lets see

app.py

@@ -2,15 +2,15 @@ import gradio as gr
 import os
 import pinecone
 import time
-# from torch import cuda
 from langchain.embeddings.huggingface import HuggingFaceEmbeddings
-# import PyPDF2
-# import re
+import torch
+import sentence_transformers
 from langchain.vectorstores import Pinecone
 from langchain import HuggingFaceHub, LLMChain
 from langchain.prompts import PromptTemplate
 from langchain.chains import RetrievalQA
 
+headers = {"Authorization": f"Bearer {os.environ.get('API_KEY')}"}
 embed_model_id = 'sentence-transformers/all-MiniLM-L6-v2'
 # device = f'cuda:{cuda.current_device()}' if cuda.is_available() else 'cpu'
 
@@ -48,8 +48,6 @@ index = pinecone.Index(index_name)
 index.describe_index_stats()
 
 
-
-
 text_field = 'text'  # field in metadata that contains text content
 
 vectorstore = Pinecone(
@@ -57,7 +55,7 @@ vectorstore = Pinecone(
 )
 
 
-hub = HuggingFaceHub(repo_id = "HuggingFaceH4/zephyr-7b-beta",huggingfacehub_api_token="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")
+hub = HuggingFaceHub(repo_id = "HuggingFaceH4/zephyr-7b-beta",huggingfacehub_api_token={os.environ.get('API_KEY')})
 print(hub)
 prompt = PromptTemplate(
     input_variables=["question"],
@@ -70,7 +68,9 @@ rag_pipeline = RetrievalQA.from_chain_type(
 )
 
 def question(question):
+    global chatbot
     answer = rag_pipeline(question)
+    chatbot = answer
     return answer
 
 demo = gr.Interface(fn=question, inputs="text", outputs="text")