Feature/export prompts (#137)

* Remove .env from tracking and add it to .gitignore

Signed-off-by: Gianluca Capuzzi <gianluca.posta78@gmail.com>

* Export prompts on files

Signed-off-by: Gianluca Capuzzi <gianluca.posta78@gmail.com>

---------

Signed-off-by: Gianluca Capuzzi <gianluca.posta78@gmail.com>

Author: gcapuzzi

.gitignore

@@ -79,4 +79,6 @@ logs/
 # Documentation build
 _build/
 _static/
-_templates/
+_templates/
+
+config.yaml

src/mvt/config.yaml

@@ -1,5 +1,5 @@
 import os
-from utils import load_yaml_file_with_db_prompts
+from utils import load_yaml_file_with_db_prompts, get_system_prompt
 from dotenv import load_dotenv, find_dotenv
 from langchain_mistralai.chat_models import ChatMistralAI
 from langchain_community.vectorstores import FAISS
@@ -60,7 +60,7 @@ def get_ragchain(filter):
     retriever = docsearch.as_retriever(search_kwargs={"k": config_data["nr_retrieved_documents"], "filter": filter})
 
     # read prompt string from config file
-    prompt_str = config_data["system_prompt"]
+    prompt_str = get_system_prompt(config_data["system_prompt"])
 
     # Answer question
     qa_system_prompt = (

src/mvt/main.py

@@ -38,15 +38,14 @@ def query_rewriting_llm(user_query, context="Founder Institute Keystone Chapter"
             temperature=0.7
         )
 
-    query_rewriting_prompt = config_data["query_rewriting_prompt"]
+    # Read query rewriting prompt from config
+    with open(config_data["query_rewriting_prompt"], "r") as f:
+        query_rewriting_prompt = f.read().strip()
 
     messages = [
         ("system", query_rewriting_prompt),
         ("human", user_query),
     ]
 
-    print(messages)
-
     response = model.invoke(messages)
-    print(response.content)
     return response.content

src/mvt/query_rewriting.py

@@ -0,0 +1 @@
+You are an assistant helping to rewrite user queries to make them more specific and effective for searching documents. The context is the Founder Institute Keystone Chapter which includes information about programs, fees, mentors, and startup guidance. Please rewrite the human query to be more specific, detailed, and optimized for document retrieval, considering the context mentioned

src/mvt/query_rewriting_prompt.txt

@@ -0,0 +1 @@
+You are an assistant designed to help answer questions about startup founders from the Founder Institute (FI) Keystone Chapter. The Keystone Chapter is part of the global network of the Founder Institute, the world's largest pre-seed startup accelerator. It supports entrepreneurs across the Delaware Valley, Philadelphia, and Princeton areas, with a mission to help them build meaningful and lasting technology companies. Use any relevant information from the retrieved context to answer questions as accurately and helpfully as possible. If the answer is not found in the provided context, simply say you don't know. Avoid vague phrases like 'based on the provided context' in your responses, and aim to be as detailed and informative as possible. If the question is ambiguous, assume the user is referring to the Founder Institute Keystone Chapter's core startup accelerator program, unless the question clearly states otherwise. If the question is not related to the Founder Institute or its programs, politely decline to answer

src/mvt/responses.txt

@@ -41,6 +41,10 @@ def load_yaml_file_with_db_prompts(path):
 
     return data
 
+# This function returns the system prompt from the file specified in the config.
+def get_system_prompt(prompt_file: str):
+    with open(prompt_file, "r") as f:
+        return f.read().strip()
 
 # This function extracts text from HTML while preserving context relationships.
 def bs4_html(html):