nixpkgs/pkgs/development/python-modules/transformers/default.nix
Daniël de Kok 2fb0ad3018 python3Packages.transformers: 3.0.2 -> 3.1.0
Transformers now has `nlp` as an optional dependency. `nlp` will
require some work to package, since it attempts to download modules
into its Python package path by default. So, the `nlp`-based test
is disable for the time being.
2020-09-01 10:33:34 -07:00

91 lines
2.1 KiB
Nix

{ buildPythonPackage
, stdenv
, fetchFromGitHub
, boto3
, filelock
, regex
, requests
, numpy
, sacremoses
, sentencepiece
, timeout-decorator
, tokenizers
, tqdm
, pytestCheckHook
}:
buildPythonPackage rec {
pname = "transformers";
version = "3.1.0";
src = fetchFromGitHub {
owner = "huggingface";
repo = pname;
rev = "v${version}";
sha256 = "0wg36qrcljmpsyhjaxpqw3s1r6276yg8cq0bjrf52l4zlc5k4xzk";
};
propagatedBuildInputs = [
boto3
filelock
numpy
regex
requests
sacremoses
sentencepiece
tokenizers
tqdm
];
checkInputs = [
pytestCheckHook
timeout-decorator
];
postPatch = ''
substituteInPlace setup.py \
--replace "tokenizers == 0.8.1.rc2" "tokenizers>=0.8"
'';
preCheck = ''
export HOME="$TMPDIR"
cd tests
# This test requires the nlp module, which we haven't
# packaged yet. However, nlp is optional for transformers
# itself
rm test_trainer.py
'';
# Disable tests that require network access.
disabledTests = [
"PegasusTokenizationTest"
"T5TokenizationTest"
"test_all_tokenizers"
"test_batch_encoding_is_fast"
"test_batch_encoding_pickle"
"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_identifier_with_correct_config"
];
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 ];
};
}