mirror of
https://github.com/open-thought/reasoning-gym.git
synced 2026-04-19 12:58:07 +00:00
6eb252ae32
New dataset categories: combinatorics, statistics, optimization, and formal languages. Extended existing algebra, arithmetic, probability, logic, and graphs packages with complex_advanced, linear_algebra, limits, number_theory, conditional_probability, set_operations, and job_scheduling. Each dataset includes config validation, deterministic seeding, custom scoring, curriculum support, and comprehensive unit tests (92 new tests).
70 lines
2 KiB
Python
70 lines
2 KiB
Python
import pytest
|
|
|
|
from reasoning_gym.optimization.knapsack import (
|
|
KnapsackConfig,
|
|
KnapsackCurriculum,
|
|
KnapsackDataset,
|
|
_solve_knapsack,
|
|
)
|
|
|
|
|
|
def test_config_validation():
|
|
with pytest.raises(AssertionError):
|
|
config = KnapsackConfig(min_items=1)
|
|
config.validate()
|
|
|
|
with pytest.raises(AssertionError):
|
|
config = KnapsackConfig(min_items=10, max_items=5)
|
|
config.validate()
|
|
|
|
|
|
def test_deterministic():
|
|
config = KnapsackConfig(seed=42, size=10)
|
|
ds1 = KnapsackDataset(config)
|
|
ds2 = KnapsackDataset(config)
|
|
for i in range(len(ds1)):
|
|
assert ds1[i] == ds2[i]
|
|
|
|
|
|
def test_item_structure():
|
|
config = KnapsackConfig(seed=42, size=50)
|
|
ds = KnapsackDataset(config)
|
|
for i in range(len(ds)):
|
|
item = ds[i]
|
|
assert isinstance(item, dict)
|
|
assert "question" in item
|
|
assert "answer" in item
|
|
assert "metadata" in item
|
|
assert item["metadata"]["source_dataset"] == "knapsack"
|
|
|
|
|
|
def test_answer_correctness():
|
|
config = KnapsackConfig(seed=42, size=50)
|
|
ds = KnapsackDataset(config)
|
|
for i in range(len(ds)):
|
|
item = ds[i]
|
|
weights = item["metadata"]["weights"]
|
|
values = item["metadata"]["values"]
|
|
capacity = item["metadata"]["capacity"]
|
|
expected = _solve_knapsack(weights, values, capacity)
|
|
assert int(item["answer"]) == expected, f"Item {i}: answer mismatch"
|
|
|
|
|
|
def test_score_answer():
|
|
config = KnapsackConfig(seed=42, size=10)
|
|
ds = KnapsackDataset(config)
|
|
for i in range(len(ds)):
|
|
item = ds[i]
|
|
score = ds.score_answer(item["answer"], item)
|
|
assert score == 1.0
|
|
|
|
|
|
def test_curriculum():
|
|
curriculum = KnapsackCurriculum()
|
|
base_value = {"size": 50, "seed": 1}
|
|
base_cfg = curriculum.generate_configuration(base_value)
|
|
assert base_cfg.seed == 1
|
|
|
|
curriculum.increment_attr_level("item_count")
|
|
increased_cfg = curriculum.generate_configuration(base_value)
|
|
assert increased_cfg.max_items >= base_cfg.max_items
|