Merge branch 'main' into rich/graphcolor

tests/test_chain_sum.py

@@ -1,6 +1,6 @@
 import pytest
 
-from reasoning_gym.arithmetic import ChainSum, ChainSumConfig
+from reasoning_gym.arithmetic import ChainSumConfig, ChainSumDataset
 from reasoning_gym.arithmetic.chain_sum import ChainSumCurriculum
 
 
@@ -18,8 +18,8 @@ def test_chain_sum_config_validation():
 def test_chain_sum_deterministic():
     """Test that dataset generates same items with same seed"""
     config = ChainSumConfig(seed=42, size=10)
-    dataset1 = ChainSum(config)
-    dataset2 = ChainSum(config)
+    dataset1 = ChainSumDataset(config)
+    dataset2 = ChainSumDataset(config)
 
     for i in range(len(dataset1)):
         assert dataset1[i] == dataset2[i]
@@ -28,7 +28,7 @@ def test_chain_sum_deterministic():
 def test_chain_sum_items():
     """Test basic properties of generated items"""
     config = ChainSumConfig(min_terms=2, max_terms=4, min_digits=1, max_digits=2, size=100, seed=42)
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
 
     for i in range(len(dataset)):
         item = dataset[i]
@@ -57,7 +57,7 @@ def test_chain_sum_number_ranges():
         size=50,
         seed=42,
     )
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
 
     for i in range(len(dataset)):
         item = dataset[i]
@@ -71,7 +71,7 @@ def test_chain_sum_number_ranges():
 
     # Test 1-digit numbers
     config = ChainSumConfig(min_terms=2, max_terms=2, min_digits=1, max_digits=1, size=50, seed=42)
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
     for i in range(len(dataset)):
         item = dataset[i]
         expression = item["metadata"]["expression"]
@@ -88,7 +88,7 @@ def test_chain_sum_negation():
     config = ChainSumConfig(
         min_terms=2, max_terms=2, min_digits=2, max_digits=2, size=100, seed=42, allow_negation=True
     )
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
 
     # Track if we see both positive and negative numbers
     has_positive = False
@@ -112,7 +112,7 @@ def test_chain_sum_negation():
 def test_chain_sum_iteration():
     """Test that iteration respects dataset size"""
     config = ChainSumConfig(min_terms=2, max_terms=2, size=5, seed=42)  # Small size for testing
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
 
     # Test manual iteration
     items = []

tests/test_coaching.py

@@ -5,7 +5,7 @@ from pathlib import Path
 
 import pytest
 
-from reasoning_gym.arithmetic.chain_sum import ChainSum, ChainSumConfig
+from reasoning_gym.arithmetic.chain_sum import ChainSumConfig, ChainSumDataset
 from reasoning_gym.arithmetic.leg_counting import LegCountingConfig
 from reasoning_gym.coaching import Coach, GroupedScores
 from reasoning_gym.composite import CompositeConfig, CompositeDataset, DatasetSpec
@@ -14,7 +14,7 @@ from reasoning_gym.composite import CompositeConfig, CompositeDataset, DatasetSp
 def test_coach_with_chain_sum():
     # Create a small ChainSum dataset
     config = ChainSumConfig(min_terms=2, max_terms=3, min_digits=1, max_digits=2, size=10, seed=42)
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
     coach = Coach(dataset)
 
     # Simulate an agent working on tasks
@@ -208,7 +208,7 @@ def test_coach_score_logging(tmp_path):
 
     # Create dataset and coach with logging
     config = ChainSumConfig(min_terms=2, max_terms=3, min_digits=1, max_digits=2, size=10, seed=42)
-    dataset = ChainSum(config)
+    dataset = ChainSumDataset(config)
     coach = Coach(dataset, score_log=log_file)
 
     # Score a few answers

tests/test_game_of_life.py

@@ -7,7 +7,7 @@ def test_game_of_life():
     """Test basic properties and solution of generated items"""
 
     # Easy
-    config = GameOfLifeConfig(seed=42, size=1, grid_size_x=20, grid_size_y=20, filled_cells=10, simulation_steps=1)
+    config = GameOfLifeConfig(seed=42, size=10, grid_size_x=20, grid_size_y=20, filled_cells=200, simulation_steps=1)
     dataset = GameOfLifeDataset(config)
 
     for item in dataset:

tests/test_polynomial_equations.py

@@ -112,7 +112,7 @@ def test_polynomial_solutions_evaluation():
             evaluated_value = poly_expr.subs(x, solution)
 
