Minor question template & score_answer improvements (#261)

* math prompt improvements
* ignore brackets in complex_arithmetic results
* improve additional instruction in prompt of polynomial_equations
* more strict tests for score_answer in polynomial_equations
* simplify special reward handling
* fix test_intermediate_integration
* fix sokoban dataset
* add common dataset score_answer consistency test
This commit is contained in:
Andreas Köpf 2025-03-04 21:55:09 +01:00 committed by GitHub
parent bf24999bb0
commit b2904ccab9
106 changed files with 403 additions and 507 deletions

View file

@ -380,20 +380,20 @@ def test_word_ladder_score_answer():
assert dataset.score_answer("COLD", entry) == 0.0
# Test wrong start word
assert dataset.score_answer("BOLD,CORD,CARD,WARD,WARM", entry) == 0.01
assert dataset.score_answer("BOLD,CORD,CARD,WARD,WARM", entry) == 0.0
# Test wrong end word
assert dataset.score_answer("COLD,CORD,CARD,WARD,WARP", entry) == 0.01
assert dataset.score_answer("COLD,CORD,CARD,WARD,WARP", entry) == 0.0
# Test wrong word length
assert dataset.score_answer("COLD,CORDS,CARDS,WARD,WARM", entry) == 0.01
assert dataset.score_answer("COLD,CORDS,CARDS,WARD,WARM", entry) == 0.0
# Test invalid transitions (more than one letter change)
assert dataset.score_answer("COLD,WARD,WARM", entry) == 0.01
assert dataset.score_answer("COLD,WARD,WARM", entry) == 0.0
# Test case insensitivity
assert dataset.score_answer("cold,cord,card,ward,warm", entry) == 1.0
# Test with unknown words (should return partial credit)
assert dataset.score_answer("COLD,COXD,CARD,WARD,WARM", entry) < 1.0
assert dataset.score_answer("COLD,COXD,CARD,WARD,WARM", entry) > 0.0
assert dataset.score_answer("COLD,COXD,CORD,CARD,WARD,WARM", entry) < 1.0
assert dataset.score_answer("COLD,COXD,CORD,CARD,WARD,WARM", entry) > 0.0