Tune parameters

src/haiku/rag/client.py

@@ -278,7 +278,7 @@ class HaikuRAG:
         return await self.document_repository.list_all(limit=limit, offset=offset)
 
     async def search(
-        self, query: str, limit: int = 3, k: int = 60, rerank=Config.RERANK
+        self, query: str, limit: int = 5, k: int = 60, rerank=Config.RERANK
     ) -> list[tuple[Chunk, float]]:
         """Search for relevant chunks using hybrid search (vector similarity + full-text search) with reranking.
 
@@ -298,7 +298,6 @@ class HaikuRAG:
         search_results = await self.chunk_repository.search_chunks_hybrid(
             query, limit * 3, k
         )
-
         # Apply reranking
         reranker = get_reranker()
         chunks = [chunk for chunk, _ in search_results]

src/haiku/rag/qa/ollama.py

@@ -4,7 +4,7 @@ from haiku.rag.client import HaikuRAG
 from haiku.rag.config import Config
 from haiku.rag.qa.base import QuestionAnswerAgentBase
 
-OLLAMA_OPTIONS = {"temperature": 0.0, "seed": 42, "num_ctx": 64000}
+OLLAMA_OPTIONS = {"temperature": 0.0, "seed": 42, "num_ctx": 16384}
 
 
 class QuestionAnswerOllamaAgent(QuestionAnswerAgentBase):

src/haiku/rag/qa/prompts.py

@@ -15,7 +15,7 @@ Guidelines:
 - Indicate when information is incomplete or when you need to search for additional context
 - If the retrieved documents don't contain sufficient information, clearly state: "I cannot find enough information in the knowledge base to answer this question."
 - For complex questions, consider breaking them down and performing multiple searches
-- Stick to the answer, do not ellaborate or provde context unless asked for it.
+- Stick to the answer, do not ellaborate or provide context unless explicitly asked for it.
 
 Be concise, and always maintain accuracy over completeness. Prefer short, direct answers that are well-supported by the documents.
 """

tests/llm_judge.py

@@ -13,7 +13,7 @@ class LLMJudgeResponseSchema(BaseModel):
 class LLMJudge:
     """LLM-as-judge for evaluating answer equivalence using Ollama."""
 
-    def __init__(self, model: str = "qwen3"):
+    def __init__(self, model: str = Config.QA_MODEL):
         self.model = model
         self.client = AsyncClient(host=Config.OLLAMA_BASE_URL)