Feat/countdown scoring

reasoning_gym/games/countdown.py

@@ -1,9 +1,10 @@
 from dataclasses import dataclass
 from random import Random
-from typing import List, Optional, Tuple
+from typing import Any, Dict, List, Optional, Tuple
 
 import sympy
 from sympy import Symbol, symbols
+from sympy.parsing.sympy_parser import parse_expr
 
 from ..factory import ProceduralDataset, register_dataset
 
@@ -158,6 +159,23 @@ def _generate_expression(self, rng: Random) -> Tuple[str, List[int], int]:
 
         raise ValueError(f"Failed to generate valid expression after {max_attempts} attempts")
 
+    def score_answer(self, answer: Optional[str], metadata: Dict[str, Any]) -> float:
+        """Determine if the solution provided solves the problem"""
+        reward = 0.0
+        if answer is not None:
+            try:
+                user_answer = int(parse_expr(answer))
+                solved = user_answer == metadata["target"]
+                if solved:
+                    reward = 1.0
+                elif len(answer.strip()) > 0:  # encourage partial solutions
+                    reward = 0.05
+                else:
+                    reward = 0.01
+            except:
+                reward = 0.01
+        return reward
+
 
 # Register the dataset
 register_dataset("countdown", CountdownDataset, CountdownConfig)

tests/test_countdown.py

@@ -64,6 +64,16 @@ def test_countdown_game_items():
 
         # Verify expression evaluates correctly
         expr = item["metadata"]["expression"]
+
+        # check score
+        assert dataset.score_answer(answer=expr, metadata=item["metadata"]) == 1.0  # correct answer
+        assert dataset.score_answer(answer="45+2", metadata=item["metadata"]) == 0.05  # wrong answer but an attempt
+        assert (
+            dataset.score_answer(answer="a wrong solution", metadata=item["metadata"]) == 0.01
+        )  # wrong answer but incorrectly formatted
+        assert dataset.score_answer(answer="", metadata=item["metadata"]) == 0.01  # wrong answer but empty string
+        assert dataset.score_answer(answer=None, metadata=item["metadata"]) == 0.0  # no answer
+
         try:
             result = eval(expr)  # Safe here since we control expression generation
             assert result == item["metadata"]["target"]