/index.html">
API
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="auto_examples/index.html">
Examples
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://blog.scikit-learn.org/">
Community
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="getting_started.html">
Getting Started
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="whats_new.html">
Release History
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="glossary.html">
Glossary
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-external" href="https://scikit-learn.org/dev/developers/index.html">
Development
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="faq.html">
FAQ
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="support.html">
Support
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="related_projects.html">
Related Projects
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="roadmap.html">
Roadmap
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="governance.html">
Governance
</a>
</li>
<li class="nav-item ">
<a class="nav-link nav-internal" href="about.html">
About us
</a>
</li>
</ul>
</nav></div>
</div>
<div class="sidebar-header-items__end">
<div class="navbar-item">
<button class="btn btn-sm nav-link pst-navbar-icon theme-switch-button pst-js-only" aria-label="Color mode" data-bs-title="Color mode" data-bs-placement="bottom" data-bs-toggle="tooltip">
<i class="theme-switch fa-solid fa-sun fa-lg" data-mode="light" title="Light"></i>
<i class="theme-switch fa-solid fa-moon fa-lg" data-mode="dark" title="Dark"></i>
<i class="theme-switch fa-solid fa-circle-half-stroke fa-lg" data-mode="auto" title="System Settings"></i>
</button></div>
<div class="navbar-item"><ul class="navbar-icon-links"
aria-label="Icon Links">
<li class="nav-item">
<a href="https://github.com/scikit-learn/scikit-learn" title="GitHub" class="nav-link pst-navbar-icon" rel="noopener" target="_blank" data-bs-toggle="tooltip" data-bs-placement="bottom"><i class="fa-brands fa-square-github fa-lg" aria-hidden="true"></i>
<span class="sr-only">GitHub</span></a>
</li>
</ul></div>
<div class="navbar-item">
<div class="version-switcher__container dropdown pst-js-only">
<button id="pst-version-switcher-button-3"
type="button"
class="version-switcher__button btn btn-sm dropdown-toggle"
data-bs-toggle="dropdown"
aria-haspopup="listbox"
aria-controls="pst-version-switcher-list-3"
aria-label="Version switcher list"
>
Choose version <!-- this text may get changed later by javascript -->
<span class="caret"></span>
</button>
<div id="pst-version-switcher-list-3"
class="version-switcher__menu dropdown-menu list-group-flush py-0"
role="listbox" aria-labelledby="pst-version-switcher-button-3">
<!-- dropdown will be populated by javascript on page load -->
</div>
</div></div>
</div>
</div>
<div class="sidebar-primary-items__start sidebar-primary__section">
<div class="sidebar-primary-item">
<nav class="bd-docs-nav bd-links"
aria-label="Section Navigation">
<p class="bd-links__title" role="heading" aria-level="1">Section Navigation</p>
<div class="bd-toc-item navbar-nav"><ul class="current nav bd-sidenav">
<li class="toctree-l1 has-children"><a class="reference internal" href="supervised_learning.html">1. Supervised learning</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/linear_model.html">1.1. Linear Models</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/lda_qda.html">1.2. Linear and Quadratic Discriminant Analysis</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/kernel_ridge.html">1.3. Kernel ridge regression</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/svm.html">1.4. Support Vector Machines</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/sgd.html">1.5. Stochastic Gradient Descent</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/neighbors.html">1.6. Nearest Neighbors</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/gaussian_process.html">1.7. Gaussian Processes</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/cross_decomposition.html">1.8. Cross decomposition</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/naive_bayes.html">1.9. Naive Bayes</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/tree.html">1.10. Decision Trees</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/ensemble.html">1.11. Ensembles: Gradient boosting, random forests, bagging, voting, stacking</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/multiclass.html">1.12. Multiclass and multioutput algorithms</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/feature_selection.html">1.13. Feature selection</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/semi_supervised.html">1.14. Semi-supervised learning</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/isotonic.html">1.15. Isotonic regression</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/calibration.html">1.16. Probability calibration</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/neural_networks_supervised.html">1.17. Neural network models (supervised)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="unsupervised_learning.html">2. Unsupervised learning</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/mixture.html">2.1. Gaussian mixture models</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/manifold.html">2.2. Manifold learning</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/clustering.html">2.3. Clustering</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/biclustering.html">2.4. Biclustering</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/decomposition.html">2.5. Decomposing signals in components (matrix factorization problems)</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/covariance.html">2.6. Covariance estimation</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/outlier_detection.html">2.7. Novelty and Outlier Detection</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/density.html">2.8. Density Estimation</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/neural_networks_unsupervised.html">2.9. Neural network models (unsupervised)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="model_selection.html">3. Model selection and evaluation</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/cross_validation.html">3.1. Cross-validation: evaluating estimator performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/grid_search.html">3.2. Tuning the hyper-parameters of an estimator</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/classification_threshold.html">3.3. Tuning the decision threshold for class prediction</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/model_evaluation.html">3.4. Metrics and scoring: quantifying the quality of predictions</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/learning_curve.html">3.5. Validation curves: plotting scores to evaluate models</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="inspection.html">4. Inspection</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/partial_dependence.html">4.1. Partial Dependence and Individual Conditional Expectation plots</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/permutation_importance.html">4.2. Permutation feature importance</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="visualizations.html">5. Visualizations</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="data_transforms.html">6. Dataset transformations</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/compose.html">6.1. Pipelines and composite estimators</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/feature_extraction.html">6.2. Feature extraction</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/preprocessing.html">6.3. Preprocessing data</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/impute.html">6.4. Imputation of missing values</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/unsupervised_reduction.html">6.5. Unsupervised dimensionality reduction</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/random_projection.html">6.6. Random Projection</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/kernel_approximation.html">6.7. Kernel Approximation</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/metrics.html">6.8. Pairwise metrics, Affinities and Kernels</a></li>
<li class="toctree-l2"><a class="reference internal" href="modules/preprocessing_targets.html">6.9. Transforming the prediction target (<code class="docutils literal notranslate"><span class="pre">y</span></code>)</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="datasets.html">7. Dataset loading utilities</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="datasets/toy_dataset.html">7.1. Toy datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="datasets/real_world.html">7.2. Real world datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="datasets/sample_generators.html">7.3. Generated datasets</a></li>
<li class="toctree-l2"><a class="reference internal" href="datasets/loading_other_datasets.html">7.4. Loading other datasets</a></li>
</ul>
</details></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="computing.html">8. Computing with scikit-learn</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="computing/scaling_strategies.html">8.1. Strategies to scale computationally: bigger data</a></li>
<li class="toctree-l2"><a class="reference internal" href="computing/computational_performance.html">8.2. Computational Performance</a></li>
<li class="toctree-l2"><a class="reference internal" href="computing/parallelism.html">8.3. Parallelism, resource management, and configuration</a></li>
</ul>
</details></li>
<li class="toctree-l1 current active"><a class="current reference internal" href="#">9. Model persistence</a></li>
<li class="toctree-l1"><a class="reference internal" href="common_pitfalls.html">10. Common pitfalls and recommended practices</a></li>
<li class="toctree-l1 has-children"><a class="reference internal" href="dispatching.html">11. Dispatching</a><details><summary><span class="toctree-toggle" role="presentation"><i class="fa-solid fa-chevron-down"></i></span></summary><ul>
