formatting

reasoning_gym/arc/arc_agi.py

@@ -99,13 +99,13 @@ class ArcAgiDataset(ProceduralDataset):
     def _create_augmentation_fn(self, rng: Random) -> Callable[[Board], Board]:
         """Create a composite augmentation function from enabled options"""
         fns = []
-        
+
         if self.config.use_rotations:
             fns.append(rng.choice(ROTATION_AUGMENTATIONS))
-            
+
         if self.config.use_mirrors:
             fns.append(rng.choice(MIRROR_AUGMENTATIONS))
-            
+
         if self.config.use_color_permutation:
             color_table = list(range(10))
             rng.shuffle(color_table)
@@ -116,7 +116,7 @@ class ArcAgiDataset(ProceduralDataset):
             for fn in fns:
                 result = fn(result)
             return result
-            
+
         return composite_fn
 
     def __getitem__(self, idx: int) -> dict:
@@ -137,16 +137,14 @@ class ArcAgiDataset(ProceduralDataset):
         # Apply augmentation to all train examples
         augmented_train = []
         for p in train:
-            augmented_train.append({
+            augmented_train.append({"input": augment(p["input"]), "output": augment(p["output"])})
-                "input": augment(p["input"]),
-                "output": augment(p["output"])
-            })
 
         examples = [
-            format_board_pair(i + 1, p, formatting_options=self.config.board_format_opts) 
+            format_board_pair(i + 1, p, formatting_options=self.config.board_format_opts)
             for i, p in enumerate(augmented_train)
         ]
         examples = "".join(examples)
+
         # Apply augmentation to test example
         augmented_test_input = augment(test["input"])
         augmented_test_output = augment(test["output"])

tests/test_arc_agi.py

@@ -52,48 +52,28 @@ def test_arc_agi_items():
 def test_arc_agi_augmentations():
     """Test that augmentations can be selectively enabled/disabled"""
     # Test with all augmentations disabled
-    config = ArcAgiConfig(
+    config = ArcAgiConfig(seed=42, size=10, use_rotations=False, use_mirrors=False, use_color_permutation=False)
-        seed=42, 
-        size=10,
-        use_rotations=False,
-        use_mirrors=False, 
-        use_color_permutation=False
-    )
     base_dataset = ArcAgiDataset(config)
     base_items = list(base_dataset)
 
     # Test with rotations only
-    rot_config = ArcAgiConfig(
+    rot_config = ArcAgiConfig(seed=42, size=10, use_rotations=True, use_mirrors=False, use_color_permutation=False)
-        seed=42,
-        size=10,
-        use_rotations=True,
-        use_mirrors=False,
-        use_color_permutation=False
-    )
     rot_dataset = ArcAgiDataset(rot_config)
     rot_items = list(rot_dataset)
 
     # Items should differ when rotations are enabled
     assert any(
-        base_items[i]["metadata"]["input"] != rot_items[i]["metadata"]["input"]
+        base_items[i]["metadata"]["input"] != rot_items[i]["metadata"]["input"] for i in range(len(base_items))
-        for i in range(len(base_items))
     ), "Rotation augmentation had no effect"
 
     # Test with color permutation only
-    color_config = ArcAgiConfig(
+    color_config = ArcAgiConfig(seed=42, size=10, use_rotations=False, use_mirrors=False, use_color_permutation=True)
-        seed=42,
-        size=10,
-        use_rotations=False,
-        use_mirrors=False,
-        use_color_permutation=True
-    )
     color_dataset = ArcAgiDataset(color_config)
     color_items = list(color_dataset)
 
     # Items should differ when color permutation is enabled
     assert any(
-        base_items[i]["metadata"]["input"] != color_items[i]["metadata"]["input"]
+        base_items[i]["metadata"]["input"] != color_items[i]["metadata"]["input"] for i in range(len(base_items))
-        for i in range(len(base_items))
     ), "Color permutation had no effect"
 
 
@@ -113,7 +93,7 @@ def test_arc_agi_scoring():
         assert dataset.score_answer(None, entry=item) == 0.0
 
         # Test wrong but valid grid format
-        wrong_answer = "0 0\n0 0"
+        wrong_answer = "1 0 0 0\n0 0 0 1"
         assert dataset.score_answer(wrong_answer, entry=item) == 0.05