open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-19 12:58:07 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Merge branch 'main' of https://github.com/open-thought/reasoning-gym into env/string-splitting

Browse source
This commit is contained in:
Zafir Stojanovski 2025-02-14 21:55:23 +01:00
commit 3275538fa1
2 changed files with 4 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

2
tests/test_arc_1d.py
View file

@ -98,7 +98,7 @@ def test_arc_1d_scoring():
assert dataset.score_answer(entry["answer"], entry) == 1.0
# Test partial match (answer contained within response)
assert dataset.score_answer(f"The answer is: {entry['answer']}", entry) == 0.5
assert dataset.score_answer(f"The answer is: {entry['answer']}", entry) > 0.5
# Test incorrect answer
assert dataset.score_answer("wrong answer", entry) == 0.01

4
tests/test_products.py
View file

@ -111,7 +111,9 @@ def test_products_scoring():
assert dataset.score_answer("wrong", item) == 0.01, "Wrong answer should score 0.01"
# Test scoring with partial match (answer contained in response)
assert dataset.score_answer(f"The answer is {item['answer']}", item) == 0.5, "Partial match should score 0.5"
assert (
dataset.score_answer(f"The answer is {item['answer']}", item) > 0.1
), "Partial match should scored len(oracle_answer)/len(answer)"
# Test scoring with None
assert dataset.score_answer(None, item) == 0.0, "None should score 0.0"