<li class="toctree-l2"><a class="reference internal" href="modules/array_api.html">11.1. Array API support (experimental)</a></li>
</ul>
</details></li>
<li class="toctree-l1"><a class="reference internal" href="machine_learning_map.html">12. Choosing the right estimator</a></li>
<li class="toctree-l1"><a class="reference internal" href="presentations.html">13. External Resources, Videos and Talks</a></li>
</ul>
</div>
</nav></div>
</div>
<div class="sidebar-primary-items__end sidebar-primary__section">
</div>
</div>
<main id="main-content" class="bd-main" role="main">
<div class="bd-content">
<div class="bd-article-container">
<div class="bd-header-article d-print-none">
<div class="header-article-items header-article__inner">
<div class="header-article-items__start">
<div class="header-article-item">
<nav aria-label="Breadcrumb" class="d-print-none">
<ul class="bd-breadcrumbs">
<li class="breadcrumb-item breadcrumb-home">
<a href="index.html" class="nav-link" aria-label="Home">
<i class="fa-solid fa-home"></i>
</a>
</li>
<li class="breadcrumb-item"><a href="user_guide.html" class="nav-link">User Guide</a></li>
<li class="breadcrumb-item active" aria-current="page"><span class="ellipsis"><span class="section-number">9. </span>Model persistence</span></li>
</ul>
</nav>
</div>
</div>
</div>
</div>
<div id="searchbox"></div>
<article class="bd-article">
<section id="model-persistence">
<span id="id1"></span><h1><span class="section-number">9. </span>Model persistence<a class="headerlink" href="#model-persistence" title="Link to this heading">#</a></h1>
<div class="pst-scrollable-table-container"><table class="table" id="id2">
<caption><span class="caption-text">Summary of model persistence methods</span><a class="headerlink" href="#id2" title="Link to this table">#</a></caption>
<colgroup>
<col style="width: 20.0%" />
<col style="width: 40.0%" />
<col style="width: 40.0%" />
</colgroup>
<thead>
<tr class="row-odd"><th class="head"><p>Persistence method</p></th>
<th class="head"><p>Pros</p></th>
<th class="head"><p>Risks / Cons</p></th>
</tr>
</thead>
<tbody>
<tr class="row-even"><td><p><a class="reference internal" href="#onnx-persistence"><span class="std std-ref">ONNX</span></a></p></td>
<td><ul class="simple">
<li><p>Serve models without a Python environment</p></li>
<li><p>Serving and training environments independent of one another</p></li>
<li><p>Most secure option</p></li>
</ul>
</td>
<td><ul class="simple">
<li><p>Not all scikit-learn models are supported</p></li>
<li><p>Custom estimators require more work to support</p></li>
<li><p>Original Python object is lost and cannot be reconstructed</p></li>
</ul>
</td>
</tr>
<tr class="row-odd"><td><p><a class="reference internal" href="#skops-persistence"><span class="std std-ref">skops.io</span></a></p></td>
<td><ul class="simple">
<li><p>More secure than <code class="docutils literal notranslate"><span class="pre">pickle</span></code> based formats</p></li>
<li><p>Contents can be partly validated without loading</p></li>
</ul>
</td>
<td><ul class="simple">
<li><p>Not as fast as <code class="docutils literal notranslate"><span class="pre">pickle</span></code> based formats</p></li>
<li><p>Supports less types than <code class="docutils literal notranslate"><span class="pre">pickle</span></code> based formats</p></li>
<li><p>Requires the same environment as the training environment</p></li>
</ul>
</td>
</tr>
<tr class="row-even"><td><p><a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a></p></td>
<td><ul class="simple">
<li><p>Native to Python</p></li>
<li><p>Can serialize most Python objects</p></li>
<li><p>Efficient memory usage with <code class="docutils literal notranslate"><span class="pre">protocol=5</span></code></p></li>
</ul>
</td>
<td><ul class="simple">
<li><p>Loading can execute arbitrary code</p></li>
<li><p>Requires the same environment as the training environment</p></li>
</ul>
</td>
</tr>
<tr class="row-odd"><td><p><a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a></p></td>
<td><ul class="simple">
<li><p>Efficient memory usage</p></li>
<li><p>Supports memory mapping</p></li>
<li><p>Easy shortcuts for compression and decompression</p></li>
</ul>
</td>
<td><ul class="simple">
<li><p>Pickle based format</p></li>
<li><p>Loading can execute arbitrary code</p></li>
<li><p>Requires the same environment as the training environment</p></li>
</ul>
</td>
</tr>
<tr class="row-even"><td><p><a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a></p></td>
<td><ul class="simple">
<li><p>Can serialize non-packaged, custom Python code</p></li>
<li><p>Comparable loading efficiency as <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> with <code class="docutils literal notranslate"><span class="pre">protocol=5</span></code></p></li>
</ul>
</td>
<td><ul class="simple">
<li><p>Pickle based format</p></li>
<li><p>Loading can execute arbitrary code</p></li>
<li><p>No forward compatibility guarantees</p></li>
<li><p>Requires the same environment as the training environment</p></li>
</ul>
</td>
