File size: 2,235 Bytes
e8bac0f
2db0d53
e8bac0f
2db0d53
e8bac0f
8ce49a6
bf799a1
e8bac0f
2db0d53
83746e4
 
2db0d53
 
 
 
83746e4
2db0d53
8ce49a6
85f74eb
2db0d53
 
85f74eb
2db0d53
85f74eb
 
 
 
 
83746e4
85f74eb
2db0d53
cde7a7b
2db0d53
a8032bb
85f74eb
 
 
 
 
 
 
 
 
 
 
 
 
 
a8032bb
 
a6549b1
 
 
 
 
 
 
 
 
2db0d53
a6549b1
2db0d53
8ce49a6
 
2db0d53
a6549b1
2db0d53
 
 
 
 
 
 
 
8ce49a6
 
a6549b1
83746e4
 
2db0d53
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
import gradio as gr
from huggingface_hub import InferenceClient
import os
import requests

# Set up the inference API client
hf_client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.3", token=os.getenv("HF_TOKEN"))

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    system_prefix = """
If the input language is Korean, respond in Korean. If it's English, respond in English.
Do not output in both languages simultaneously. Always respond in Korean to Korean questions and in English to English questions.
    """
    
    messages = [{"role": "system", "content": f"{system_prefix} {system_message}"}]

    # Ensure alternating user/assistant messages
    for user_msg, assistant_msg in history:
        messages.append({"role": "user", "content": user_msg})
        if assistant_msg:  # Only add assistant message if it exists
            messages.append({"role": "assistant", "content": assistant_msg})

    # Add the current user message
    messages.append({"role": "user", "content": message})

    response = ""

    try:
        for message in hf_client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            token = message.choices[0].delta.content
            if token is not None:
                response += token.strip("")
            yield response
    except Exception as e:
        yield f"An error occurred: {str(e)}"


theme = "Nymbo/Nymbo_Theme"

css = """
footer {
    visibility: hidden;
}
"""

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="""
You are an AI assistant.
""", label="System Prompt"),
        gr.Slider(minimum=1, maximum=2000, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
    theme=theme,  # Apply theme
    css=css  # Apply CSS
)

if __name__ == "__main__":
    demo.launch()