Spaces:
Sleeping
Sleeping
File size: 5,156 Bytes
4d55d8d e71c4e6 e031d5d e71c4e6 e19713c 0c47d68 4d55d8d e71c4e6 4d55d8d e71c4e6 e19713c e71c4e6 4d55d8d e71c4e6 e19713c e71c4e6 e19713c e71c4e6 e19713c e71c4e6 4d55d8d e71c4e6 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 |
import json
import os
from threading import Lock
from typing import Any, Dict, Optional, Tuple
import gradio as gr
from langchain.chains import ConversationalRetrievalChain
from langchain.chat_models import ChatOpenAI
from langchain.memory import ConversationBufferMemory
from langchain.prompts.chat import (ChatPromptTemplate,
HumanMessagePromptTemplate,
SystemMessagePromptTemplate)
from src.core.chunking import chunk_file
from src.core.embedding import embed_files
from src.core.parsing import read_file
VECTOR_STORE = "faiss"
MODEL = "openai"
EMBEDDING = "openai"
MODEL = "gpt-3.5-turbo-16k"
K = 5
USE_VERBOSE = True
API_KEY = os.environ["OPENAI_API_KEY"]
system_template = """
The context below contains excerpts from 'Croatia,' by Insight Guides. You must only use the information in the context below to formulate your response. If there is not enough information to formulate a response, you must respond with
"I'm sorry, but I can't find the answer to your question in, the book Croatia by Insight Guides."
Begin context:
{context}
End context.
{chat_history}
"""
# Create the chat prompt templates
messages = [
SystemMessagePromptTemplate.from_template(system_template),
HumanMessagePromptTemplate.from_template("{question}")
]
qa_prompt = ChatPromptTemplate.from_messages(messages)
class AnswerConversationBufferMemory(ConversationBufferMemory):
def save_context(self, inputs: Dict[str, Any], outputs: Dict[str, str]) -> None:
return super(AnswerConversationBufferMemory, self).save_context(inputs,{'response': outputs['answer']})
def getretriever():
with open("./resources/RG_Croatia_9ed_FINAL-919-745-7.pdf", 'rb') as uploaded_file:
try:
file = read_file(uploaded_file)
except Exception as e:
print(e)
chunked_file = chunk_file(file, chunk_size=512, chunk_overlap=0)
folder_index = embed_files(
files=[chunked_file],
embedding=EMBEDDING,
vector_store=VECTOR_STORE,
openai_api_key=API_KEY,
)
return folder_index.index.as_retriever(verbose=True, search_type="similarity", search_kwargs={"k": K})
retriever = getretriever()
def predict(message):
print(message)
msgJson = json.loads(message)
print(msgJson)
messages = [
SystemMessagePromptTemplate.from_template(system_template),
HumanMessagePromptTemplate.from_template("{question}")
]
qa_prompt = ChatPromptTemplate.from_messages(messages)
llm = ChatOpenAI(
openai_api_key=API_KEY,
model_name=MODEL,
verbose=True)
memory = AnswerConversationBufferMemory(memory_key="chat_history", return_messages=True)
for msg in msgJson["history"]:
memory.save_context({"input": msg[0]}, {"answer": msg[1]})
chain = ConversationalRetrievalChain.from_llm(
llm,
retriever=retriever,
return_source_documents=USE_VERBOSE,
memory=memory,
verbose=USE_VERBOSE,
combine_docs_chain_kwargs={"prompt": qa_prompt})
chain.rephrase_question = False
lock = Lock()
lock.acquire()
try:
output = chain({"question": msgJson["question"]})
output = output["answer"]
except Exception as e:
print(e)
raise e
finally:
lock.release()
return output
def getanswer(chain, question, history):
if hasattr(chain, "value"):
chain = chain.value
if hasattr(history, "value"):
history = history.value
if hasattr(question, "value"):
question = question.value
history = history or []
lock = Lock()
lock.acquire()
try:
output = chain({"question": question})
output = output["answer"]
history.append((question, output))
except Exception as e:
raise e
finally:
lock.release()
return history, history, gr.update(value="")
def load_chain(inputs = None):
llm = ChatOpenAI(
openai_api_key=API_KEY,
model_name=MODEL,
verbose=True)
chain = ConversationalRetrievalChain.from_llm(
llm,
retriever=retriever,
return_source_documents=USE_VERBOSE,
memory=AnswerConversationBufferMemory(memory_key="chat_history", return_messages=True),
verbose=USE_VERBOSE,
combine_docs_chain_kwargs={"prompt": qa_prompt})
return chain
with gr.Blocks() as block:
with gr.Row():
with gr.Column(scale=0.75):
with gr.Row():
gr.Markdown("<h1>Croatia</h1>")
with gr.Row():
gr.Markdown("by Insight Guides")
chatbot = gr.Chatbot(elem_id="chatbot").style(height=600)
with gr.Row():
message = gr.Textbox(
label="",
placeholder="Ask Croatia...",
lines=1,
)
with gr.Row():
submit = gr.Button(value="Send", variant="primary", scale=1)
state = gr.State()
chain_state = gr.State(load_chain)
submit.click(getanswer, inputs=[chain_state, message, state], outputs=[chatbot, state, message])
message.submit(getanswer, inputs=[chain_state, message, state], outputs=[chatbot, state, message])
with gr.Column(scale=0.25):
predictBtn = gr.Button(value="Predict", visible=False)
predictBtn.click(predict, inputs=[message], outputs=[message])
block.launch(debug=True) |