File size: 1,750 Bytes
a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 7049d4a a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c a2561e3 71fdc5c c2fbb38 |
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 |
import os
import gradio as gr
from openai import OpenAI
# Load API key securely from Hugging Face secrets or environment
api_key = os.getenv("NV_API_KEY")
if not api_key:
raise ValueError("Please set the NV_API_KEY environment variable in your Hugging Face Space.")
# NVIDIA-compatible OpenAI client
client = OpenAI(
base_url="https://integrate.api.nvidia.com/v1",
api_key=api_key
)
# System message
system_prompt = {
"role": "system",
"content": "You are a helpful assistant to answer user queries."
}
# Main chat function with memory from Gradio
def get_text_response(user_message, history):
# Convert Gradio history to OpenAI format
formatted_history = [{"role": "user" if i % 2 == 0 else "assistant", "content": msg}
for i, msg in enumerate(sum(history, []))]
# Combine system prompt, history, and current user input
messages = [system_prompt] + formatted_history + [{"role": "user", "content": user_message}]
# Stream the response
response = ""
completion = client.chat.completions.create(
model="nvidia/llama-3.1-nemotron-70b-instruct",
messages=messages,
temperature=0.5,
top_p=1,
max_tokens=100,
stream=True
)
for chunk in completion:
delta = chunk.choices[0].delta
if delta and delta.content:
response += delta.content
return response
# Gradio Chat Interface
demo = gr.ChatInterface(
fn=get_text_response,
theme="soft",
textbox=gr.Textbox(placeholder="Ask me anything...", container=False),
examples=["How are you doing?", "What are your interests?", "Which places do you like to visit?"]
)
if __name__ == "__main__":
demo.queue().launch(share=True)
|