this repo has no description
anil.recoil.org
1opam-version: "2.0"
2authors: "Francois Berenger"
3maintainer: "unixjunkie@sdf.org"
4homepage: "https://github.com/UnixJunkie/linwrap"
5bug-reports: "https://github.com/UnixJunkie/linwrap/issues"
6dev-repo: "git+https://github.com/UnixJunkie/linwrap.git"
7license: "BSD-3-Clause"
8build: ["dune" "build" "-p" name "-j" jobs]
9install: [
10 ["cp" "bin/ecfp6.py" "%{bin}%/linwrap_ecfp6.py"]
11]
12depends: [
13 "base-unix"
14 "batteries"
15 "conf-liblinear-tools"
16 "conf-python-3"
17 "conf-rdkit"
18 "cpm" {>= "4.0.0"}
19 "dokeysto_camltc"
20 "dolog" {>= "4.0.0" & < "5.0.0"}
21 "dune" {>= "1.10"}
22 "minicli" {>= "5.0.0"}
23 "parany" {>= "6.0.0" & < "10.0.0"}
24]
25synopsis: "Wrapper around liblinear-tools"
26description: """
27Only L2-regularized logistic regression is supported currently.
28When using bagging, each model is trained on balanced bootstraps
29from the training set (one bootstrap for the positive class,
30one for the negative class).
31The size of the bootstrap is the size of the smallest (under-represented)
32class.
33
34usage: linwrap
35 -i <filename>: training set or DB to screen
36 [-o <filename>]: predictions output file
37 [-np <int>]: ncores
38 [-c <float>]: fix C
39 [-w <float>]: fix w1
40 [-k <int>]: number of bags for bagging (default=off)
41 [-n <int>]: folds of cross validation
42 [--seed <int>]: fix random seed
43 [-p <float>]: training set portion (in [0.0:1.0])
44 [--train <train.liblin>]: training set (overrides -p)
45 [--valid <valid.liblin>]: validation set (overrides -p)
46 [--test <test.liblin>]: test set (overrides -p)
47 [{-l|--load} <filename>]: prod. mode; use trained models
48 [{-s|--save} <filename>]: train. mode; save trained models
49 [-f]: force overwriting existing model file
50 [--scan-c]: scan for best C
51 [--scan-w]: scan weight to counter class imbalance
52 [--scan-k]: scan number of bags (advice: optim. k rather than w)
53"""
54url {
55 src: "https://github.com/UnixJunkie/linwrap/archive/v5.0.0.tar.gz"
56 checksum: [
57 "sha256=2e2f573ae5102a575ad70acdeaf84673a0b8ec6eac60dd3601dd79dfddc5c596"
58 "md5=3407f24bc31cfdbafa6141aced22ce81"
59 ]
60}