Robust Judge API parsing and schemas validation default values for 0.5B model support

backend/models/schemas.py

@@ -167,12 +167,12 @@ class JudgeResponse(BaseModel):
     chunk_a_score: ChunkScore
     chunk_b_score: ChunkScore
 
-    winner_reason: str
-    chunk_a_strengths: List[str]
-    chunk_b_strengths: List[str]
-    chunk_a_weaknesses: List[str]
-    chunk_b_weaknesses: List[str]
-    deciding_dimension: str
+    winner_reason: str = "No reason provided."
+    chunk_a_strengths: List[str] = []
+    chunk_b_strengths: List[str] = []
+    chunk_a_weaknesses: List[str] = []
+    chunk_b_weaknesses: List[str] = []
+    deciding_dimension: str = "query_relevance"
 
     @model_validator(mode="after")
     def check_winner_consistency(self) -> "JudgeResponse":

backend/routers/retrieval_router.py

@@ -34,13 +34,12 @@ async def retrieve(request: QueryRequest, http_request: Request):
         if not limiter.allow(client_ip):
             raise HTTPException(
                 status_code=429,
-                detail="Rate limit exceeded. You can only make 2 LLM requests per minute."
+                detail="Rate limit exceeded. You can only make 2 LLM requests per minute.",
             )
         search_text = await get_hyde_text(request.search_text)
     else:
         search_text = request.search_text
 
-
     limit = request.top_k * 4 if request.use_reranking else request.top_k
 
     retrieved_chunks, embeddings = await process_retrieval(
@@ -83,7 +82,7 @@ async def compare(request: CompareRequest, http_request: Request):
         if not limiter.allow(client_ip):
             raise HTTPException(
                 status_code=429,
-                detail="Rate limit exceeded. You can only make 2 LLM requests per minute."
+                detail="Rate limit exceeded. You can only make 2 LLM requests per minute.",
             )
         retrieval_text = await get_hyde_text(request.search_text)
     else:
@@ -144,10 +143,9 @@ async def judge(request: JudgeRequest, http_request: Request):
     if not limiter.allow(client_ip):
         raise HTTPException(
             status_code=429,
-            detail="Rate limit exceeded. You can only make 2 LLM requests per minute."
+            detail="Rate limit exceeded. You can only make 2 LLM requests per minute.",
         )
     prompt = system_instructions.format(
-
         search_query=request.search_query,
         chunk_a=request.chunk_a,
         chunk_b=request.chunk_b,