</tr>
</tbody>
</table>
</div>
<p>After training a scikit-learn model, it is desirable to have a way to persist
the model for future use without having to retrain. Based on your use-case,
there are a few different ways to persist a scikit-learn model, and here we
help you decide which one suits you best. In order to make a decision, you need
to answer the following questions:</p>
<ol class="arabic simple">
<li><p>Do you need the Python object after persistence, or do you only need to
persist in order to serve the model and get predictions out of it?</p></li>
</ol>
<p>If you only need to serve the model and no further investigation on the Python
object itself is required, then <a class="reference internal" href="#onnx-persistence"><span class="std std-ref">ONNX</span></a> might be the
best fit for you. Note that not all models are supported by ONNX.</p>
<p>In case ONNX is not suitable for your use-case, the next question is:</p>
<ol class="arabic simple" start="2">
<li><p>Do you absolutely trust the source of the model, or are there any security
concerns regarding where the persisted model comes from?</p></li>
</ol>
<p>If you have security concerns, then you should consider using <a class="reference internal" href="#skops-persistence"><span class="std std-ref">skops.io</span></a> which gives you back the Python object, but unlike
<code class="docutils literal notranslate"><span class="pre">pickle</span></code> based persistence solutions, loading the persisted model doesn’t
automatically allow arbitrary code execution. Note that this requires manual
investigation of the persisted file, which <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#module-skops.io" title="(in skops)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">skops.io</span></code></a> allows you to do.</p>
<p>The other solutions assume you absolutely trust the source of the file to be
loaded, as they are all susceptible to arbitrary code execution upon loading
the persisted file since they all use the pickle protocol under the hood.</p>
<ol class="arabic simple" start="3">
<li><p>Do you care about the performance of loading the model, and sharing it
between processes where a memory mapped object on disk is beneficial?</p></li>
</ol>
<p>If yes, then you can consider using <a class="reference internal" href="#pickle-persistence"><span class="std std-ref">joblib</span></a>. If this
is not a major concern for you, then you can use the built-in <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a>
module.</p>
<ol class="arabic simple" start="4">
<li><p>Did you try <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> or <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a> and found that the model cannot
be persisted? It can happen for instance when you have user defined
functions in your model.</p></li>
</ol>
<p>If yes, then you can use <a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a> which can serialize certain objects
which cannot be serialized by <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> or <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a>.</p>
<section id="workflow-overview">
<h2><span class="section-number">9.1. </span>Workflow Overview<a class="headerlink" href="#workflow-overview" title="Link to this heading">#</a></h2>
<p>In a typical workflow, the first step is to train the model using scikit-learn
and scikit-learn compatible libraries. Note that support for scikit-learn and
third party estimators varies across the different persistence methods.</p>
<section id="train-and-persist-the-model">
<h3><span class="section-number">9.1.1. </span>Train and Persist the Model<a class="headerlink" href="#train-and-persist-the-model" title="Link to this heading">#</a></h3>
<p>Creating an appropriate model depends on your use-case. As an example, here we
train a <a class="reference internal" href="modules/generated/sklearn.ensemble.HistGradientBoostingClassifier.html#sklearn.ensemble.HistGradientBoostingClassifier" title="sklearn.ensemble.HistGradientBoostingClassifier"><code class="xref py py-class docutils literal notranslate"><span class="pre">sklearn.ensemble.HistGradientBoostingClassifier</span></code></a> on the iris
dataset:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn</span> <span class="kn">import</span> <span class="n">ensemble</span>
<span class="gp">>>> </span><span class="kn">from</span> <span class="nn">sklearn</span> <span class="kn">import</span> <span class="n">datasets</span>
<span class="gp">>>> </span><span class="n">clf</span> <span class="o">=</span> <span class="n">ensemble</span><span class="o">.</span><span class="n">HistGradientBoostingClassifier</span><span class="p">()</span>
<span class="gp">>>> </span><span class="n">X</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="n">datasets</span><span class="o">.</span><span class="n">load_iris</span><span class="p">(</span><span class="n">return_X_y</span><span class="o">=</span><span class="kc">True</span><span class="p">)</span>
