File size: 2,706 Bytes
e8bac0f
 
 
 
 
 
 
 
 
 
 
 
 
bcac619
83746e4
 
 
6ab04f4
83746e4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
e8bcde6
83746e4
 
 
6bda5d8
bcac619
83746e4
7826a10
bcac619
 
7826a10
 
 
 
bcac619
 
900fa0f
 
7826a10
900fa0f
 
 
 
93d46ac
a6549b1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
83746e4
 
a6549b1
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
import gradio as gr
import aiohttp
import os
import json
from collections import deque

TOKEN = os.getenv("HUGGINGFACE_API_TOKEN")

if not TOKEN:
    raise ValueError("API token is not set. Please set the HUGGINGFACE_API_TOKEN environment variable.")

memory = deque(maxlen=10)

async def respond(
    message,
    history: list[tuple[str, str]],
    system_message="AI Assistant Role",
    max_tokens=512,
    temperature=0.7,
    top_p=0.95,
):
    system_prefix = "System: ์ž…๋ ฅ์–ด์˜ ์–ธ์–ด(์˜์–ด, ํ•œ๊ตญ์–ด, ์ค‘๊ตญ์–ด, ์ผ๋ณธ์–ด ๋“ฑ)์— ๋”ฐ๋ผ ๋™์ผํ•œ ์–ธ์–ด๋กœ ๋‹ต๋ณ€ํ•˜๋ผ."
    full_system_message = f"{system_prefix}{system_message}"

    memory.append((message, None))
    messages = [{"role": "system", "content": full_system_message}]
    for val in memory:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    headers = {
        "Authorization": f"Bearer {TOKEN}",
        "Content-Type": "application/json"
    }
    payload = {
        "model": "mistralai/Mistral-Nemo-Instruct-2407",
        "max_tokens": max_tokens,
        "temperature": temperature,
        "top_p": top_p,
        "messages": messages,
        "stream": True
    }

    async with aiohttp.ClientSession() as session:
        async with session.post("https://api-inference.huggingface.co/v1/chat/completions", headers=headers, json=payload) as response:
            try:
                async for chunk in response.content:
                    if chunk:
                        chunk_data = chunk.decode('utf-8')
                        response_json = json.loads(chunk_data)
                        if "choices" in response_json:
                            content = response_json["choices"][0]["message"]["content"]
                            yield content
            except json.JSONDecodeError:
                pass
            except StopAsyncIteration:
                pass
            finally:
                pass  # ์ŠคํŠธ๋ฆผ ์ข…๋ฃŒ ์‹œ ์•„๋ฌด๊ฒƒ๋„ ํ•˜์ง€ ์•Š์Œ
                
theme = "Nymbo/Nymbo_Theme"

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

demo = gr.ChatInterface(
    css=css,
    fn=respond,
    theme=theme,
    additional_inputs=[
        gr.Textbox(value="AI Assistant Role", label="System message"),
        gr.Slider(minimum=1, maximum=2048, 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)"),
    ]
)

if __name__ == "__main__":
    demo.queue().launch(max_threads=20)