open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-23 16:55:05 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

[wip] more flexible api

Browse source
This commit is contained in:
Rich Jones 2025-02-19 03:29:40 +01:00
parent bd56a7b1fe
commit c3bc7937dc
2 changed files with 51 additions and 37 deletions
Download patch file Download diff file Expand all files Collapse all files

42
tests/test_decimal_arithmetic.py
View file

@ -7,7 +7,9 @@ def test_decimal_arithmetic():
"""Test basic properties and solution of generated items"""
# Easy
config = DecimalArithmeticDatasetConfig(seed=42, size=2000, num_decimal_places=3)
config = DecimalArithmeticDatasetConfig(
seed=42, size=999000, min_num_decimal_places=3, max_num_decimal_places=13, terms=13
)
dataset = DecimalArithmeticDataset(config)
for item in dataset:
@ -16,29 +18,31 @@ def test_decimal_arithmetic():
assert "answer" in item
assert "metadata" in item
print(item["answer"])
# Test the scoring
assert dataset.score_answer(answer=item["answer"], entry=item) == 1.0
# M
config = DecimalArithmeticDatasetConfig(seed=42, size=2000, num_decimal_places=8)
dataset = DecimalArithmeticDataset(config)
# # M
# config = DecimalArithmeticDatasetConfig(seed=42, size=2000, num_decimal_places=8)
# dataset = DecimalArithmeticDataset(config)
for item in dataset:
assert isinstance(item, dict)
assert "question" in item
assert "answer" in item
assert "metadata" in item
# for item in dataset:
# assert isinstance(item, dict)
# assert "question" in item
# assert "answer" in item
# assert "metadata" in item
assert dataset.score_answer(answer=item["answer"], entry=item) == 1.0
# assert dataset.score_answer(answer=item["answer"], entry=item) == 1.0
# H
config = DecimalArithmeticDatasetConfig(seed=42, size=2000, num_decimal_places=15)
dataset = DecimalArithmeticDataset(config)
# # H
# config = DecimalArithmeticDatasetConfig(seed=42, size=2000, num_decimal_places=15)
# dataset = DecimalArithmeticDataset(config)
for item in dataset:
assert isinstance(item, dict)
assert "question" in item
assert "answer" in item
assert "metadata" in item
# for item in dataset:
# assert isinstance(item, dict)
# assert "question" in item
# assert "answer" in item
# assert "metadata" in item
assert dataset.score_answer(answer=item["answer"], entry=item) == 1.0
# assert dataset.score_answer(answer=item["answer"], entry=item) == 1.0