<span class="gp">>>> </span><span class="n">clf</span><span class="o">.</span><span class="n">fit</span><span class="p">(</span><span class="n">X</span><span class="p">,</span> <span class="n">y</span><span class="p">)</span>
<span class="go">HistGradientBoostingClassifier()</span>
</pre></div>
</div>
<p>Once the model is trained, you can persist it using your desired method, and
then you can load the model in a separate environment and get predictions from
it given input data. Here there are two major paths depending on how you
persist and plan to serve the model:</p>
<ul class="simple">
<li><p><a class="reference internal" href="#onnx-persistence"><span class="std std-ref">ONNX</span></a>: You need an <code class="docutils literal notranslate"><span class="pre">ONNX</span></code> runtime and an environment
with appropriate dependencies installed to load the model and use the runtime
to get predictions. This environment can be minimal and does not necessarily
even require Python to be installed to load the model and compute
predictions. Also note that <code class="docutils literal notranslate"><span class="pre">onnxruntime</span></code> typically requires much less RAM
than Python to compute predictions from small models.</p></li>
<li><p><a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#module-skops.io" title="(in skops)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">skops.io</span></code></a>, <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a>, <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a>, <a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a>: You need a
Python environment with the appropriate dependencies installed to load the
model and get predictions from it. This environment should have the same
<strong>packages</strong> and the same <strong>versions</strong> as the environment where the model was
trained. Note that none of these methods support loading a model trained with
a different version of scikit-learn, and possibly different versions of other
dependencies such as <code class="docutils literal notranslate"><span class="pre">numpy</span></code> and <code class="docutils literal notranslate"><span class="pre">scipy</span></code>. Another concern would be running
the persisted model on a different hardware, and in most cases you should be
able to load your persisted model on a different hardware.</p></li>
</ul>
</section>
</section>
<section id="onnx">
<span id="onnx-persistence"></span><h2><span class="section-number">9.2. </span>ONNX<a class="headerlink" href="#onnx" title="Link to this heading">#</a></h2>
<p><code class="docutils literal notranslate"><span class="pre">ONNX</span></code>, or <a class="reference external" href="https://onnx.ai/">Open Neural Network Exchange</a> format is best
suitable in use-cases where one needs to persist the model and then use the
persisted artifact to get predictions without the need to load the Python
object itself. It is also useful in cases where the serving environment needs
to be lean and minimal, since the <code class="docutils literal notranslate"><span class="pre">ONNX</span></code> runtime does not require <code class="docutils literal notranslate"><span class="pre">python</span></code>.</p>
<p><code class="docutils literal notranslate"><span class="pre">ONNX</span></code> is a binary serialization of the model. It has been developed to improve
the usability of the interoperable representation of data models. It aims to
facilitate the conversion of the data models between different machine learning
frameworks, and to improve their portability on different computing
architectures. More details are available from the <a class="reference external" href="https://onnx.ai/get-started.html">ONNX tutorial</a>. To convert scikit-learn model to <code class="docutils literal notranslate"><span class="pre">ONNX</span></code>
<a class="reference external" href="http://onnx.ai/sklearn-onnx/">sklearn-onnx</a> has been developed. However,
not all scikit-learn models are supported, and it is limited to the core
scikit-learn and does not support most third party estimators. One can write a
custom converter for third party or custom estimators, but the documentation to
do that is sparse and it might be challenging to do so.</p>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="using-onnx">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">Using ONNX<a class="headerlink" href="#using-onnx" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<p class="sd-card-text">To convert the model to <code class="docutils literal notranslate"><span class="pre">ONNX</span></code> format, you need to give the converter some
information about the input as well, about which you can read more <a class="reference external" href="http://onnx.ai/sklearn-onnx/index.html">here</a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">skl2onnx</span> <span class="kn">import</span> <span class="n">to_onnx</span>
