Issue generating response from the knowledge base: "I'm sorry, but I need more information to be able to..."

[mshedeed85]

Hi, I am using the Docker version of lightRAG "LightRAG Server v1.4.6/0196 " with Ollama for LLM and Embedding. I ingested 3 documents each were chunked to 2 Chunks each. The retrieval works fine. but when I ingested another document that had 38 Chunks, and tried to retrieve an info, the answer I get is always like "I'm sorry, but I need more information to proceed. Could you please clarify what knowledge graph (KG) you are referring to? Is it a specific application or system that you have access to, such as a company's internal knowledge base, a research project, or an online service like ConceptNet or YAGO?"

Although from the lightRag logs I see there were entities are returned and also the relationships:
INFO: Process 1 building query context...
INFO: Query nodes: KG documents, List all documents, top_k: 200, cosine: 0.2
INFO: limit_async: 2 new workers initialized
INFO: Local query: 200 entites, 165 relations
INFO: Query edges: Document retrieval, Knowledge Graph, top_k: 200, cosine: 0.2
INFO: Global query: 192 entites, 200 relations
INFO: Naive query: 44 chunks (chunk_top_k: 200)
INFO: Truncated KG query results: 89 entities, 90 relations
INFO: KG related chunks: 39 from entitys, 31 from relations
INFO: Final context: 89 entities, 90 relations, 7 chunks
INFO: limit_async: 1 new workers initialized
INFO: 172.21.0.1:59860 - "POST /query/stream HTTP/1.1" 200

I checked the llm.py in order to update the rag_response prompt, but no difference...

What to check and how to fix that ?