allow more separators

Author: hynky1999

pyproject.toml

@@ -109,7 +109,7 @@ multilingual = [
     "jieba", # for chinese tokenizer
     "pyvi", # for vietnamese tokenizer
 ]
-math = ["latex2sympy2_extended==1.0.5"]
+math = ["latex2sympy2_extended==1.0.6"]
 
 [project.urls]
 Homepage = "https://github.com/huggingface/lighteval"

src/lighteval/metrics/utils/extractive_match_utils.py

@@ -27,7 +27,7 @@
 from typing import Any, Literal, Sequence
 
 import sympy
-from sympy import Basic, FiniteSet, MatrixBase, Number
+from sympy import FiniteSet, Number
 from sympy.parsing import parse_expr
 
 from lighteval.metrics.utils.math_comparison import should_treat_as_complex
@@ -234,7 +234,7 @@ def lazy_latex_regex(latex_config: LatexExtractionConfig, language: Language) ->
     and_word = translation_literal.and_word
     or_word = translation_literal.or_word
     next_groups = "".join(
-        [rf"(?:\s*(?:{and_word}|{or_word})\s*{make_latex_env_pattern(f'next{i}_')})?" for i in range(1, 6)]
+        [rf"(?:\s*(?:{and_word}|{or_word}|,)\s*{make_latex_env_pattern(f'next{i}_')})?" for i in range(1, 6)]
     )
 
     latex_envs_re = rf"(?:{first_latex_group}{next_groups})"
@@ -266,7 +266,7 @@ def lazy_latex_regex(latex_config: LatexExtractionConfig, language: Language) ->
         latex_re_boxed = make_latex_env_pattern(prefix="first_", context="boxed")
         next_groups = "".join(
             [
-                rf"(?:\s*(?:{and_word}|{or_word})\s*{make_latex_env_pattern(f'next{i}_', context='boxed')})?"
+                rf"(?:\s*(?:{and_word}|{or_word}|,)\s*{make_latex_env_pattern(f'next{i}_', context='boxed')})?"
                 for i in range(1, 6)
             ]
         )
@@ -487,9 +487,7 @@ def extract_latex(
     return latex_exprs[0], latex_strs[0]
 
 
-def extract_match(
-    match: re.Match, target_type: ExtractionTarget, timeout_seconds: int
-):
+def extract_match(match: re.Match, target_type: ExtractionTarget, timeout_seconds: int):
     """Extracts the match from the regex match.
 
     Args:

tests/metrics/test_extractive_match.py

@@ -1035,7 +1035,7 @@ def test_math_extraction_additional_cases(gold, pred, expected):
         ),
         (
             r"$(2,1),(1,2),(-1,-20),(-20,-1)$",
-            r"solutions are:\n\n\\[\n\\boxed{(1, 2)}, \\boxed{(2, 1)}, \\boxed{(-1, -20)}, \\boxed{(-20, -1)}\n\\]",
+            "solutions are:\n\n\\[\n\\boxed{(1, 2)}, \\boxed{(2, 1)}, \\boxed{(-1, -20)}, \\boxed{(-20, -1)}\n\\]",
             1,
         ),
         (
@@ -1121,6 +1121,16 @@ def test_math_extraction_additional_cases(gold, pred, expected):
             r"$\boxed{10^{\frac{\sqrt{13} - 5}{6}}} \quad \text{and} \quad \boxed{10^{-\frac{5 + \sqrt{13}}{6}}}$",
             1,
         ),
+        (
+            r"\boxed{1} and and and or thus but \boxed{2} and \boxed{3}",
+            r"$\boxed{2,3}$",
+            1,
+        ),
+        (
+            r"\boxed{1} and and and or thus but \boxed{2} and \boxed{3}",
+            r"$\boxed{1,2,3}$",
+            0,
+        ),
     ],
 )
 def test_math_numina_cases(gold, pred, expected):