Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
update to use client
Browse files- app_allenai.py +78 -9
app_allenai.py
CHANGED
@@ -1,13 +1,82 @@
|
|
|
|
1 |
import gradio as gr
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
|
12 |
|
13 |
|
|
|
1 |
+
from gradio_client import Client
|
2 |
import gradio as gr
|
3 |
+
|
4 |
+
MODELS = {
|
5 |
+
"Olmo": "akhaliq/olmo-anychat",
|
6 |
+
"Allen Test": "akhaliq/allen-test"
|
7 |
+
}
|
8 |
+
|
9 |
+
def create_chat_fn(client):
|
10 |
+
def chat(message, history):
|
11 |
+
response = client.predict(
|
12 |
+
message=message,
|
13 |
+
system_prompt="You are a helpful AI assistant.",
|
14 |
+
temperature=0.7,
|
15 |
+
max_new_tokens=1024,
|
16 |
+
top_k=40,
|
17 |
+
repetition_penalty=1.1,
|
18 |
+
top_p=0.95,
|
19 |
+
api_name="/chat"
|
20 |
+
)
|
21 |
+
return response
|
22 |
+
return chat
|
23 |
+
|
24 |
+
def set_client_for_session(model_name: str, request: gr.Request):
|
25 |
+
headers = {}
|
26 |
+
if request and hasattr(request, 'request') and hasattr(request.request, 'headers'):
|
27 |
+
x_ip_token = request.request.headers.get('x-ip-token')
|
28 |
+
if x_ip_token:
|
29 |
+
headers["X-IP-Token"] = x_ip_token
|
30 |
+
|
31 |
+
return Client(MODELS[model_name], headers=headers)
|
32 |
+
|
33 |
+
def safe_chat_fn(message, history, client):
|
34 |
+
if client is None:
|
35 |
+
return "Error: Client not initialized. Please refresh the page."
|
36 |
+
return create_chat_fn(client)(message, history)
|
37 |
+
|
38 |
+
with gr.Blocks() as interface:
|
39 |
+
gr.Markdown("# AI Chat Interface")
|
40 |
+
|
41 |
+
client = gr.State()
|
42 |
+
|
43 |
+
model_dropdown = gr.Dropdown(
|
44 |
+
choices=list(MODELS.keys()),
|
45 |
+
value="Olmo",
|
46 |
+
label="Select Model",
|
47 |
+
interactive=True
|
48 |
+
)
|
49 |
+
|
50 |
+
chat_interface = gr.ChatInterface(
|
51 |
+
fn=safe_chat_fn,
|
52 |
+
title="",
|
53 |
+
description="Chat with AI",
|
54 |
+
examples=[
|
55 |
+
["Hello! How are you?", None],
|
56 |
+
["What is machine learning?", None]
|
57 |
+
],
|
58 |
+
cache_examples=False,
|
59 |
+
additional_inputs=[client]
|
60 |
+
)
|
61 |
+
|
62 |
+
# Update client when model changes
|
63 |
+
def update_model(model_name, request):
|
64 |
+
return set_client_for_session(model_name, request)
|
65 |
+
|
66 |
+
model_dropdown.change(
|
67 |
+
fn=update_model,
|
68 |
+
inputs=[model_dropdown],
|
69 |
+
outputs=[client],
|
70 |
+
)
|
71 |
+
|
72 |
+
# Initialize client on page load
|
73 |
+
interface.load(
|
74 |
+
fn=set_client_for_session,
|
75 |
+
inputs=gr.State("Olmo"), # Default model
|
76 |
+
outputs=client,
|
77 |
+
)
|
78 |
+
|
79 |
+
interface.launch(show_error=True)
|
80 |
|
81 |
|
82 |
|