Spaces:
Sleeping
Sleeping
cindyangelira
commited on
Commit
·
a77ebb6
1
Parent(s):
ca4207c
Update space
Browse files
app.py
CHANGED
@@ -1,9 +1,6 @@
|
|
1 |
import gradio as gr
|
2 |
from huggingface_hub import InferenceClient
|
3 |
|
4 |
-
"""
|
5 |
-
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
|
6 |
-
"""
|
7 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
8 |
|
9 |
|
@@ -15,31 +12,35 @@ def respond(
|
|
15 |
temperature,
|
16 |
top_p,
|
17 |
):
|
|
|
18 |
messages = [{"role": "system", "content": system_message}]
|
19 |
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
|
|
25 |
|
|
|
26 |
messages.append({"role": "user", "content": message})
|
27 |
|
|
|
28 |
response = ""
|
29 |
-
|
30 |
-
for message in client.chat_completion(
|
31 |
messages,
|
32 |
max_tokens=max_tokens,
|
33 |
stream=True,
|
34 |
temperature=temperature,
|
35 |
top_p=top_p,
|
36 |
-
)
|
37 |
-
token = message.choices[0].delta.content
|
38 |
|
|
|
|
|
39 |
response += token
|
40 |
yield response
|
41 |
|
42 |
-
# Define the Gradio interface with
|
43 |
def create_demo():
|
44 |
with gr.Blocks() as demo:
|
45 |
# Add logo at the top
|
@@ -50,16 +51,10 @@ def create_demo():
|
|
50 |
gr.ChatInterface(
|
51 |
respond,
|
52 |
additional_inputs=[
|
53 |
-
gr.Textbox(value="You are a Singaporean Auntie Chatbot. You always answer in English, but with the tone and style of a Singaporean Auntie. Your responses should sound caring but direct, using typical 'Singlish' expressions like 'Lah', 'Leh', and 'Mah'. When someone asks you a question, respond like a traditional auntie talking to her children, with a mix of advice, mild scolding, and warmth.
|
54 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
55 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
56 |
-
gr.Slider(
|
57 |
-
minimum=0.1,
|
58 |
-
maximum=1.0,
|
59 |
-
value=0.95,
|
60 |
-
step=0.05,
|
61 |
-
label="Top-p (nucleus sampling)",
|
62 |
-
),
|
63 |
],
|
64 |
)
|
65 |
|
@@ -67,4 +62,4 @@ def create_demo():
|
|
67 |
|
68 |
if __name__ == "__main__":
|
69 |
demo = create_demo()
|
70 |
-
demo.launch()
|
|
|
1 |
import gradio as gr
|
2 |
from huggingface_hub import InferenceClient
|
3 |
|
|
|
|
|
|
|
4 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
|
5 |
|
6 |
|
|
|
12 |
temperature,
|
13 |
top_p,
|
14 |
):
|
15 |
+
# Initialize messages with the system message hidden from user input
|
16 |
messages = [{"role": "system", "content": system_message}]
|
17 |
|
18 |
+
# Append previous conversation history
|
19 |
+
for user_msg, assistant_msg in history:
|
20 |
+
if user_msg:
|
21 |
+
messages.append({"role": "user", "content": user_msg})
|
22 |
+
if assistant_msg:
|
23 |
+
messages.append({"role": "assistant", "content": assistant_msg})
|
24 |
|
25 |
+
# Append the latest user message
|
26 |
messages.append({"role": "user", "content": message})
|
27 |
|
28 |
+
# Get the response in a single call, avoiding multiple replies for the same input
|
29 |
response = ""
|
30 |
+
message_stream = client.chat_completion(
|
|
|
31 |
messages,
|
32 |
max_tokens=max_tokens,
|
33 |
stream=True,
|
34 |
temperature=temperature,
|
35 |
top_p=top_p,
|
36 |
+
)
|
|
|
37 |
|
38 |
+
for message in message_stream:
|
39 |
+
token = message.choices[0].delta.content
|
40 |
response += token
|
41 |
yield response
|
42 |
|
43 |
+
# Define the Gradio interface with a clean setup
|
44 |
def create_demo():
|
45 |
with gr.Blocks() as demo:
|
46 |
# Add logo at the top
|
|
|
51 |
gr.ChatInterface(
|
52 |
respond,
|
53 |
additional_inputs=[
|
54 |
+
gr.Textbox(value="You are a Singaporean Auntie Chatbot. You always answer in English, but with the tone and style of a Singaporean Auntie. Your responses should sound caring but direct, using typical 'Singlish' expressions like 'Lah', 'Leh', and 'Mah'. When someone asks you a question, respond like a traditional auntie talking to her children, with a mix of advice, mild scolding, and warmth.", label="System message", visible=False),
|
55 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
56 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
57 |
+
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
],
|
59 |
)
|
60 |
|
|
|
62 |
|
63 |
if __name__ == "__main__":
|
64 |
demo = create_demo()
|
65 |
+
demo.launch()
|