Update env/graders.py

env/graders.py

@@ -122,11 +122,11 @@ def _score_error_location(submitted_location: str, expected_location: str) -> fl
 def grade_easy(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     """
     Easy task grader — syntax errors.
-    Max score: 0.999 (strictly less than 1.0)
+    Max score: 1.0
     DETERMINISTIC: same input always returns same score.
     """
     if action is None or action.payload is None:
-        return 0.001, {"error": "null_action"}, "No action provided."
+        return 0.0, {"error": "null_action"}, "No action provided."
 
     payload   = action.payload
     score     = 0.0
@@ -186,8 +186,7 @@ def grade_easy(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     score     += conf_score
     breakdown["confidence"] = round(conf_score, 4)
 
-    # Clamp strictly between 0 and 1 exclusive
-    final_score = round(max(0.001, min(0.999, score)), 4)
+    final_score = round(max(0.0, min(1.0, score)), 4)
     feedback    = " ".join(feedback_parts) if feedback_parts else "No valid response provided."
     return final_score, breakdown, feedback
 
@@ -195,11 +194,11 @@ def grade_easy(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
 def grade_medium(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     """
     Medium task grader — logic errors.
-    Max score: 0.999 (strictly less than 1.0)
+    Max score: 1.0
     DETERMINISTIC: same input always returns same score.
     """
     if action is None or action.payload is None:
-        return 0.001, {"error": "null_action"}, "No action provided."
+        return 0.0, {"error": "null_action"}, "No action provided."
 
     payload        = action.payload
     score          = 0.0
@@ -276,8 +275,7 @@ def grade_medium(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     else:
         breakdown["impact_analysis"] = 0.0
 
-    # Clamp strictly between 0 and 1 exclusive
-    final_score = round(max(0.001, min(0.999, score)), 4)
+    final_score = round(max(0.0, min(1.0, score)), 4)
     feedback    = " ".join(feedback_parts) if feedback_parts else "No valid response provided."
     return final_score, breakdown, feedback
 
@@ -285,12 +283,12 @@ def grade_medium(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
 def grade_hard(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     """
     Hard task grader — performance issues.
-    Max score: 0.999 (strictly less than 1.0)
+    Max score: 1.0
     Frontier models expected ~0.10-0.20.
     DETERMINISTIC: same input always returns same score.
     """
     if action is None or action.payload is None:
-        return 0.001, {"error": "null_action"}, "No action provided."
+        return 0.0, {"error": "null_action"}, "No action provided."
 
     payload        = action.payload
     score          = 0.0
@@ -383,8 +381,7 @@ def grade_hard(action: Action, ground_truth: dict) -> tuple[float, dict, str]:
     score     += conf_score
     breakdown["confidence"] = round(conf_score, 4)
 
-    # Clamp strictly between 0 and 1 exclusive
-    final_score = round(max(0.001, min(0.999, score)), 4)
+    final_score = round(max(0.0, min(1.0, score)), 4)
     feedback    = " ".join(feedback_parts) if feedback_parts else "Performance issue not identified."
     return final_score, breakdown, feedback
 
@@ -398,14 +395,14 @@ def grade(action: Action, task_id: str) -> tuple[float, dict, str]:
     Main grader entry point.
     Looks up ground truth, dispatches to correct grader by difficulty.
     ALWAYS returns (float, dict, str) — never crashes.
-    Score is always strictly between 0.001 and 0.999.
+    Score is always between 0.0 and 1.0.
     """
     if action is None:
-        return 0.001, {"error": "null_action"}, "No action provided."
+        return 0.0, {"error": "null_action"}, "No action provided."
 
     ground_truth = task_manager.get_ground_truth(task_id)
     if ground_truth is None:
-        return 0.001, {"error": "unknown_task"}, f"Task '{task_id}' not found."
+        return 0.0, {"error": "unknown_task"}, f"Task '{task_id}' not found."
 
     difficulty = ground_truth.get("id", "").split("_")[0]
 
@@ -417,6 +414,6 @@ def grade(action: Action, task_id: str) -> tuple[float, dict, str]:
         elif difficulty == "hard":
             return grade_hard(action, ground_truth)
         else:
-            return 0.001, {"error": "unknown_difficulty"}, f"Unknown difficulty: {difficulty}"
+            return 0.0, {"error": "unknown_difficulty"}, f"Unknown difficulty: {difficulty}"
     except Exception as e:
-        return 0.001, {"error": str(e)}, f"Grader error: {str(e)}"
+        return 0.0, {"error": str(e)}, f"Grader error: {str(e)}"