Spaces:
Sleeping
Sleeping
File size: 5,689 Bytes
b0b485a 8776349 b0b485a 321b65c b0b485a baf43d1 b0b485a 32f7b36 6e2cc63 b0b485a d11a7ca b0b485a 71a59eb b0b485a f435b5e 26ff4af f435b5e 26ff4af f435b5e b0b485a f435b5e b0b485a |
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 |
import streamlit as st
# from langchain_community.llms import HuggingFaceTextGenInference
import os, pickle
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
from langchain.schema import StrOutputParser
from custom_llm import CustomLLM, custom_chain_with_history, custom_combined_chain, custom_dataframe_chain, format_df,custom_unique_df_chain
import pandas as pd
API_TOKEN = os.getenv('HF_INFER_API')
from typing import Optional
from langchain.prompts import ChatPromptTemplate, MessagesPlaceholder
from langchain_community.chat_models import ChatAnthropic
from langchain_core.chat_history import BaseChatMessageHistory
from langchain.memory import ConversationBufferMemory
from langchain_core.runnables.history import RunnableWithMessageHistory
@st.cache_data(persist=False)
def get_df():
return pickle.load(open("ebesha_ticket_df.pkl", "rb"))
@st.cache_resource
def get_llm_chain():
llm = CustomLLM(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", model_type='text-generation', api_token=API_TOKEN, stop=["\n<|","<|"])
dataframe_chain = custom_dataframe_chain(llm=llm, df=st.session_state.df, unique_values=st.session_state.unique_values)
memory_chain = custom_chain_with_history(llm=llm, memory=st.session_state.memory)
return custom_combined_chain(llm=CustomLLM(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", model_type='text-generation', api_token=API_TOKEN, stop=["\n<|","<|"], max_new_tokens=4), df_chain=dataframe_chain, memory_chain=memory_chain)
if 'memory' not in st.session_state:
st.session_state['memory'] = ConversationBufferMemory(return_messages=True)
st.session_state.memory.chat_memory.add_ai_message("Hello there! I'm AI assistant of Lintas Media Danawa. How can I help you today?")
if 'df' not in st.session_state:
st.session_state['df'] = get_df()
if 'unique_values' not in st.session_state:
# exec(custom_unique_df_chain(llm=CustomLLM(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", model_type='text-generation', api_token=API_TOKEN, stop=["\n<|","<|"]), df=st.session_state.df).invoke({"df_example":format_df(st.session_state.df.head(4))}))
# st.session_state.unique_values = response
df = st.session_state.df
st.session_state.unique_values = {
'request_mode': df['request_mode'].unique().tolist(),
'service_category': df['service_category'].unique().tolist(),
'child_service_1': df['child_service_1'].unique().tolist(),
'child_service_2': df['child_service_2'].unique().tolist(),
'child_service_3': df['child_service_3'].unique().tolist(),
'child_service_4': df['child_service_4'].unique().tolist(),
'request_status': df['request_status'].unique().tolist(),
'request_type': df['request_type'].unique().tolist(),
'priority': df['priority'].unique().tolist(),
'fcr': df['fcr'].unique().tolist(),
}
if 'chain' not in st.session_state:
# st.session_state['chain'] = custom_chain_with_history(llm=CustomLLM(repo_id="mistralai/Mixtral-8x7B-Instruct-v0.1", model_type='text-generation', api_token=API_TOKEN, stop=["\n<|","<|"]), memory=st.session_state.memory)
st.session_state['chain'] = get_llm_chain()
# st.session_state['chain'] = custom_chain_with_history(llm=InferenceClient("https://api-inference.huggingface.co/models/mistralai/Mixtral-8x7B-Instruct-v0.1", headers = {"Authorization": f"Bearer {API_TOKEN}"}, stream=True, max_new_tokens=512, temperature=0.01), memory=st.session_state.memory)
st.title("LMD Chatbot V3")
st.subheader("Combination of Ticket Submission and WI/User Guide Knowledge")
# Initialize chat history
if "messages" not in st.session_state:
st.session_state.messages = [{"role":"assistant", "content":"Hello there! I'm AI assistant of Lintas Media Danawa. How can I help you today?"}]
# Display chat messages from history on app rerun
for message in st.session_state.messages:
with st.chat_message(message["role"]):
st.markdown(message["content"])
# React to user input
if prompt := st.chat_input("Ask me anything.."):
# Display user message in chat message container
st.chat_message("User").markdown(prompt)
# Add user message to chat history
st.session_state.messages.append({"role": "User", "content": prompt})
# full_response = st.session_state.chain.invoke(prompt).split("\n<|")[0]
full_response = st.session_state.chain.invoke({"question":prompt, "memory":st.session_state.memory, "df_example":format_df(st.session_state.df.head(4))}).split("\n<|")[0]
print(len(full_response))
try :
df = st.session_state.df
exec(full_response)
full_response = "Here is the python code: \n\n```python"+ full_response +"\n```\n\nGenerated Response: \n\n"+ str(response)
except Exception as e:
print(e)
with st.chat_message("assistant"):
st.markdown(full_response)
# Display assistant response in chat message container
# with st.chat_message("assistant"):
# message_placeholder = st.empty()
# full_response = ""
# for chunk in st.session_state.chain.stream(prompt):
# full_response += chunk + " "
# message_placeholder.markdown(full_response + " ")
# if full_response[-4:] == "\n<|":
# break
# st.markdown(full_response)
st.session_state.memory.save_context({"question":prompt}, {"output":full_response})
st.session_state.memory.chat_memory.messages = st.session_state.memory.chat_memory.messages[-15:]
# Add assistant response to chat history
st.session_state.messages.append({"role": "assistant", "content": full_response}) |