<span class="n">onx</span> <span class="o">=</span> <span class="n">to_onnx</span><span class="p">(</span><span class="n">clf</span><span class="p">,</span> <span class="n">X</span><span class="p">[:</span><span class="mi">1</span><span class="p">]</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="n">numpy</span><span class="o">.</span><span class="n">float32</span><span class="p">),</span> <span class="n">target_opset</span><span class="o">=</span><span class="mi">12</span><span class="p">)</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="s2">"filename.onnx"</span><span class="p">,</span> <span class="s2">"wb"</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">f</span><span class="o">.</span><span class="n">write</span><span class="p">(</span><span class="n">onx</span><span class="o">.</span><span class="n">SerializeToString</span><span class="p">())</span>
</pre></div>
</div>
<p class="sd-card-text">You can load the model in Python and use the <code class="docutils literal notranslate"><span class="pre">ONNX</span></code> runtime to get
predictions:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">onnxruntime</span> <span class="kn">import</span> <span class="n">InferenceSession</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="s2">"filename.onnx"</span><span class="p">,</span> <span class="s2">"rb"</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">onx</span> <span class="o">=</span> <span class="n">f</span><span class="o">.</span><span class="n">read</span><span class="p">()</span>
<span class="n">sess</span> <span class="o">=</span> <span class="n">InferenceSession</span><span class="p">(</span><span class="n">onx</span><span class="p">,</span> <span class="n">providers</span><span class="o">=</span><span class="p">[</span><span class="s2">"CPUExecutionProvider"</span><span class="p">])</span>
<span class="n">pred_ort</span> <span class="o">=</span> <span class="n">sess</span><span class="o">.</span><span class="n">run</span><span class="p">(</span><span class="kc">None</span><span class="p">,</span> <span class="p">{</span><span class="s2">"X"</span><span class="p">:</span> <span class="n">X_test</span><span class="o">.</span><span class="n">astype</span><span class="p">(</span><span class="n">numpy</span><span class="o">.</span><span class="n">float32</span><span class="p">)})[</span><span class="mi">0</span><span class="p">]</span>
</pre></div>
</div>
</div>
</details></section>
<section id="skops-io">
<span id="skops-persistence"></span><h2><span class="section-number">9.3. </span><code class="docutils literal notranslate"><span class="pre">skops.io</span></code><a class="headerlink" href="#skops-io" title="Link to this heading">#</a></h2>
<p><a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#module-skops.io" title="(in skops)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">skops.io</span></code></a> avoids using <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> and only loads files which have types
and references to functions which are trusted either by default or by the user.
Therefore it provides a more secure format than <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a>, <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a>,
and <a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a>.</p>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="using-skops">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">Using skops<a class="headerlink" href="#using-skops" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<p class="sd-card-text">The API is very similar to <a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a>, and you can persist your models as
explained in the <a class="reference external" href="https://skops.readthedocs.io/en/stable/persistence.html">documentation</a> using
<a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.dump" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.dump</span></code></a> and <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.dumps" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.dumps</span></code></a>:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">import</span> <span class="nn">skops.io</span> <span class="k">as</span> <span class="nn">sio</span>
<span class="n">obj</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">dump</span><span class="p">(</span><span class="n">clf</span><span class="p">,</span> <span class="s2">"filename.skops"</span><span class="p">)</span>
</pre></div>
</div>
<p class="sd-card-text">And you can load them back using <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.load" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.load</span></code></a> and
<a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.loads" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.loads</span></code></a>. However, you need to specify the types which are
trusted by you. You can get existing unknown types in a dumped object / file
using <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#skops.io.get_untrusted_types" title="(in skops)"><code class="xref py py-func docutils literal notranslate"><span class="pre">skops.io.get_untrusted_types</span></code></a>, and after checking its contents,
