consy's picture
more parameters for chat_completion
1e125e4 verified
import gradio as gr
import random
#import lines go at the top!
from huggingface_hub import InferenceClient
client = InferenceClient("microsoft/phi-4")
# name of llm chatbot accessed ^^ or can use ' microsoft/phi-4 that's connected to the microsoft phi gen model
def respond(message,history):
messages = [{'role': 'system','content':'You are a friendly chatbot.'}]
#creating dictionary containing key value pairs; first key value pair is system message, second is user message
if history:
messages.extend(history)
messages.append({'role': 'user','content': message})
response = client.chat_completion(messages, max_tokens = 500, top_p=0.8)
#max tokens is a parameter to determine how long the message should be
return response['choices'][0]['message']['content'].strip()
#def yes_or_no(message,history):
# return random.choice(['Yes','No'])
#def echo(message, history):
#always need two inputs
# return message
#print("Hello, World")
chatbot = gr.ChatInterface(respond, type='messages')
#defining my chatbot so user can interact, see their conversation and send new messages
chatbot.launch()