open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-25 17:10:51 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add GO hints, legend, disallow numeric answer, store expected string answer

Browse source
This commit is contained in:
Andreas Koepf 2025-02-07 15:18:01 +01:00
parent 2458d3a646
commit 81cb7aa42b
3 changed files with 150 additions and 32 deletions
Download patch file Download diff file Expand all files Collapse all files

11
tests/test_tsumego.py
View file

@ -1,7 +1,6 @@
"""Tests for Ttsumego problem generation"""
import pytest
from random import Random
from reasoning_gym.games.tsumego import TsumegoConfig, TsumegoDataset
@ -30,7 +29,7 @@ def test_dataset_item_properties():
assert key in item
metadata = item["metadata"]
for key in ["board_size", "board", "solution"]:
for key in ["difficulty", "board", "solution"]:
assert key in metadata
board = metadata["board"]
@ -101,17 +100,11 @@ def test_liberties_and_move():
def test_score_answer():
config = TsumegoConfig(min_board_size=9, max_board_size=9, max_stones=10)
dataset = TsumegoDataset(config)
metadata = {"board_size": 9, "solution": "5,5"}
# Correct numeric answer
assert dataset.score_answer("5,5", metadata) == 1.0
metadata = {"board_size": 9, "solution": (4, 4)}
# Correct letter-number answer (E corresponds to 5)
assert dataset.score_answer("E5", metadata) == 1.0
# Valid but incorrect numeric move
assert dataset.score_answer("4,4", metadata) == 0.05
# Valid but incorrect letter-number move (D corresponds to 4)
assert dataset.score_answer("D4", metadata) == 0.05