text-classification/sntxtclassify.py at master · silknow/text-classification · GitHub

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
#!/usr/bin/env python
# -*- coding: utf-8 -*-
# License: MIT
# author: Luis Rei < me@luisrei.com >

import argparse

from dnnhelper import (
    train_model,
    create_embeddings,
    eval_model,
    classify_csv,
    hidden_for_csv,
)


def parse_args():
    parser = argparse.ArgumentParser(
        description="SilkNOW Text Classifier",
        usage="""sntxtclassify <command> [<args>]

The available commands are:
   train        Train a text classification model
   embeddings   Create embeddings file
   evaluate     Evaluate a text classification model
   classify     Classify text samples
""",
    )
    subparsers = parser.add_subparsers(required=True, dest="command")

    #
    # TRAIN
    #
    train_parser = subparsers.add_parser("train")
    train_parser.add_argument(
        "--data-train", type=str, help="train CSV file", required=True
    )

    train_parser.add_argument(
        "--target", type=str, help="CSV column target", required=True
    )

    train_parser.add_argument(
        "--pretrained-embeddings",
        type=str,
        help="pretrained embeddings path",
        default="embeddings",
    )

    train_parser.add_argument(
        "--all-embeddings", action="store_true", help="Keep all word vectors"
    )

    train_parser.add_argument(
        "--multimodal", action="store_true", help="Multimodal model."
    )

    train_parser.add_argument(
        "--model-save", type=str, help="save model path", required=True
    )
    #
    # Embeddings
    #
    emb_parser = subparsers.add_parser("embeddings")
    emb_parser.add_argument(
        "--data-train", type=str, help="train CSV file", required=True
    )
    emb_parser.add_argument(
        "--pretrained-embeddings",
        type=str,
        help="pretrained embeddings path",
        required=True,
    )
    emb_parser.add_argument(
        "--save", type=str, help="save embeddings path", required=True
    )

    #
    # Evaluate
    #
    eval_parser = subparsers.add_parser("evaluate")
    eval_parser.add_argument(
        "--model-load", type=str, help="model path", required=True
    )
    eval_parser.add_argument(
        "--data-test", type=str, help="test CSV path", required=True
    )
    eval_parser.add_argument(
        "--target", type=str, help="CSV column target", required=True
    )

    #
    # Classify
    #
    classify_parser = subparsers.add_parser("classify")
    classify_parser.add_argument("--model-load", type=str, help="model path")
    classify_parser.add_argument(
        "--data-input", type=str, help="input CSV path", required=True
    )
    classify_parser.add_argument(
        "--data-output", type=str, help="output file path", required=True
    )
    classify_parser.add_argument(
        "--output-hidden", action="store_true", help="output hidden vectors"
    )
    classify_parser.add_argument(
        "--scores", action="store_true", help="output prediction scores"
    )

    args = parser.parse_args()
    return args


def main():
    args = parse_args()

    if args.command == "train":
        train_model(
            args.data_train,
            args.pretrained_embeddings,
            args.target,
            args.model_save,
            args.all_embeddings,
            args.multimodal,
        )
    elif args.command == "embeddings":
        create_embeddings(
            args.data_train, args.pretrained_embeddings, args.save
        )
    elif args.command == "evaluate":
        eval_model(args.model_load, args.data_test, args.target)
    elif args.command == "classify":
        if args.output_hidden:
            hidden_for_csv(args.model_load, args.data_input, args.data_output)
        else:
            classify_csv(
                args.model_load, args.data_input, args.data_output, args.scores
            )


if __name__ == "__main__":
    main()