mirror of
https://github.com/open-thought/reasoning-gym.git
synced 2026-04-19 12:58:07 +00:00
d9638df79c
* updated algorithmic datasets * added changes to symbolic and power * updated power function test
64 lines
2 KiB
Python
64 lines
2 KiB
Python
"""Tests for Power Function questions generation"""
|
|
|
|
import pytest
|
|
|
|
from reasoning_gym.arithmetic import PowerFunctionConfig, PowerFunctionDataset
|
|
|
|
|
|
def test_power_function_dataset_deterministic():
|
|
"""Test that dataset generates same items with same seed"""
|
|
config = PowerFunctionConfig(seed=42, size=10)
|
|
dataset1 = PowerFunctionDataset(config)
|
|
dataset2 = PowerFunctionDataset(config)
|
|
|
|
for i in range(len(dataset1)):
|
|
assert dataset1[i] == dataset2[i]
|
|
|
|
|
|
def test_power_function_dataset_items():
|
|
"""Test basic properties of generated items"""
|
|
config = PowerFunctionConfig(min_base=-100, max_base=-100, min_exponent=-10, max_exponent=10, size=10, seed=42)
|
|
dataset = PowerFunctionDataset(config)
|
|
|
|
for i in range(len(dataset)):
|
|
item = dataset[i]
|
|
# Check item structure
|
|
assert isinstance(item, dict)
|
|
assert "question" in item
|
|
assert "answer" in item
|
|
assert "metadata" in item
|
|
|
|
# Check metadata
|
|
assert "base" in item["metadata"]
|
|
assert "exponent" in item["metadata"]
|
|
|
|
base = item["metadata"]["base"]
|
|
exponent = item["metadata"]["exponent"]
|
|
solution = item["metadata"]["solution"]
|
|
|
|
# Verify values
|
|
assert config.min_base <= base <= config.max_base
|
|
assert config.min_exponent <= exponent <= config.max_exponent
|
|
assert solution == pow(base, exponent)
|
|
|
|
|
|
def test_power_function_dataset_iteration():
|
|
"""Test that iteration respects dataset size"""
|
|
config = PowerFunctionConfig(size=5, seed=42)
|
|
dataset = PowerFunctionDataset(config)
|
|
|
|
items = list(dataset)
|
|
assert len(items) == config.size
|
|
|
|
# Test multiple iterations yield same items
|
|
assert items == list(dataset)
|
|
|
|
|
|
def test_power_function_score_function():
|
|
"""Test score function"""
|
|
config = PowerFunctionConfig(seed=42)
|
|
dataset = PowerFunctionDataset(config)
|
|
|
|
for item in dataset:
|
|
answer = item["answer"]
|
|
assert dataset.score_answer(answer, item) == 1.0
|