try

app.py

@@ -1,114 +1,60 @@
 import gradio as gr
-import random
 import os
 from huggingface_hub import InferenceClient
 from sentence_transformers import SentenceTransformer
 import torch
 
+# Load and prepare knowledge base
 with open("knowledge.txt", "r", encoding="utf-8") as file:
-  recent = file.read()
-  # opens the text, saves as "file"
-  # reads the text and saves as water_cycle_text variable
-
-
-
-cleaned_text = recent.strip()
-# cleaning up the text
-chunks = cleaned_text.split("\n")
-# seperating the text into one sentence pieces
-cleaned_chunks = []
-# creating an empty list to put the cleaned chunks in
-
-for chunk in chunks:
-  stripped_chunk = chunk.strip()
-  if stripped_chunk:
-    cleaned_chunks.append(stripped_chunk)
-    # loop through chunks and add not empty chunks to cleaned_chunks list
-
+    recent = file.read()
 
+chunks = [chunk.strip() for chunk in recent.strip().split("\n") if chunk.strip()]
 model = SentenceTransformer('all-MiniLM-L6-v2')
+chunk_embeddings = model.encode(chunks, convert_to_tensor=True)
 
-chunk_embeddings = model.encode(cleaned_chunks, convert_to_tensor=True)
-# encode the model, pass through my cleaned chunks and convert to vector embeddings (not arrays)
-
-
-def get_top_chunks(query):
-  # create my function taking query as parameter
-  query_embedding = model.encode(query, convert_to_tensor=True)
-  # encode query to vector embedding for comparison
-  query_embedding_normalized = query_embedding / query_embedding.norm()
-  # normalize my query to 1; allows for comparison of meaning
-  chunk_embeddings_normalized = chunk_embeddings / chunk_embeddings.norm(dim=1, keepdim=True)
-  # normailizing chunks for comparison of meaning
-
-  similarities = torch.matmul(chunk_embeddings_normalized, query_embedding_normalized)
-
-  # using matmul (matrix multiplication method) to compare query to chunks
-  top_indices = torch.topk(similarities, k=3).indices
-
-  # get the indices of the chunks that are most similar to query
-
-  top_chunks = []
-
-  for i in top_indices:
-    chunk = chunks[i]
-    # for each index number in top_indices, get back the text
-    top_chunks.append(chunk)
-    # values of each index number is added to top_chunks
-  return top_chunks
-    
+# Hugging Face Inference Client (add token if required)
 client = InferenceClient(
-    model='Qwen/Qwen2.5-72B-Instruct'
-    #token = os.getenv('HF_TOKEN')
-) 
+    model='Qwen/Qwen2.5-72B-Instruct',
+    # token=os.getenv("HF_TOKEN")
+)
 
-#client is where you can change the LLM model!
-def respond(message,history):
-    if not message.strip():
-        return "Hello!"
-    gift_ideas = get_top_chunks(message)
-    messages = [{'role': 'system', 'content': f'You give really good gift ideas and are super helpful! You also tell me the price of each item. Give me 5 gift ideas if I ask. Use the following database for gift ideas: {gift_ideas}'}]
-    
-    if history:
-        messages.extend(history)
-    
-    messages.append({"role": "user", "content": message})
+# Top similar chunks
+def get_top_chunks(query):
+    query_embedding = model.encode(query, convert_to_tensor=True)
+    query_embedding /= query_embedding.norm()
+    chunk_embeddings_normalized = chunk_embeddings / chunk_embeddings.norm(dim=1, keepdim=True)
+    similarities = torch.matmul(chunk_embeddings_normalized, query_embedding)
+    top_indices = torch.topk(similarities, k=3).indices
+    return [chunks[i] for i in top_indices]
+
+# Main response function
+def respond(user_message, chat_history):
+    gift_ideas = get_top_chunks(user_message)
+    messages = [{"role": "system", "content": f"You are helpful and give 5 gift ideas with prices. Use this database: {gift_ideas}"}]
+    for user, bot in chat_history:
+        messages.append({"role": "user", "content": user})
+        messages.append({"role": "assistant", "content": bot})
+    messages.append({"role": "user", "content": user_message})
 
     response = ""
-    for message in client.chat_completion(
-        #max_tokens controls how many words your responses is 
-        messages,
-        max_tokens = 500,
-        # stream = True,
+    for part in client.chat_completion(messages, max_tokens=500):
+        token = part.choices[0].delta.content
+        if token:
+            response += token
+
+    chat_history.append((user_message, response))
+    return "", chat_history
+
+# Gradio app
+with gr.Blocks(theme='hmb/amethyst') as demo:
+    gr.Markdown("""
+        # 🎁 Introducing WrapIT!
+        **WrapIT** helps users find personalized gift ideas and craft thoughtful card messages**  
+        by inputting details like the recipient's interests, celebration type, and budget.
         
-        #temperature = 0.8, #code a decimal between 0-2
-        #top_p = .65 #code a decimal between 0-1
-    ):
-        token = message.choices[0].delta.content
-        response += token
-        yield response
-    #print(response["choices"][0]["message"]["content"].strip())
-    
-    #yield response["choices"][0]["message"]["content"].strip()
-    
-chatbot = gr.ChatInterface(respond, type='messages')
+        ✨ *All you have to do is wrap it.*
+    """)
 
-with gr.Blocks(theme='hmb/amethyst') as chatbot:
-    with gr.Row(equal_height=True):
-        with gr.Column(scale=10):
-            gr.Markdown(
-                """
-                # 🎁 Introducing WrapIT!
-                **WrapIT** helps users find personalized gift ideas and craft thoughtful card messages  
-                by inputting details like the recipient's interests, celebration type, and budget.
-                
-                ✨ *All you have to do is wrap it.*
-                """
-            )
-            #gr.chatInterface()
-
-  
-    
     chatbot = gr.Chatbot()
     msg = gr.Textbox(placeholder="Say hi to WrapIT here!", label="Message")
     send = gr.Button("Send")
@@ -116,5 +62,4 @@ with gr.Blocks(theme='hmb/amethyst') as chatbot:
     msg.submit(respond, [msg, chatbot], [msg, chatbot])
     send.click(respond, [msg, chatbot], [msg, chatbot])
 
-
-chatbot.launch()
+demo.launch()