Aishu-7's picture
Update app.py
be5f43d verified
Raw
History Blame Contribute Delete
3.46 kB
from sentence_transformers import SentenceTransformer
import gradio as gr
from huggingface_hub import InferenceClient
import numpy as np
import torch
import os
import gradio as gr
#pip install https://gradio-builds.s3.amazonaws.com/75c684efb87624bee2fb63b08122564e6538509e/gradio-6.17.3-py3-none-any.whl
def image_classifier(inp):
return {'cat': 0.3, 'dog': 0.7}
demo = gr.Interface(fn=image_classifier, inputs="image", outputs="label")
demo.launch()
with open("knowledge.txt", "r", encoding="utf-8") as file:
knowledge_base = file.read()
def preprocess_text(text):
cleaned_text = text.strip()
chunks = cleaned_text.split("\n")
cleaned_chunks = []
for chunk in chunks:
stripped_chunk = chunk.strip()
if len (stripped_chunk)>0:
cleaned_chunks.append(stripped_chunk)
#print(cleaned_chunks)
#print (len(cleaned_chunks))
return cleaned_chunks
cleaned_chunks = preprocess_text(knowledge_base)
model = SentenceTransformer('all-MiniLM-L6-v2')
def create_embeddings(text_chunks):
chunk_embeddings = model.encode(text_chunks, convert_to_tensor=True) # Replace ... with the text_chunks list
#print(chunk_embeddings)
#print(chunk_embeddings.shape)
return chunk_embeddings
chunk_embeddings = create_embeddings(cleaned_chunks)# Complete this line
def get_top_chunks(query, chunk_embeddings, text_chunks):
query_embedding = model.encode(query, convert_to_tensor=True) # Complete this line
query_embedding_normalized = 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_normalized) # Complete this line
#print(similarities)
top_indices = torch.topk(similarities, k=3).indices
#print(top_indices)
top_chunks = []
for i in top_indices:
chunk = text_chunks[i]
top_chunks.append(chunk)
return top_chunks
top_results = get_top_chunks("Your account has been compromised", chunk_embeddings, cleaned_chunks) # Complete this line
#print(top_results)
#with gr.Blocks(theme=gr.themes.Default(primary_hue=gr.themes.colors.red, secondary_hue=gr.themes.colors.pink)) as demo:
cleaned_chunks = preprocess_text(knowledge_base)
client = InferenceClient("Qwen/Qwen2.5-7B-Instruct", token=os.getenv("ByteShield_Token"))
def respond(message, history):
top_chunks = get_top_chunks(message, chunk_embeddings, cleaned_chunks)
context = "\n".join(top_chunks)
messages = [{"role": "system","content": f"You are a friendly, tech expert chatbot. Use this context to answer:\n{context}"}]
if history:
messages.extend(history)
messages.append({"role": "user", "content": message})
response = client.chat_completion(
messages,
max_tokens=1000
)
return response.choices[0].message.content.strip()
with gr.Blocks(theme=gr.themes.Ocean()) as demo:
chatbot = gr.ChatInterface(
fn = respond,
cache_examples = False,
textbox=gr.Textbox(placeholder="Ask me anything!", container=False, scale=7),
title = "ByteShield - Your AI Gaurdian for Online Safety",
description = "Ask me anything about online safety!",
examples = ["Generate me some strong passwords to use.", "What are some security measures I can take to stay safe online?", "Explain how a data breach works.", "How do I know if a message is a scam or not?"]
)
demo.launch()