Delete app.py
Browse files
app.py
DELETED
|
@@ -1,158 +0,0 @@
|
|
| 1 |
-
# -*- coding: utf-8 -*-
|
| 2 |
-
"""Doc_chat_vegleges.ipynb
|
| 3 |
-
|
| 4 |
-
Automatically generated by Colaboratory.
|
| 5 |
-
|
| 6 |
-
Original file is located at
|
| 7 |
-
https://colab.research.google.com/drive/1G34ZCuupJZxNy-CFxjMNIa4_I3jynKqC
|
| 8 |
-
|
| 9 |
-
# Setting up environment
|
| 10 |
-
"""
|
| 11 |
-
|
| 12 |
-
from PyPDF2 import PdfReader
|
| 13 |
-
from langchain.embeddings.openai import OpenAIEmbeddings
|
| 14 |
-
from langchain.text_splitter import CharacterTextSplitter
|
| 15 |
-
from langchain.vectorstores import ElasticVectorSearch, Pinecone, Weaviate, FAISS
|
| 16 |
-
|
| 17 |
-
# Get your API keys from openai, you will need to create an account.
|
| 18 |
-
# Here is the link to get the keys: https://platform.openai.com/account/billing/overview
|
| 19 |
-
import os
|
| 20 |
-
|
| 21 |
-
"""# Preprocessing document"""
|
| 22 |
-
|
| 23 |
-
# location of the pdf file/files.
|
| 24 |
-
reader = PdfReader('constitution.pdf')
|
| 25 |
-
#reader = PdfReader('/content/WOW.pdf')
|
| 26 |
-
#reader = PdfReader('/content/the_little_prince.pdf')
|
| 27 |
-
|
| 28 |
-
# read data from the file
|
| 29 |
-
raw_text = ''
|
| 30 |
-
for i, page in enumerate(reader.pages):
|
| 31 |
-
text = page.extract_text()
|
| 32 |
-
if text:
|
| 33 |
-
raw_text += text
|
| 34 |
-
|
| 35 |
-
# We need to split the text that we read into smaller chunks so that during information retreival we don't hit the token size limits.
|
| 36 |
-
|
| 37 |
-
text_splitter = CharacterTextSplitter(
|
| 38 |
-
separator = "\n",
|
| 39 |
-
chunk_size = 1000,
|
| 40 |
-
chunk_overlap = 200,
|
| 41 |
-
length_function = len,
|
| 42 |
-
)
|
| 43 |
-
texts = text_splitter.split_text(raw_text)
|
| 44 |
-
|
| 45 |
-
len(texts)
|
| 46 |
-
|
| 47 |
-
"""## Setting up doc search"""
|
| 48 |
-
|
| 49 |
-
embeddings = OpenAIEmbeddings()
|
| 50 |
-
doc_search = FAISS.from_texts(texts, embeddings)
|
| 51 |
-
|
| 52 |
-
"""# Setting up chatbot"""
|
| 53 |
-
|
| 54 |
-
from langchain.chains.question_answering import load_qa_chain
|
| 55 |
-
from langchain.memory import ConversationBufferWindowMemory
|
| 56 |
-
from langchain.prompts import PromptTemplate
|
| 57 |
-
from langchain_openai import OpenAI
|
| 58 |
-
|
| 59 |
-
template = """You are a chatbot having a conversation with a human.
|
| 60 |
-
|
| 61 |
-
Given the following extracted parts of a long document and a question, create a final answer based on the document ONLY and NOTHING else.
|
| 62 |
-
Any questions outside of the document is irrelevant and you certanly dont know!
|
| 63 |
-
|
| 64 |
-
{context}
|
| 65 |
-
|
| 66 |
-
{chat_history}
|
| 67 |
-
Human: {human_input}
|
| 68 |
-
Chatbot:"""
|
| 69 |
-
|
| 70 |
-
prompt = PromptTemplate(
|
| 71 |
-
input_variables=["chat_history", "human_input", "context"], template=template
|
| 72 |
-
)
|
| 73 |
-
|
| 74 |
-
memory = ConversationBufferWindowMemory(memory_key="chat_history", input_key="human_input",k=3)
|
| 75 |
-
chain = load_qa_chain( OpenAI(), chain_type="stuff", memory=memory, prompt=prompt)
|
| 76 |
-
|
| 77 |
-
"""## The chatbot should know the answer"""
|
| 78 |
-
|
| 79 |
-
query = "Who wrote the constitution?"
