feat: update example code and template processing in agent and common graders (#49)

* feat: update example code and template processing in agent and common graders

* feat: fix grader template long text format issue

Author: jc200808

openjudge/graders/agent/action/action_alignment.py

@@ -165,7 +165,7 @@ class ActionAlignmentGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     plan="I will open drawer 1 to find the key.",
         ...     action="open drawer 1"
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/action/action_loop.py

@@ -21,10 +21,11 @@ class ActionLoopDetectionGrader(BaseGrader):
     all pairs of actions for similarity and penalizing based on the proportion
     of similar action pairs found.
     Example:
+        >>> import asyncio
         >>> grader = ActionLoopDetectionGrader(similarity_threshold=1.0)
-        >>> result = await grader.aevaluate(
+        >>> result = asyncio.run(grader.aevaluate(
         ...     messages=[...],
-        ... )
+        ... ))
         >>> print(f"Loop detection score: {result.score}")
     """
 

openjudge/graders/agent/memory/memory_accuracy.py

@@ -165,7 +165,7 @@ class MemoryAccuracyGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     observation="You see a closed cabinet.",
         ...     memory="The cabinet is closed."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/memory/memory_detail_preservation.py

@@ -165,7 +165,7 @@ class MemoryDetailPreservationGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     observation="Cabinet 1 at coordinates (3.5, 2.1) contains 5 red apples.",
         ...     memory="Cabinet 1 at (3.5, 2.1) has 5 red apples."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/memory/memory_retrieval_effectiveness.py

@@ -167,7 +167,7 @@ class MemoryRetrievalEffectivenessGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     observation="You see a closed cabinet.",
         ...     memory="The cabinet is closed."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/observation/observation_information_gain.py

@@ -23,10 +23,11 @@ class ObservationInformationGainGrader(BaseGrader):
     Attributes:
         similarity_threshold: Threshold for considering observations as redundant
     Example:
+        >>> import asyncio
         >>> grader = ObservationInformationGainGrader(similarity_threshold=0.5)
-        >>> result = await grader.aevaluate(
+        >>> result = asyncio.run( grader.aevaluate(
         ...     messages=[...],  # List of message dicts
-        ... )
+        ... ))
         >>> print(f"Info gain score: {result.score}")
     """
 

openjudge/graders/agent/plan/plan_feasibility.py

@@ -168,7 +168,7 @@ class PlanFeasibilityGrader(LLMGrader):
         ...     plan="I will first open the drawer to get the key, then use it to unlock the door.",
         ...     observation="The drawer is closed. You don't have any items.",
         ...     memory="The key is inside the drawer."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/reflection/reflection_accuracy.py

@@ -165,7 +165,7 @@ class ReflectionAccuracyGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     observation="You see a closed cabinet.",
         ...     reflection="I observed a closed cabinet."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/reflection/reflection_outcome_understanding.py

@@ -289,7 +289,7 @@ class ReflectionOutcomeUnderstandingGrader(LLMGrader):
         >>> result = asyncio.run(grader.aevaluate(
         ...     observation="The drawer is now open.",
         ...     reflection="I successfully opened the drawer."
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """
 

openjudge/graders/agent/reflection/reflection_progress_awareness.py

@@ -206,7 +206,7 @@ class ReflectionProgressAwarenessGrader(LLMGrader):
         ...     observation="Cabinet 1 now has apples. Task complete.",
         ...     reflection="Good progress! I've successfully found the apples.",
         ...     context="Task: Find apples in cabinets"
-        ... )
+        ... ))
         >>> print(f"Score: {result.score}")  # Expected: 1.0
     """