Update app.py

app.py

@@ -1,3 +1,9 @@
+
+
+Share
+
+
+You said:
 import gradio as gr
 from sentence_transformers import SentenceTransformer, util
 from transformers import GPT2LMHeadModel, GPT2Tokenizer
@@ -40,75 +46,18 @@ def load_and_preprocess_text(filename):
 
 segments = load_and_preprocess_text(filename)
 
-def find_relevant_segment(user_query, segments, similarity_threshold=0.5):
-    """
-    Find the most relevant text segment based on a user query.
-    Parameters:
-    - user_query (str): The user's query.
-    - segments (list[str]): List of text segments to search within.
-    - similarity_threshold (float): Minimum similarity required to consider a segment relevant.
-    Returns:
-    - str: The most relevant text segment.
-    """
+def find_relevant_segment(user_query, segments):
     try:
-        query_embedding = retrieval_model.encode(user_query)
+        lower_query = user_query.lower()
+        query_embedding = retrieval_model.encode(lower_query)
         segment_embeddings = retrieval_model.encode(segments)
         similarities = util.pytorch_cos_sim(query_embedding, segment_embeddings)[0]
         best_idx = similarities.argmax()
-        if similarities[best_idx].item() >= similarity_threshold:
-            return segments[best_idx]
-        else:
-            return "Sorry, I couldn't find a specific match. Here are some general tips to help you:"
+        return segments[best_idx]
     except Exception as e:
-        print(f"Error finding relevant segment: {e}")
+        print(f"Error in finding relevant segment: {e}")
         return ""
 
-def clean_up_response(response, segment):
-    """
-    Clean up the generated response to ensure it is tidy and presentable.
-    Parameters:
-    - response (str): The initial response generated by the model.
-    - segment (str): The segment used to generate the response.
-    Returns:
-    - str: A cleaned and formatted response.
-    """
-    sentences = response.split('.')
-    cleaned_sentences = [sentence.strip() for sentence in sentences if sentence.strip() and sentence.strip().lower() not in segment.lower()]
-    cleaned_response = '. '.join(cleaned_sentences).strip()
-    if cleaned_response and not cleaned_response.endswith((".", "!", "?")):
-        cleaned_response += "."
-    return cleaned_response
-
-
-def generate_response_with_context(user_query, relevant_segment):
-    """
-    Generate a response based on a user query and a relevant segment.
-    
-    Parameters:
-    - user_query (str): The user's query.
-    - relevant_segment (str): A relevant fact or detail.
-    
-    Returns:
-    - str: Formatted response incorporating the relevant segment.
-    """
-    try:
-        # Prepare the prompt incorporating the relevant segment
-        prompt = f"User: {user_query}\n\nAssistant: Here is some helpful information based on your topic: {relevant_segment}"
-        
-        # Calculate the maximum tokens allowed for the response
-        max_tokens = len(tokenizer(prompt)['input_ids']) + 100
-        
-        # Generate the response using the model
-        response = gpt_model(prompt, max_length=max_tokens, temperature=0.7)[0]['generated_text']
-        
-        # Clean up the response for better formatting and clarity
-        return clean_up_response(response, relevant_segment)
-    
-    except Exception as e:
-        print(f"Error generating response: {e}")
-        return "I'm sorry, but there was an error generating your response. Please try again."
-
-
 def generate_response(user_query, relevant_segment):
     try:
         user_message = f"Here's the information on your request: {relevant_segment}"