pass it to the load function:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">unknown_types</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">get_untrusted_types</span><span class="p">(</span><span class="n">file</span><span class="o">=</span><span class="s2">"filename.skops"</span><span class="p">)</span>
<span class="c1"># investigate the contents of unknown_types, and only load if you trust</span>
<span class="c1"># everything you see.</span>
<span class="n">clf</span> <span class="o">=</span> <span class="n">sio</span><span class="o">.</span><span class="n">load</span><span class="p">(</span><span class="s2">"filename.skops"</span><span class="p">,</span> <span class="n">trusted</span><span class="o">=</span><span class="n">unknown_types</span><span class="p">)</span>
</pre></div>
</div>
<p class="sd-card-text">Please report issues and feature requests related to this format on the <a class="reference external" href="https://github.com/skops-dev/skops/issues">skops
issue tracker</a>.</p>
</div>
</details></section>
<section id="pickle-joblib-and-cloudpickle">
<span id="pickle-persistence"></span><h2><span class="section-number">9.4. </span><code class="docutils literal notranslate"><span class="pre">pickle</span></code>, <code class="docutils literal notranslate"><span class="pre">joblib</span></code>, and <code class="docutils literal notranslate"><span class="pre">cloudpickle</span></code><a class="headerlink" href="#pickle-joblib-and-cloudpickle" title="Link to this heading">#</a></h2>
<p>These three modules / packages, use the <code class="docutils literal notranslate"><span class="pre">pickle</span></code> protocol under the hood, but
come with slight variations:</p>
<ul class="simple">
<li><p><a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> is a module from the Python Standard Library. It can serialize
and deserialize any Python object, including custom Python classes and
objects.</p></li>
<li><p><a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a> is more efficient than <code class="docutils literal notranslate"><span class="pre">pickle</span></code> when working with large machine
learning models or large numpy arrays.</p></li>
<li><p><a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a> can serialize certain objects which cannot be serialized by
<a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> or <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a>, such as user defined functions and lambda
functions. This can happen for instance, when using a
<a class="reference internal" href="modules/generated/sklearn.preprocessing.FunctionTransformer.html#sklearn.preprocessing.FunctionTransformer" title="sklearn.preprocessing.FunctionTransformer"><code class="xref py py-class docutils literal notranslate"><span class="pre">FunctionTransformer</span></code></a> and using a custom
function to transform the data.</p></li>
</ul>
<details class="sd-sphinx-override sd-dropdown sd-card sd-mb-3" id="using-pickle,-joblib,-or-cloudpickle">
<summary class="sd-summary-title sd-card-header">
<span class="sd-summary-text">Using <code class="docutils literal notranslate"><span class="pre">pickle</span></code>, <code class="docutils literal notranslate"><span class="pre">joblib</span></code>, or <code class="docutils literal notranslate"><span class="pre">cloudpickle</span></code><a class="headerlink" href="#using-pickle,-joblib,-or-cloudpickle" title="Link to this dropdown">#</a></span><span class="sd-summary-state-marker sd-summary-chevron-right"><svg version="1.1" width="1.5em" height="1.5em" class="sd-octicon sd-octicon-chevron-right" viewBox="0 0 24 24" aria-hidden="true"><path d="M8.72 18.78a.75.75 0 0 1 0-1.06L14.44 12 8.72 6.28a.751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018l6.25 6.25a.75.75 0 0 1 0 1.06l-6.25 6.25a.75.75 0 0 1-1.06 0Z"></path></svg></span></summary><div class="sd-summary-content sd-card-body docutils">
<p class="sd-card-text">Depending on your use-case, you can choose one of these three methods to
persist and load your scikit-learn model, and they all follow the same API:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># Here you can replace pickle with joblib or cloudpickle</span>
<span class="kn">from</span> <span class="nn">pickle</span> <span class="kn">import</span> <span class="n">dump</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="s2">"filename.pkl"</span><span class="p">,</span> <span class="s2">"wb"</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">dump</span><span class="p">(</span><span class="n">clf</span><span class="p">,</span> <span class="n">f</span><span class="p">,</span> <span class="n">protocol</span><span class="o">=</span><span class="mi">5</span><span class="p">)</span>
</pre></div>
</div>
<p class="sd-card-text">Using <code class="docutils literal notranslate"><span class="pre">protocol=5</span></code> is recommended to reduce memory usage and make it faster to
store and load any large NumPy array stored as a fitted attribute in the model.
