improved app.py

app.py

@@ -13,6 +13,7 @@ import chardet
 import gradio as gr
 import pandas as pd
 import json
+import re
 
 # Enable logging for debugging
 logging.basicConfig(level=logging.DEBUG)
@@ -70,13 +71,27 @@ def load_documents(file_paths):
             logger.error(f"Error processing file {file_path}: {e}")
     return docs
 
+# Function to ensure the response ends with a complete sentence
+def ensure_complete_sentences(text):
+    # Use regex to find all complete sentences
+    sentences = re.findall(r'[^.!?]*[.!?]', text)
+    if sentences:
+        return sentences[-1].strip()
+    return text  # Return as is if no complete sentence is found
+
 # Initialize the LLM using ChatGroq with GROQ's API
 def initialize_llm(model, temperature, max_tokens):
     try:
+        # Allocate some tokens for the prompt (e.g., 50 tokens)
+        prompt_tokens = 50
+        response_max_tokens = max_tokens - prompt_tokens
+        if response_max_tokens <= 0:
+            raise ValueError("max_tokens is too small to allocate for the response.")
+
         llm = ChatGroq(
             model=model,
             temperature=temperature,
-            max_tokens=max_tokens,
+            max_tokens=response_max_tokens,  # Adjusted max_tokens
             api_key=api_key  # Ensure the API key is passed correctly
         )
         logger.debug("LLM initialized successfully.")
@@ -114,7 +129,7 @@ def create_rag_pipeline(file_paths, model, temperature, max_tokens):
         custom_prompt_template = PromptTemplate(
             input_variables=["context", "question"],
             template="""
-            You are an AI assistant with expertise in daily wellness and your aim is to give detailed solutions regarding it.
+            You are an AI assistant with expertise in daily wellness. Your aim is to provide detailed yet concise solutions regarding daily wellness topics.
 
             Context:
             {context}
@@ -122,7 +137,7 @@ def create_rag_pipeline(file_paths, model, temperature, max_tokens):
             Question:
             {question}
 
-            Provide a detailed answer, including relevant examples and a suggested schedule.
+            Provide a detailed but concise answer, ensuring that it is complete and does not end abruptly. Include relevant examples and a suggested schedule.
             """
         )
 
@@ -138,7 +153,7 @@ def create_rag_pipeline(file_paths, model, temperature, max_tokens):
         logger.error(f"Error creating RAG pipeline: {e}")
         return None, f"Error creating RAG pipeline: {e}"
 
-# Function to answer questions
+# Function to answer questions with post-processing
 def answer_question(file_paths, model, temperature, max_tokens, question):
     rag_chain, message = create_rag_pipeline(file_paths, model, temperature, max_tokens)
     if rag_chain is None:
@@ -146,7 +161,9 @@ def answer_question(file_paths, model, temperature, max_tokens, question):
     try:
         answer = rag_chain.run(question)
         logger.debug("Question answered successfully.")
-        return answer
+        # Post-process to ensure the answer ends with a complete sentence
+        complete_answer = ensure_complete_sentences(answer)
+        return complete_answer
     except Exception as e:
         logger.error(f"Error during RAG pipeline execution: {e}")
         return f"Error during RAG pipeline execution: {e}"
@@ -162,7 +179,7 @@ interface = gr.Interface(
     inputs=[
         gr.Textbox(label="Model Name", value="llama3-8b-8192"),
         gr.Slider(label="Temperature", minimum=0, maximum=1, step=0.01, value=0.7),
-        gr.Slider(label="Max Tokens", minimum=1, maximum=1024, step=1, value=500),
+        gr.Slider(label="Max Tokens", minimum=100, maximum=1024, step=1, value=500),
         gr.Textbox(label="Question")
     ],
     outputs="text",