aboutsummaryrefslogtreecommitdiff
path: root/infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix
diff options
context:
space:
mode:
Diffstat (limited to 'infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix')
-rw-r--r--infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix22
1 files changed, 14 insertions, 8 deletions
diff --git a/infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix b/infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix
index fdb807bd0a1e..39fd7a11d0df 100644
--- a/infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix
+++ b/infra/libkookie/nixpkgs/pkgs/development/python-modules/transformers/default.nix
@@ -2,13 +2,14 @@
, stdenv
, fetchFromGitHub
, boto3
+, cookiecutter
, filelock
, regex
, requests
, numpy
, parameterized
+, protobuf
, sacremoses
-, sentencepiece
, timeout-decorator
, tokenizers
, tqdm
@@ -17,23 +18,23 @@
buildPythonPackage rec {
pname = "transformers";
- version = "3.3.1";
+ version = "4.0.0";
src = fetchFromGitHub {
owner = "huggingface";
repo = pname;
rev = "v${version}";
- sha256 = "1j9nzhl0zw5z9rnvzfih7v6bax353rxp05b3f0cvkii3b5dbkc2j";
+ sha256 = "17djq32pq8d6vqip7i9pda0ldigmzckbbcd278llmpxdriqd4llg";
};
propagatedBuildInputs = [
- boto3
+ cookiecutter
filelock
numpy
+ protobuf
regex
requests
sacremoses
- sentencepiece
tokenizers
tqdm
];
@@ -46,7 +47,7 @@ buildPythonPackage rec {
postPatch = ''
substituteInPlace setup.py \
- --replace "tokenizers == 0.8.1.rc2" "tokenizers>=0.8"
+ --replace "tokenizers == 0.9.4" "tokenizers"
'';
preCheck = ''
@@ -67,11 +68,15 @@ buildPythonPackage rec {
# Disable tests that require network access.
disabledTests = [
- "PegasusTokenizationTest"
- "T5TokenizationTest"
+ "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"
@@ -86,6 +91,7 @@ buildPythonPackage rec {
"test_tokenizer_from_pretrained"
"test_tokenizer_from_tokenizer_class"
"test_tokenizer_identifier_with_correct_config"
+ "test_tokenizer_identifier_non_existent"
];
meta = with stdenv.lib; {