Convert personality txt files to yaml, replace print with log, and update app.py

ai_ref_kits/agentic_llm_rag/app.py

@@ -1,3 +1,4 @@
+
 import argparse
 import io
 import logging
@@ -13,6 +14,7 @@
 import openvino.properties.hint as hints
 import openvino.properties.streams as streams
 import requests
+import yaml
 from llama_index.core import PromptTemplate
 from llama_index.core import SimpleDirectoryReader
 from llama_index.core import VectorStoreIndex, Settings
@@ -149,7 +151,7 @@ def _handle_user_message(user_message, history):
 
 
     def _generate_response(chat_history, log_history):
-        print("log_history", log_history)
+        log.info(f"log_history {log_history}")
         if not isinstance(log_history, list):
             log_history = []
 
@@ -215,7 +217,7 @@ def run():
                 chat_window = gr.Chatbot(
                     label="Paint Purchase Helper",
                     avatar_images=(None, "https://docs.openvino.ai/2024/_static/favicon.ico"),
-		                height=400,  # Adjust height as per your preference
+                                height=400,  # Adjust height as per your preference
                     scale=2  # Set a higher scale value for Chatbot to make it wider
                     #autoscroll=True,  # Enable auto-scrolling for better UX
                 )
@@ -283,7 +285,7 @@ def run():
     parser.add_argument("--chat_model", type=str, default="model/llama3.1-8B-INT4", help="Path to the chat model directory")
     parser.add_argument("--embedding_model", type=str, default="model/bge-large-FP32", help="Path to the embedding model directory")
     parser.add_argument("--rag_pdf", type=str, default="test_painting_llm_rag.pdf", help="Path to a RAG PDF file with additional knowledge the chatbot can rely on.")
-    parser.add_argument("--personality", type=str, default="paint_concierge_personality.txt", help="Path to the TXT file with chatbot personality")
+    parser.add_argument("--personality", type=str, default="paint_concierge_personality.yaml", help="Path to the yaml file with chatbot personality")
 
     args = parser.parse_args()
 
@@ -299,7 +301,7 @@ def run():
     # Step 4: Load documents and create the VectorStoreIndex
     text_example_en_path = Path(args.rag_pdf)
     index = load_documents(text_example_en_path)
-    print("loading in", index)
+    log.info(f"loading in {index}")
     vector_tool = QueryEngineTool(
         index.as_query_engine(streaming=True),
         metadata=ToolMetadata(
@@ -313,11 +315,12 @@ def run():
 
     # Load agent config
     personality_file_path = Path(args.personality)
-    with open(personality_file_path) as f:
-        chatbot_config = f.read()
 
-    react_system_prompt = PromptTemplate(chatbot_config)
-    print("react_system_prompt", react_system_prompt)
+    with open(personality_file_path, "rb") as f:
+        chatbot_config = yaml.safe_load(f)
+
+    react_system_prompt = PromptTemplate(chatbot_config['system_configuration'])
+    log.info(f"react_system_prompt {react_system_prompt}")
     #Define agent and available tools
     agent = ReActAgent.from_tools([multiply_tool, divide_tool, add_tool, subtract_tool, paint_cost_calculator, vector_tool],
                                   llm=llm,

ai_ref_kits/agentic_llm_rag/paint_concierge_personality.yaml

@@ -0,0 +1,8 @@
+system_configuration: >
+  You are a helpful, respectful, and knowledgeable Paint Concierge working at a retail store, where customer experience is absolutely crucial.
+  Your role is to assist customers with inquiries about paint suggestions, price details, supply calculations, and product recommendations based on the knowledge and documents provided to you.
+  You are strongly encouraged to use various tools that have been provided to you, including the vector_search tool, various math tools, and a paint calculation tool (calculate_paint_cost tool).
+  Answer questions with the information available from the RAG document (vector_search tool) or other shared knowledge, but if you're unsure or don't have specific details, politely inform the customer to check with the store staff or the official product website for further information.
+  When answering questions about the number of gallons of paint needed or the cost of paint or similar queries, always consult the RAG document first.
+  Do not assume or provide any speculative information or estimates outside the shared knowledge base. Always encourage customers to verify with store associates for unavailable or unknown details.
+  Do not ask for personal information or provide any responses that are inappropriate or unethical. Always remain professional, empathetic, and polite.

ai_ref_kits/agentic_llm_rag/store_employee_personality.yaml

@@ -0,0 +1,9 @@
+system_configuration: >
+  You are a helpful, respectful, and knowledgeable Paint Employee Concierge working at a retail store, designed to help new employees with their onboarding experience and complex questions from customers.
+  Your role is to assist new employees with inquiries about paint suggestions, price details, supply calculations, and product recommendations based on the knowledge and documents provided to you.
+  You may be asked to test employees' knowledge on paint, or guide them towards accurate answers.
+  You are strongly encouraged to use various tools that have been provided to you, including the vector_search tool, various math tools, and a paint calculation tool (calculate_paint_cost tool).
+  Answer questions with the information available from the RAG document (vector_search tool) or other shared knowledge, but if you're unsure or don't have specific details, politely inform the customer to check with senior store staff or the official product website for further information.
+  When answering questions about the number of gallons of paint needed or the cost of paint or similar queries, always consult the RAG document first.
+  Do not assume or provide any speculative information or estimates outside the shared knowledge base. Always encourage junior employees to verify with store associates for unavailable or unknown details.
+  Do not ask for personal information or provide any responses that are inappropriate or unethical. Always remain professional, empathetic, and polite.