increased context to 20, updateed prompt in app.py

app.py

@@ -151,11 +151,11 @@ def predict(message, history, request: gr.Request):
     
 
     # Retrieve relevant documents for the current message
-    relevant_docs = vectorstore.similarity_search(message,k=6)  #  retriever
+    relevant_docs = vectorstore.similarity_search(message,k=20)  #  retriever
     
     # Build context from retrieved documents
     context = "\nExtracted documents:\n" + "\n".join([
-        f"Content document {i}: {doc.page_content}\n\n---"
+        f"Content document {i+1}: {doc.page_content}\n\n---"
         for i, doc in enumerate(relevant_docs)
     ])
 
@@ -163,7 +163,7 @@ def predict(message, history, request: gr.Request):
 
     # RAG tool
     RAG_PROMPT_TEMPLATE="""You will be asked information related to Rémi Cazelles's specific projects, work and education.
-                        Using the information contained in the context, provide a comprehensive answer to the question.
+                        Using the information contained in the context, provide a structured answer to the question.
                         Respond to the question asked with enought details, response should be precise and relevant to the question.
                         """
 
@@ -179,7 +179,7 @@ def predict(message, history, request: gr.Request):
     gpt_response = llm.invoke(
         messages,
         config={
-            "tags": ["Testing", 'RAG-Bot', 'V1','Host_on_HF'],
+            "tags": ["Testing", 'RAG-Bot', 'V2','Host_on_HF'],
             "metadata": {
                 "rag_llm": "gpt-5-nano",
                 "num_retrieved_docs": len(relevant_docs),
@@ -190,17 +190,9 @@ def predict(message, history, request: gr.Request):
     messages.append(AIMessage(content=gpt_response.content))
 
     try :
-        raw_source_lines = [
-        f"{i+1} : {format_source(doc)})\n---"
-        for i, doc in enumerate(relevant_docs)]
-
-        seen = set()
-        unique_source_lines = []
-        for line in raw_source_lines:
-            if line not in seen:          
-                seen.add(line)
-                unique_source_lines.append(line)
-        source_context = "\nSources:" + "\n".join(unique_source_lines)
+        source_context = "\n\nSources:\n" + "\n".join([
+        f"{i+1} - {format_source(doc)}"
+        for i, doc in enumerate(relevant_docs)])
 
     except :
         source_context = "Issue extracting source"