Attempting to debug

app.py

@@ -12,20 +12,18 @@ import torch
 import numpy as np
 
 # upload knowledge base - from sentiment analysis lab
-with open("essay_writing.txt", "r", encoding="utf-8") as file: 
-    essay_writing = file.read()
+with open("essay_writing.txt", "r", encoding="utf-8") as f: 
+    essay_writing = f.read()
 
 # split the text into chunks 
-cleaned_text = essay_writing.strip()
-chunks = cleaned_text.split("\n")
-cleaned_chunks = [chunk.strip() for chunk in chunks if chunk.strip()]
+cleaned_chunks = [chunk.strip() for chunk in essay_writing.split("\n\n") if chunk.strip()]
 
 # load an embedding model 
 model = SentenceTransformer('all-MiniLM-L6-v2')
 
 chunk_embeddings = model.encode(cleaned_chunks, convert_to_tensor=True)
 
-def pull_relevant_info(query): 
+def pull_relevant_info(query, top_k=3): 
     query_embedding = model.encode(query, convert_to_tensor=True)
     query_embedding_normalized = query_embedding / query_embedding.norm()
 
@@ -33,9 +31,9 @@ def pull_relevant_info(query):
 
     similarities = torch.matmul(chunk_embeddings_normalized, query_embedding_normalized)
 
-    top_indices = torch.topk(similarities, k=3).indices.cpu().numpy()
+    top_indices = torch.topk(similarities, k=top_k).indices.cpu().numpy()
 
-    relevant_info = "\n".join([chunks[i] for i in top_indices])
+    relevant_info = "\n\n".join([chunks[i] for i in top_indices])
 
     return relevant_info
 
@@ -43,7 +41,7 @@ client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
 def respond(message, history): 
     
-    info = pull_relevant_info(message)
+    info = pull_relevant_info(message, top_k=3)
     system_message = (f"You are a friendly chatbot. Use the following information to help answer the user's question:\n{info}\n")
     messages = [{"role": "system", "content": system_message}]