Spaces:
Running
Running
starting with existing framework, adding new parameters and modernizing a bit
Browse files
app.py
ADDED
@@ -0,0 +1,86 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from openai import OpenAI
|
3 |
+
import os
|
4 |
+
|
5 |
+
ACCESS_TOKEN = os.getenv("HF_TOKEN")
|
6 |
+
|
7 |
+
print("Access token loaded.")
|
8 |
+
|
9 |
+
client = OpenAI(
|
10 |
+
base_url="https://api-inference.huggingface.co/v1/",
|
11 |
+
api_key=ACCESS_TOKEN,
|
12 |
+
)
|
13 |
+
|
14 |
+
print("OpenAI client initialized.")
|
15 |
+
|
16 |
+
def respond(
|
17 |
+
message,
|
18 |
+
history: list[tuple[str, str]],
|
19 |
+
system_message,
|
20 |
+
max_tokens,
|
21 |
+
temperature,
|
22 |
+
top_p,
|
23 |
+
):
|
24 |
+
print(f"Received message: {message}")
|
25 |
+
print(f"History: {history}")
|
26 |
+
print(f"System message: {system_message}")
|
27 |
+
print(f"Max tokens: {max_tokens}, Temperature: {temperature}, Top-P: {top_p}")
|
28 |
+
|
29 |
+
messages = [{"role": "system", "content": system_message}]
|
30 |
+
|
31 |
+
for val in history:
|
32 |
+
if val[0]:
|
33 |
+
messages.append({"role": "user", "content": val[0]})
|
34 |
+
print(f"Added user message to context: {val[0]}")
|
35 |
+
if val[1]:
|
36 |
+
messages.append({"role": "assistant", "content": val[1]})
|
37 |
+
print(f"Added assistant message to context: {val[1]}")
|
38 |
+
|
39 |
+
messages.append({"role": "user", "content": message})
|
40 |
+
|
41 |
+
response = ""
|
42 |
+
print("Sending request to OpenAI API.")
|
43 |
+
|
44 |
+
for message in client.chat.completions.create(
|
45 |
+
model="meta-llama/Llama-3.3-70B-Instruct",
|
46 |
+
max_tokens=max_tokens,
|
47 |
+
stream=True,
|
48 |
+
temperature=temperature,
|
49 |
+
top_p=top_p,
|
50 |
+
messages=messages,
|
51 |
+
):
|
52 |
+
token = message.choices[0].delta.content
|
53 |
+
print(f"Received token: {token}")
|
54 |
+
response += token
|
55 |
+
yield response
|
56 |
+
|
57 |
+
print("Completed response generation.")
|
58 |
+
|
59 |
+
chatbot = gr.Chatbot(height=600)
|
60 |
+
|
61 |
+
print("Chatbot interface created.")
|
62 |
+
|
63 |
+
demo = gr.ChatInterface(
|
64 |
+
respond,
|
65 |
+
additional_inputs=[
|
66 |
+
gr.Textbox(value="", label="System message"),
|
67 |
+
gr.Slider(minimum=1, maximum=4096, value=512, step=1, label="Max new tokens"),
|
68 |
+
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
69 |
+
gr.Slider(
|
70 |
+
minimum=0.1,
|
71 |
+
maximum=1.0,
|
72 |
+
value=0.95,
|
73 |
+
step=0.05,
|
74 |
+
label="Top-P",
|
75 |
+
),
|
76 |
+
|
77 |
+
],
|
78 |
+
fill_height=True,
|
79 |
+
chatbot=chatbot,
|
80 |
+
theme="Nymbo/Nymbo_Theme",
|
81 |
+
)
|
82 |
+
print("Gradio interface initialized.")
|
83 |
+
|
84 |
+
if __name__ == "__main__":
|
85 |
+
print("Launching the demo application.")
|
86 |
+
demo.launch()
|