Update Space (evaluate main: e4a27243)
Browse files- exact_match.py +25 -11
- requirements.txt +1 -1
exact_match.py
CHANGED
@@ -14,6 +14,8 @@
|
|
14 |
"""Exact Match metric."""
|
15 |
import re
|
16 |
import string
|
|
|
|
|
17 |
|
18 |
import datasets
|
19 |
import numpy as np
|
@@ -83,13 +85,29 @@ _CITATION = """
|
|
83 |
"""
|
84 |
|
85 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
86 |
@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
|
87 |
-
class
|
88 |
-
|
|
|
|
|
|
|
|
|
89 |
return evaluate.MetricInfo(
|
90 |
description=_DESCRIPTION,
|
91 |
citation=_CITATION,
|
92 |
inputs_description=_KWARGS_DESCRIPTION,
|
|
|
93 |
features=datasets.Features(
|
94 |
{
|
95 |
"predictions": datasets.Value("string", id="sequence"),
|
@@ -103,30 +121,26 @@ class ExactMatch(evaluate.Metric):
|
|
103 |
self,
|
104 |
predictions,
|
105 |
references,
|
106 |
-
regexes_to_ignore=None,
|
107 |
-
ignore_case=False,
|
108 |
-
ignore_punctuation=False,
|
109 |
-
ignore_numbers=False,
|
110 |
):
|
111 |
|
112 |
-
if regexes_to_ignore is not None:
|
113 |
-
for s in regexes_to_ignore:
|
114 |
predictions = np.array([re.sub(s, "", x) for x in predictions])
|
115 |
references = np.array([re.sub(s, "", x) for x in references])
|
116 |
else:
|
117 |
predictions = np.asarray(predictions)
|
118 |
references = np.asarray(references)
|
119 |
|
120 |
-
if ignore_case:
|
121 |
predictions = np.char.lower(predictions)
|
122 |
references = np.char.lower(references)
|
123 |
|
124 |
-
if ignore_punctuation:
|
125 |
repl_table = string.punctuation.maketrans("", "", string.punctuation)
|
126 |
predictions = np.char.translate(predictions, table=repl_table)
|
127 |
references = np.char.translate(references, table=repl_table)
|
128 |
|
129 |
-
if ignore_numbers:
|
130 |
repl_table = string.digits.maketrans("", "", string.digits)
|
131 |
predictions = np.char.translate(predictions, table=repl_table)
|
132 |
references = np.char.translate(references, table=repl_table)
|
|
|
14 |
"""Exact Match metric."""
|
15 |
import re
|
16 |
import string
|
17 |
+
from dataclasses import dataclass
|
18 |
+
from typing import Optional
|
19 |
|
20 |
import datasets
|
21 |
import numpy as np
|
|
|
85 |
"""
|
86 |
|
87 |
|
88 |
+
@dataclass
|
89 |
+
class ExactMatchConfig(evaluate.info.Config):
|
90 |
+
|
91 |
+
name: str = "default"
|
92 |
+
|
93 |
+
regexes_to_ignore: Optional[str] = None
|
94 |
+
ignore_case: bool = False
|
95 |
+
ignore_punctuation: bool = False
|
96 |
+
ignore_numbers: bool = False
|
97 |
+
|
98 |
+
|
99 |
@evaluate.utils.file_utils.add_start_docstrings(_DESCRIPTION, _KWARGS_DESCRIPTION)
|
100 |
+
class ExactMatchConfig(evaluate.Metric):
|
101 |
+
|
102 |
+
CONFIG_CLASS = ExactMatchConfig
|
103 |
+
ALLOWED_CONFIG_NAMES = ["default"]
|
104 |
+
|
105 |
+
def _info(self, config):
|
106 |
return evaluate.MetricInfo(
|
107 |
description=_DESCRIPTION,
|
108 |
citation=_CITATION,
|
109 |
inputs_description=_KWARGS_DESCRIPTION,
|
110 |
+
config=config,
|
111 |
features=datasets.Features(
|
112 |
{
|
113 |
"predictions": datasets.Value("string", id="sequence"),
|
|
|
121 |
self,
|
122 |
predictions,
|
123 |
references,
|
|
|
|
|
|
|
|
|
124 |
):
|
125 |
|
126 |
+
if self.config.regexes_to_ignore is not None:
|
127 |
+
for s in self.config.regexes_to_ignore:
|
128 |
predictions = np.array([re.sub(s, "", x) for x in predictions])
|
129 |
references = np.array([re.sub(s, "", x) for x in references])
|
130 |
else:
|
131 |
predictions = np.asarray(predictions)
|
132 |
references = np.asarray(references)
|
133 |
|
134 |
+
if self.config.ignore_case:
|
135 |
predictions = np.char.lower(predictions)
|
136 |
references = np.char.lower(references)
|
137 |
|
138 |
+
if self.config.ignore_punctuation:
|
139 |
repl_table = string.punctuation.maketrans("", "", string.punctuation)
|
140 |
predictions = np.char.translate(predictions, table=repl_table)
|
141 |
references = np.char.translate(references, table=repl_table)
|
142 |
|
143 |
+
if self.config.ignore_numbers:
|
144 |
repl_table = string.digits.maketrans("", "", string.digits)
|
145 |
predictions = np.char.translate(predictions, table=repl_table)
|
146 |
references = np.char.translate(references, table=repl_table)
|
requirements.txt
CHANGED
@@ -1 +1 @@
|
|
1 |
-
git+https://github.com/huggingface/evaluate@
|
|
|
1 |
+
git+https://github.com/huggingface/evaluate@e4a2724377909fe2aeb4357e3971e5a569673b39
|