|
| 80 |
-
docs = doc_search.similarity_search(query)
|
| 81 |
-
|
| 82 |
-
chain({"input_documents": docs, "human_input": query}, return_only_outputs=True)['output_text']
|
| 83 |
-
|
| 84 |
-
# query = "Acronyms?"
|
| 85 |
-
# docs = doc_search.similarity_search(query)
|
| 86 |
-
|
| 87 |
-
# chain({"input_documents": docs, "human_input": query}, return_only_outputs=True)['output_text']
|
| 88 |
-
|
| 89 |
-
# query = "Say 3 of them"
|
| 90 |
-
# docs = doc_search.similarity_search(query)
|
| 91 |
-
|
| 92 |
-
# chain({"input_documents": docs, "human_input": query}, return_only_outputs=True)['output_text']
|
| 93 |
-
|
| 94 |
-
"""## The chatbot should not know the answer."""
|
| 95 |
-
|
| 96 |
-
query = "What is the capital of France?"
|
| 97 |
-
docs = doc_search.similarity_search(query)
|
| 98 |
-
|
| 99 |
-
chain({"input_documents": docs, "human_input": query}, return_only_outputs=True)['output_text']
|
| 100 |
-
|
| 101 |
-
"""## Previous chatbot (deprecated)"""
|
| 102 |
-
|
| 103 |
-
#print(chain.memory.buffer)
|
| 104 |
-
|
| 105 |
-
# from langchain.chains.question_answering import load_qa_chain
|
| 106 |
-
# from langchain.llms import OpenAI
|
| 107 |
-
|
| 108 |
-
# embeddings = OpenAIEmbeddings()
|
| 109 |
-
# doc_search = FAISS.from_texts(texts, embeddings)
|
| 110 |
-
|
| 111 |
-
# chain = load_qa_chain(OpenAI(), chain_type="stuff")
|
| 112 |
-
|
| 113 |
-
# query = "Who wrote the constitution?"
|
| 114 |
-
|
| 115 |
-
|
| 116 |
-
# answer = chain.run(input_documents=docs, question=query)
|
| 117 |
-
# print(answer)
|
| 118 |
-
|
| 119 |
-
# query = "What is the capital of france?"
|
| 120 |
-
# answer = chain.run(input_documents=docs, question=query)
|
| 121 |
-
# print(answer)
|
| 122 |
-
|
| 123 |
-
"""# Demo
|
| 124 |
-
|
| 125 |
-
## Setting up methods
|
| 126 |
-
"""
|
| 127 |
-
|
| 128 |
-
def chat(query,history):
|
| 129 |
-
docs = doc_search.similarity_search(query)
|
| 130 |
-
return chain({"input_documents": docs, "human_input": query}, return_only_outputs=True)['output_text']
|
| 131 |
-
|
| 132 |
-
"""## Setting up UI with gradio"""
|
| 133 |
-
|
| 134 |
-
import gradio as gr
|
| 135 |
-
|
| 136 |
-
css = """
|
| 137 |
-
body {
|
| 138 |
-
background-color: #FFFFFF; /* White background */
|
| 139 |
-
}
|
| 140 |
-
|
| 141 |
-
.gradio-chat-interface, .gradio-chat-input {
|
| 142 |
-
background-color: #E0FFFF; /* Light blue-green background */
|
| 143 |
-
}
|
| 144 |
-
"""
|
| 145 |
-
|
| 146 |
-
gr.ChatInterface(
|
| 147 |
-
chat,
|
| 148 |
-
chatbot=gr.Chatbot(height=500),
|
| 149 |
-
title="Doc-chat",
|
| 150 |
-
description="Ask about the constitution!",
|
| 151 |
-
theme="soft",
|
| 152 |
-
examples=["Who wrote the constitution?","What is the capital of France?"],
|
| 153 |
-
cache_examples=True,
|
| 154 |
-
retry_btn=None,
|
| 155 |
-
undo_btn="Delete Previous",
|
| 156 |
-
clear_btn="Clear",
|
| 157 |
-
css=css
|
| 158 |
-
).launch()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|