             # Ensure the evaluated value is close to zero (numerical stability threshold)
-            assert abs(evaluated_value) < 1e-6, (
+            assert abs(evaluated_value) < 1e-5, (
                 f"Solution {solution} does not satisfy the polynomial {poly_str}. "
                 f"Evaluated value: {evaluated_value}"
             )

tests/test_products.py

@@ -0,0 +1,144 @@
+import pytest
+
+from reasoning_gym.arithmetic import ProductsConfig, ProductsDataset
+from reasoning_gym.arithmetic.products import ProductsCurriculum
+
+
+def test_products_config_validation():
+    """Test that invalid configs raise appropriate errors"""
+    with pytest.raises(AssertionError):
+        config = ProductsConfig(min_terms=0)
+        config.validate()
+
+    with pytest.raises(AssertionError):
+        config = ProductsConfig(min_terms=3, max_terms=2)
+        config.validate()
+
+
+def test_products_deterministic():
+    """Test that dataset generates same items with same seed"""
+    config = ProductsConfig(seed=42, size=10)
+    dataset1 = ProductsDataset(config)
+    dataset2 = ProductsDataset(config)
+
+    for i in range(len(dataset1)):
+        assert dataset1[i] == dataset2[i]
+
+
+def test_products_items():
+    """Test basic properties of generated items"""
+    config = ProductsConfig(min_terms=2, max_terms=4, min_digits=1, max_digits=2, size=100, seed=42)
+    dataset = ProductsDataset(config)
+
+    for i in range(len(dataset)):
+        item = dataset[i]
+        assert isinstance(item, dict)
+        assert "question" in item
+        assert "answer" in item
+        assert "metadata" in item
+
+        # Verify only * is used
+        expression = item["metadata"]["expression"]
+        assert all(op in ["*", " "] or op.isdigit() for op in expression)
+
+        # Verify the answer matches the expression
+        answer = eval(expression)  # Safe here as we control the expression
+        assert str(answer) == item["answer"]
+
+
+def test_products_number_ranges():
+    """Test that generated numbers respect digit constraints"""
+    # Test 3-digit numbers
+    config = ProductsConfig(
+        min_terms=2,
+        max_terms=2,  # Fix to 2 terms for easier testing
+        min_digits=3,  # Should generate numbers >= 100
+        max_digits=3,  # Should generate numbers <= 999
+        size=50,
+        seed=42,
+    )
+    dataset = ProductsDataset(config)
+
+    for i in range(len(dataset)):
+        item = dataset[i]
+        expression = item["metadata"]["expression"]
+        numbers = [int(n) for n in expression.split() if n.isdigit()]
+        for num in numbers:
+            assert 100 <= num <= 999, f"Number {num} outside valid range for 3 digits"
+
+    # Test 1-digit numbers
+    config = ProductsConfig(min_terms=2, max_terms=2, min_digits=1, max_digits=1, size=50, seed=42)
+    dataset = ProductsDataset(config)
+    for i in range(len(dataset)):
+        item = dataset[i]
+        expression = item["metadata"]["expression"]
+        numbers = [int(n) for n in expression.split() if n.isdigit()]
+        for num in numbers:
+            assert 0 <= num <= 9, f"Number {num} outside valid range for 1 digit"
+
+
+def test_products_iteration():
+    """Test that iteration respects dataset size"""
+    config = ProductsConfig(min_terms=2, max_terms=2, size=5, seed=42)  # Small size for testing
+    dataset = ProductsDataset(config)
+
+    # Test manual iteration
+    items = []
+    for item in dataset:
+        items.append(item)
+    assert len(items) == config.size, "Iterator should yield exactly size items"
+
+    # Test list conversion
+    items = list(dataset)
+    assert len(items) == config.size, "Iterator should yield exactly size items"
+
+    # Test multiple iterations
+    first_items = list(dataset)
+    second_items = list(dataset)
+    assert first_items == second_items, "Multiple iterations should yield same items"
+
+
+def test_products_scoring():
+    """Test that scoring works correctly"""
+    config = ProductsConfig(min_terms=2, max_terms=2, size=10, seed=42)
+    dataset = ProductsDataset(config)
+
+    # Test scoring with exact match
+    item = dataset[0]
+    assert dataset.score_answer(item["answer"], item) == 1.0, "Exact match should score 1.0"
+
+    # Test scoring with wrong answer
+    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"
+
+    # Test scoring with None
+    assert dataset.score_answer(None, item) == 0.0, "None should score 0.0"
+
+
+def test_products_curriculum():
+    curriculum = ProductsCurriculum()
+
+    base_value = {"size": 150, "seed": 1}
+
+    base_cfg: ProductsConfig = curriculum.generate_configuration(base_value)
+    assert base_cfg.seed == 1
+    assert base_cfg.size == 150
+    assert base_cfg.min_digits == 1 and base_cfg.max_digits == 1
+    assert base_cfg.min_terms == 2 and base_cfg.max_terms == 2
+
+    # test incrementing attribute levels for num_terms & num_digits attributes
+    curriculum.increment_attr_level("num_terms")
+    curriculum.increment_attr_level("num_digits")
+
+    increased_cfg = curriculum.generate_configuration(base_value)
+    assert increased_cfg.min_digits == 1 and increased_cfg.max_digits == 2
+    assert increased_cfg.min_terms == 2 and increased_cfg.max_terms == 3
+
+    # test decrementing attribute level for num_digits again
+    curriculum.decrement_attr_level("num_digits")
+
+    partially_decreased_cfg = curriculum.generate_configuration(base_value)
+    assert partially_decreased_cfg.min_digits == 1 and partially_decreased_cfg.max_digits == 1
+    assert partially_decreased_cfg.min_terms == 2 and partially_decreased_cfg.max_terms == 3