File size: 3,167 Bytes
2e9166f
 
 
 
 
 
 
 
 
4e3e56f
284e795
7954b88
284e795
2e9166f
2593eec
c87e284
2e9166f
 
 
 
 
 
 
 
 
 
 
 
0551f07
2e9166f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ec2f726
10c0de8
2e9166f
10c0de8
2e9166f
c87e284
 
2e9166f
 
 
c87e284
e3213ee
10c0de8
af24468
e3213ee
2e9166f
 
c87e284
c39746a
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
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
from langchain import HuggingFaceHub
from langchain.llms.base import LLM
from langchain.memory import ConversationBufferMemory,ConversationBufferWindowMemory
from langchain.chains import LLMChain, ConversationChain
from langchain_community.llms.huggingface_pipeline import HuggingFacePipeline
from langchain_community.llms import HuggingFaceEndpoint
from langchain.prompts import PromptTemplate, ChatPromptTemplate
from langchain_core.prompts.chat import SystemMessagePromptTemplate, HumanMessagePromptTemplate
import os
import gradio as gr
import spaces

your_endpoint_url = "https://kp4xdy196cw81uf3.us-east-1.aws.endpoints.huggingface.cloud"
token = os.environ["API_TOKEN"]

llm = HuggingFaceEndpoint(
    endpoint_url=f"{your_endpoint_url}",
    huggingfacehub_api_token = f"{token}",
    task = "text-generation",
    max_new_tokens=128,
    top_k=10,
    top_p=0.95,
    typical_p=0.95,
    temperature=0.01,
    repetition_penalty=1.03
)
#print(llm)

def chat_template_prompt():
    template = """
    Do not repeat questions and do not generate answer for user/human.
    
    You are a helpful hotel booking asssitant.
    Below is an instruction that describes a task.
    Write a response that appropriately completes the request.
    Reply with the most helpful and logic answer. During the conversation you need to ask the user
    the following questions to complete the hotel booking task.
    1) Where would you like to stay and when?
    2) How many people are staying in the room?
    3) Do you prefer any ammenities like breakfast included or gym?
    4) What is your name, your email address and phone number?
    
    When the booking task is completed, respond with "Thank you for choosing us.".
    
    {history}
    
    """
    
    system_prompt = SystemMessagePromptTemplate.from_template(template)
    human_prompt = HumanMessagePromptTemplate.from_template("{input}")
    chat_prompt = ChatPromptTemplate.from_messages([system_prompt, human_prompt])
    return chat_prompt

def chain():
    #memory = ConversationBufferMemory(memory_key="history")
    chat_prompt = chat_template_prompt()
    memory = ConversationBufferWindowMemory(k=3) #memory_key="history"
    llm_chain = LLMChain(llm=llm, memory = memory, prompt = chat_prompt)
    memory.load_memory_variables({}) #Initialize memory
    return llm_chain

@spaces.GPU
def chat_output(message, history):
    llm_chaim = chain()
    result = llm_chaim.predict(input = message)
    return result

with gr.Blocks() as demo:
    
    chatbot_component = gr.Chatbot(height=300, label = "history")
    textbox_component = gr.Textbox(placeholder="Can I help you to book a hotel?", container=False, label = "input", scale=7)

    demo.chatbot_interface = gr.ChatInterface(
        fn=chat_output,
        examples = ["Hello I would like to book a hotel room.", "Hello I want to stay in Nuremberg in 30th of May." ],
        #outputs=chatbot_component,
        title = "Hotel Booking Assistant Chat 🤗",
        description = "I am your hotel booking assistant. Feel free to start chatting with me."
    )
demo.launch()