You can alternatively pass <code class="docutils literal notranslate"><span class="pre">protocol=pickle.HIGHEST_PROTOCOL</span></code> which is
equivalent to <code class="docutils literal notranslate"><span class="pre">protocol=5</span></code> in Python 3.8 and later (at the time of writing).</p>
<p class="sd-card-text">And later when needed, you can load the same object from the persisted file:</p>
<div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># Here you can replace pickle with joblib or cloudpickle</span>
<span class="kn">from</span> <span class="nn">pickle</span> <span class="kn">import</span> <span class="n">load</span>
<span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="s2">"filename.pkl"</span><span class="p">,</span> <span class="s2">"rb"</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span>
<span class="n">clf</span> <span class="o">=</span> <span class="n">load</span><span class="p">(</span><span class="n">f</span><span class="p">)</span>
</pre></div>
</div>
</div>
</details></section>
<section id="security-maintainability-limitations">
<span id="persistence-limitations"></span><h2><span class="section-number">9.5. </span>Security & Maintainability Limitations<a class="headerlink" href="#security-maintainability-limitations" title="Link to this heading">#</a></h2>
<p><a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a> (and <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a> and <code class="xref py py-mod docutils literal notranslate"><span class="pre">clouldpickle</span></code> by extension), has
many documented security vulnerabilities by design and should only be used if
the artifact, i.e. the pickle-file, is coming from a trusted and verified
source. You should never load a pickle file from an untrusted source, similarly
to how you should never execute code from an untrusted source.</p>
<p>Also note that arbitrary computations can be represented using the <code class="docutils literal notranslate"><span class="pre">ONNX</span></code>
format, and it is therefore recommended to serve models using <code class="docutils literal notranslate"><span class="pre">ONNX</span></code> in a
sandboxed environment to safeguard against computational and memory exploits.</p>
<p>Also note that there are no supported ways to load a model trained with a
different version of scikit-learn. While using <a class="reference external" href="https://skops.readthedocs.io/en/stable/modules/classes.html#module-skops.io" title="(in skops)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">skops.io</span></code></a>, <a class="reference external" href="https://joblib.readthedocs.io/en/latest/index.html#module-joblib" title="(in joblib v1.5.dev0)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">joblib</span></code></a>,
<a class="reference external" href="https://docs.python.org/3/library/pickle.html#module-pickle" title="(in Python v3.13)"><code class="xref py py-mod docutils literal notranslate"><span class="pre">pickle</span></code></a>, or <a class="reference external" href="https://github.com/cloudpipe/cloudpickle">cloudpickle</a>, models saved using one version of
scikit-learn might load in other versions, however, this is entirely
unsupported and inadvisable. It should also be kept in mind that operations
performed on such data could give different and unexpected results, or even
crash your Python process.</p>
<p>In order to rebuild a similar model with future versions of scikit-learn,
additional metadata should be saved along the pickled model:</p>
<ul class="simple">
<li><p>The training data, e.g. a reference to an immutable snapshot</p></li>
<li><p>The Python source code used to generate the model</p></li>
<li><p>The versions of scikit-learn and its dependencies</p></li>
<li><p>The cross validation score obtained on the training data</p></li>
</ul>
<p>This should make it possible to check that the cross-validation score is in the
same range as before.</p>
<p>Aside for a few exceptions, persisted models should be portable across
operating systems and hardware architectures assuming the same versions of
dependencies and Python are used. If you encounter an estimator that is not
portable, please open an issue on GitHub. Persisted models are often deployed
in production using containers like Docker, in order to freeze the environment
and dependencies.</p>
<p>If you want to know more about these issues, please refer to these talks:</p>
<ul class="simple">
<li><p><a class="reference external" href="https://www.youtube.com/watch?v=9w_H5OSTO9A">Adrin Jalali: Let’s exploit pickle, and skops to the rescue! | PyData
Amsterdam 2023</a>.</p></li>