mii/tokenizers
1
0
Fork 0
mirror of https://github.com/mii443/tokenizers.git synced 2025-08-22 16:25:30 +00:00
Code Issues Packages Projects Releases Wiki Activity

make tests happy

Browse Source
This commit is contained in:
Arthur Zucker
2023-09-05 15:37:09 +00:00
parent 531b06f6db
commit 08af8ea9c3
7 changed files with 15 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
bindings/python/py_src/tokenizers/__init__.py
View File

@ -89,6 +89,7 @@ from .tokenizers import (
pre_tokenizers,
processors,
trainers,
__version__,
)
from .implementations import (
BertWordPieceTokenizer,

2
bindings/python/src/tokenizer.rs
View File

@ -75,7 +75,7 @@ impl PyAddedToken {
single_word: None,
lstrip: None,
rstrip: None,
normalized: None,
normalized: Some(!special.unwrap_or(true)),
}
}

2
bindings/python/src/trainers.rs
View File

@ -226,7 +226,7 @@ impl PyBpeTrainer {
if let Ok(content) = token.extract::<String>() {
Ok(tk::tokenizer::AddedToken::from(content, true))
} else if let Ok(mut token) = token.extract::<PyRefMut<PyAddedToken>>() {
token.special = false;
token.special = true;
Ok(token.get_token())
} else {
Err(exceptions::PyTypeError::new_err(

2
bindings/python/tests/bindings/test_tokenizer.py
View File

@ -24,7 +24,7 @@ class TestAddedToken:
assert added_token.special == False
added_token.special = True
assert added_token.special == True
added_token.special = False
assert str(added_token) == "<mask>"
assert (
repr(added_token)

16
bindings/python/tests/bindings/test_trainers.py
View File

@ -34,8 +34,8 @@ class TestBpeTrainer:
assert trainer.min_frequency == 12
assert trainer.show_progress == False
assert trainer.special_tokens == [
AddedToken("1"),
AddedToken("2"),
AddedToken("1", special = True),
AddedToken("2", special = True),
]
assert trainer.limit_alphabet == 13
assert sorted(trainer.initial_alphabet) == ["a", "b", "c"]
@ -91,8 +91,8 @@ class TestWordPieceTrainer:
assert trainer.min_frequency == 12
assert trainer.show_progress == False
assert trainer.special_tokens == [
AddedToken("1"),
AddedToken("2"),
AddedToken("1", special = True),
AddedToken("2", special = True),
]
assert trainer.limit_alphabet == 13
assert sorted(trainer.initial_alphabet) == ["a", "b", "c"]
@ -131,8 +131,8 @@ class TestWordLevelTrainer:
assert trainer.min_frequency == 12
assert trainer.show_progress == False
assert trainer.special_tokens == [
AddedToken("1"),
AddedToken("2"),
AddedToken("1", special = True),
AddedToken("2", special = True),
]
# Modify these
@ -272,8 +272,8 @@ class TestUnigram:
assert trainer.vocab_size == 12345
assert trainer.show_progress == False
assert trainer.special_tokens == [
AddedToken("1", normalized=False),
AddedToken("2", lstrip=True, normalized=False),
AddedToken("1", normalized=False, special = True),
AddedToken("2", lstrip=True, normalized=False, special = True),
]
assert sorted(trainer.initial_alphabet) == ["a", "b", "c"]

4
tokenizers/src/tokenizer/added_vocabulary.rs
View File

@ -609,7 +609,7 @@ mod tests {
);
assert_eq!(vocab.len(), 3);
assert_eq!(vocab.get_vocab_r()[&0], added_token);
assert_eq!(vocab.get_added_tokens_decoder()[&0], added_token);
}
#[test]
@ -650,7 +650,7 @@ mod tests {
assert_eq!(vocab.len(), 3); // New token was added
assert!(vocab.is_special_token("test"));
assert_eq!(
*vocab.get_vocab_r(),
*vocab.get_added_tokens_decoder(),
HashMap::from([
(0, AddedToken::from("test", true)),
(2, AddedToken::from("added_token_1", true)),

2
tokenizers/src/tokenizer/mod.rs
View File

@ -669,7 +669,7 @@ where
// TODO ArthurZ THIS IS WRONG! We need to measure the length of the `set` because
// now some tokens can be both in the added_tokens_encoder and in the vocab
if with_added_tokens {
self.get_vocab(with_added_tokens).len()
self.get_vocab(true).len()
} else {
self.model.get_vocab_size()
}