Python - Test Decoders

This commit is contained in:
Anthony MOI
2020-03-31 18:20:45 -04:00
parent 0de9885da8
commit a9f4c5950a
2 changed files with 62 additions and 1 deletions

View File

@@ -107,7 +107,7 @@ impl BPEDecoder {
#[new]
#[args(kwargs = "**")]
fn new(obj: &PyRawObject, kwargs: Option<&PyDict>) -> PyResult<()> {
let mut suffix = String::from("</w");
let mut suffix = String::from("</w>");
if let Some(kwargs) = kwargs {
for (key, value) in kwargs {

View File

@@ -0,0 +1,61 @@
import pytest
from tokenizers.decoders import Decoder, ByteLevel, WordPiece, Metaspace, BPEDecoder
class TestByteLevel:
def test_instantiate(self):
assert ByteLevel() is not None
assert isinstance(ByteLevel(), Decoder)
def test_decoding(self):
decoder = ByteLevel()
assert decoder.decode(["My", "Ġname", "Ġis", "ĠJohn"]) == "My name is John"
class TestWordPiece:
def test_instantiate(self):
assert WordPiece() is not None
assert WordPiece(prefix="__") is not None
assert WordPiece(cleanup=True) is not None
assert isinstance(WordPiece(), Decoder)
def test_decoding(self):
decoder = WordPiece()
assert decoder.decode(["My", "na", "##me", "is", "Jo", "##hn"]) == "My name is John"
assert decoder.decode(["I", "'m", "Jo", "##hn"]) == "I'm John"
decoder = WordPiece(prefix="__", cleanup=False)
assert decoder.decode(["My", "na", "__me", "is", "Jo", "__hn"]) == "My name is John"
assert decoder.decode(["I", "'m", "Jo", "__hn"]) == "I 'm John"
class TestMetaspace:
def test_instantiate(self):
assert Metaspace() is not None
assert Metaspace(replacement="-") is not None
with pytest.raises(Exception, match="replacement must be a character"):
Metaspace(replacement="")
assert Metaspace(add_prefix_space=True) is not None
assert isinstance(Metaspace(), Decoder)
def test_decoding(self):
decoder = Metaspace()
assert decoder.decode(["▁My", "▁name", "▁is", "▁John"]) == "My name is John"
decoder = Metaspace(replacement="-", add_prefix_space=False)
assert decoder.decode(["-My", "-name", "-is", "-John"]) == " My name is John"
class TestBPEDecoder:
def test_instantiate(self):
assert BPEDecoder() is not None
assert BPEDecoder(suffix="_") is not None
assert isinstance(BPEDecoder(), Decoder)
def test_decoding(self):
decoder = BPEDecoder()
assert (
decoder.decode(["My</w>", "na", "me</w>", "is</w>", "Jo", "hn</w>"])
== "My name is John"
)
decoder = BPEDecoder(suffix="_")
assert decoder.decode(["My_", "na", "me_", "is_", "Jo", "hn_"]) == "My name is John"