rohangbs's picture
Update app.py
0132409 verified
import streamlit as st
import os
import numpy as np
from sentence_transformers import SentenceTransformer
import faiss
from openai import OpenAI
from PIL import Image
class IntegratedChatSystem:
def __init__(self, api_key: str, model_name: str, embedding_dim: int = 384):
self.api_key = api_key
self.model_name = model_name
self.embedding_model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
self.embedding_dim = embedding_dim
self.index = faiss.IndexFlatIP(embedding_dim)
self.metadata = []
self.client = OpenAI(api_key=api_key)
def _add_to_index(self, vector: np.ndarray, metadata: dict):
self.index.add(vector)
self.metadata.append(metadata)
def add_image(self, image_path: str, context_text: str):
filename = os.path.basename(image_path)
if not os.path.exists(image_path):
raise FileNotFoundError(f"Image not found: {image_path}")
embedding = self.embedding_model.encode(context_text)
embedding = np.expand_dims(embedding, axis=0)
self._add_to_index(embedding, {"filepath": filename, "context": context_text})
def chat(self, user_message: str, similarity_threshold: float = 0.7, top_k: int = 3):
message_embedding = self.embedding_model.encode(user_message)
message_embedding = np.expand_dims(message_embedding, axis=0)
distances, indices = self.index.search(message_embedding, top_k)
relevant_images = [
self.metadata[i] for i, distance in zip(indices[0], distances[0])
if i != -1 and distance >= similarity_threshold
]
system_prompt = """You are an assistant chatbot. You should help the user by answering their question."""
enhanced_message = user_message
if relevant_images:
image_contexts = "\n".join(f"- {img['context']}" for img in relevant_images)
enhanced_message = f"{user_message}\n\nContext from relevant images:\n{image_contexts}"
try:
completion = self.client.chat.completions.create(
model=self.model_name,
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": enhanced_message}
]
)
response = completion.choices[0].message.content
return {
"response": response,
"images": relevant_images if relevant_images else None
}
except Exception as e:
print(f"Error calling OpenAI API: {str(e)}")
return {
"response": "I apologize, but I encountered an error processing your request.",
"images": None
}
# Initialize the chat system
api_key = ""
model_name = "ft:gpt-3.5-turbo-0125:brenin::AlVMkeUb"
chat_system = IntegratedChatSystem(api_key, model_name)
# Add images
image_folder = "images"
chat_system.add_image(os.path.join(image_folder, "sequence diagram.png"), "A diagram showing the sequence of how it is supposed to work. What is the sequence?")
chat_system.add_image(os.path.join(image_folder, "UX workflow.png"), "A flowchart of showing the UX workflow.What is the UX workflow")
chat_system.add_image(os.path.join(image_folder, "UI.png"), "A diagram the UI. What is the UI? ")
chat_system.add_image(os.path.join(image_folder, "workflow.png"), "A flowchart of showing the workflow. What is the workflow?")
# Streamlit UI
st.title("Chat with Integrated Image Context")
st.sidebar.title("Chat System")
user_message = st.text_input("Your message:", placeholder="Type your message here...")
if st.button("Send"):
if user_message.strip():
result = chat_system.chat(user_message)
st.write(f"**Assistant:** {result['response']}")
if result["images"]:
st.write("Relevant Images:")
for img in result["images"]:
image_path = os.path.join(image_folder, img["filepath"])
if os.path.exists(image_path):
st.image(Image.open(image_path), caption=img["context"])
else:
st.write(f"Image not found: {img['filepath']}")
else:
st.error("Please enter a message.")