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
|
{ buildPythonPackage
, stdenv
, fetchFromGitHub
, boto3
, cookiecutter
, filelock
, regex
, requests
, numpy
, parameterized
, protobuf
, sacremoses
, timeout-decorator
, tokenizers
, tqdm
, pytestCheckHook
}:
buildPythonPackage rec {
pname = "transformers";
version = "4.0.0";
src = fetchFromGitHub {
owner = "huggingface";
repo = pname;
rev = "v${version}";
sha256 = "17djq32pq8d6vqip7i9pda0ldigmzckbbcd278llmpxdriqd4llg";
};
propagatedBuildInputs = [
cookiecutter
filelock
numpy
protobuf
regex
requests
sacremoses
tokenizers
tqdm
];
checkInputs = [
parameterized
pytestCheckHook
timeout-decorator
];
postPatch = ''
substituteInPlace setup.py \
--replace "tokenizers == 0.9.4" "tokenizers"
'';
preCheck = ''
export HOME="$TMPDIR"
# This test requires the `datasets` module to download test
# data. However, since we cannot download in the Nix sandbox
# and `dataset` is an optional dependency for transformers
# itself, we will just remove the tests files that import
# `dataset`.
rm tests/test_retrieval_rag.py
rm tests/test_trainer.py
'';
# We have to run from the main directory for the tests. However,
# letting pytest discover tests leads to errors.
pytestFlagsArray = [ "tests" ];
# Disable tests that require network access.
disabledTests = [
"BlenderbotSmallTokenizerTest"
"Blenderbot3BTokenizerTests"
"GetFromCacheTests"
"TokenizationTest"
"TestTokenizationBart"
"test_all_tokenizers"
"test_batch_encoding_is_fast"
"test_batch_encoding_pickle"
"test_batch_encoding_word_to_tokens"
"test_config_from_model_shortcut"
"test_config_model_type_from_model_identifier"
"test_from_pretrained_use_fast_toggle"
"test_hf_api"
"test_outputs_can_be_shorter"
"test_outputs_not_longer_than_maxlen"
"test_padding_accepts_tensors"
"test_pretokenized_tokenizers"
"test_tokenizer_equivalence_en_de"
"test_tokenizer_from_model_type"
"test_tokenizer_from_model_type"
"test_tokenizer_from_pretrained"
"test_tokenizer_from_tokenizer_class"
"test_tokenizer_identifier_with_correct_config"
"test_tokenizer_identifier_non_existent"
];
meta = with stdenv.lib; {
homepage = "https://github.com/huggingface/transformers";
description = "State-of-the-art Natural Language Processing for TensorFlow 2.0 and PyTorch";
changelog = "https://github.com/huggingface/transformers/releases/tag/v${version}";
license = licenses.asl20;
platforms = platforms.unix;
maintainers = with maintainers; [ danieldk pashashocky ];
};
}
|
