naeval

1

2
from os.path import (
3
    expanduser,
4
    join as join_path
5
)
6

7
from IPython.display import HTML
8
from tqdm.notebook import tqdm as log_progress
9

10
from naeval.log import log
11
from naeval.const import (
12
    FACTRU, NE5,
13
    GAREEV, WIKINER, BSNLP,
14

15
    DEEPPAVLOV, DEEPPAVLOV_BERT, DEEPPAVLOV_SLAVIC,
16
    MITIE, PULLENTI,
17
    TEXTERRA, TOMITA,
18
    SLOVNET, SLOVNET_BERT, SPACY,
19
    STANZA,
20

21
    DATASET, JL, GZ,
22

23
    GPU, KB, MB, GB
24
)
25
from naeval.record import (
26
    from_jsons,
27
    as_jsons
28
)
29
from naeval.io import (
30
    format_jl,
31
    parse_jl,
32
    dump_gz_lines,
33
    load_gz_lines
34
)
35
from naeval.readme import patch_readme
36
from naeval.sent import iter_sents
37
from naeval.docker import docker_client
38

39
from naeval.ner.datasets import (
40
    load_factru,
41
    load_bsnlp,
42
    load_ne5,
43
    load_gareev,
44
    load_wikiner
45
)
46
from naeval.ner.models import (
47
    TomitaModel,
48
    PullentiModel,
49
    MitieModel,
50
    TexterraModel,
51
    DeeppavlovModel,
52
    DeeppavlovBERTModel,
53
    DeeppavlovSlavicModel,
54
    SlovnetModel,
55
    SlovnetBERTModel,
56
    SpacyModel,
57
    StanzaModel,
58
)
59
from naeval.ner.markup import (
60
    Markup,
61
    show_markup
62
)
63
from naeval.ner.score import score_markups
64
from naeval.ner.report import (
65
    scores_report_table,
66
    format_scores_report,
67
    format_github_scores_report,
68

69
    Bench,
70
    bench_report_table,
71
    format_bench_report,
72
)
73

74

75
CORUS_DATA_DIR = expanduser('~/proj/corus-data')
76
CORUS_FILES = {
77
    FACTRU: 'factRuEval-2016-master',
78
    BSNLP: 'bsnlp',
79
    NE5: 'Collection5',
80
    GAREEV: 'rus-ner-news-corpus.iob',
81
    WIKINER: 'aij-wikiner-ru-wp3.bz2',
82
}
83
NER1, NER2 = 'ner1', 'ner2'
84
README = expanduser('~/proj/naeval/README.md')
85
SLOVNET_README = expanduser('~/proj/slovnet/README.md')
86

87
DATASETS = {
88
    FACTRU: load_factru,
89
    GAREEV: load_gareev,
90
    NE5: load_ne5,
91
    BSNLP: load_bsnlp,
92
}
93
MODELS = {
94
    SLOVNET: SlovnetModel,
95
    SLOVNET_BERT: SlovnetBERTModel,
96
    DEEPPAVLOV: DeeppavlovModel,
97
    DEEPPAVLOV_BERT: DeeppavlovBERTModel,
98
    DEEPPAVLOV_SLAVIC: DeeppavlovSlavicModel,
99
    PULLENTI: PullentiModel,
100
    SPACY: SpacyModel,
101
    STANZA: StanzaModel,
102
    TEXTERRA: TexterraModel,
103
    TOMITA: TomitaModel,
104
    MITIE: MitieModel,
105
}
106

107
DATA_DIR = expanduser('~/proj/naeval/data/ner')
108