Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -2,41 +2,55 @@ 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:
|
|
|
6 |
"""
|
7 |
client = InferenceClient("unsloth/Llama-3.2-1B-Instruct")
|
8 |
|
9 |
|
10 |
def respond(
|
11 |
message,
|
12 |
-
history: list[tuple[str, str]],
|
13 |
-
# system_message,
|
14 |
-
# max_tokens,
|
15 |
-
# temperature,
|
16 |
-
# top_p,
|
17 |
):
|
18 |
-
|
19 |
-
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
24 |
max_tokens = 512
|
25 |
temperature = 0.7
|
26 |
top_p = 0.95
|
27 |
-
|
|
|
28 |
messages = [{"role": "system", "content": system_message}]
|
29 |
|
|
|
30 |
for val in history:
|
31 |
if val[0]:
|
32 |
messages.append({"role": "user", "content": val[0]})
|
33 |
if val[1]:
|
34 |
messages.append({"role": "assistant", "content": val[1]})
|
35 |
|
|
|
36 |
messages.append({"role": "user", "content": message})
|
37 |
|
38 |
response = ""
|
39 |
|
|
|
40 |
for message in client.chat_completion(
|
41 |
messages,
|
42 |
max_tokens=max_tokens,
|
@@ -45,33 +59,15 @@ def respond(
|
|
45 |
top_p=top_p,
|
46 |
):
|
47 |
token = message.choices[0].delta.content
|
48 |
-
|
49 |
response += token
|
50 |
yield response
|
51 |
|
52 |
|
53 |
-
|
54 |
-
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
|
55 |
-
"""
|
56 |
with gr.Blocks() as demo:
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
# gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
|
61 |
-
# gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
62 |
-
# gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
63 |
-
# gr.Slider(
|
64 |
-
# minimum=0.1,
|
65 |
-
# maximum=1.0,
|
66 |
-
# value=0.95,
|
67 |
-
# step=0.05,
|
68 |
-
# label="Top-p (nucleus sampling)",
|
69 |
-
# ),
|
70 |
-
# ],
|
71 |
-
)
|
72 |
-
# Send a default message when the interface loads
|
73 |
-
chat.send_message("Hi there! I'm your Dietician Assistant, here to help you with general advice on diet, nutrition, and healthy eating habits. Let's explore your questions.")
|
74 |
-
|
75 |
|
76 |
if __name__ == "__main__":
|
77 |
demo.launch()
|
|
|
2 |
from huggingface_hub import InferenceClient
|
3 |
|
4 |
"""
|
5 |
+
For more information on `huggingface_hub` Inference API support, please check the docs:
|
6 |
+
https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
|
7 |
"""
|
8 |
client = InferenceClient("unsloth/Llama-3.2-1B-Instruct")
|
9 |
|
10 |
|
11 |
def respond(
|
12 |
message,
|
13 |
+
history: list[tuple[str, str]] = None, # Default history as None to avoid mutable issues
|
|
|
|
|
|
|
|
|
14 |
):
|
15 |
+
if history is None:
|
16 |
+
history = [
|
17 |
+
(
|
18 |
+
None,
|
19 |
+
"Hi there! I'm your Dietician Assistant, here to help with general advice on diet, "
|
20 |
+
"nutrition, and healthy eating habits. Let's explore your questions.",
|
21 |
+
)
|
22 |
+
]
|
23 |
+
|
24 |
+
# System message describing the assistant's role
|
25 |
+
system_message = (
|
26 |
+
"You are a Dietician Assistant specializing in providing general guidance on diet, "
|
27 |
+
"nutrition, and healthy eating habits. Answer questions thoroughly with scientifically "
|
28 |
+
"backed advice, practical tips, and easy-to-understand explanations. Keep in mind that "
|
29 |
+
"your role is to assist, not replace a registered dietitian, so kindly remind users to "
|
30 |
+
"consult a professional for personalized advice when necessary."
|
31 |
+
)
|
32 |
+
|
33 |
+
# Define model parameters
|
34 |
max_tokens = 512
|
35 |
temperature = 0.7
|
36 |
top_p = 0.95
|
37 |
+
|
38 |
+
# Initialize the message history with the system message
|
39 |
messages = [{"role": "system", "content": system_message}]
|
40 |
|
41 |
+
# Add previous history to the message chain
|
42 |
for val in history:
|
43 |
if val[0]:
|
44 |
messages.append({"role": "user", "content": val[0]})
|
45 |
if val[1]:
|
46 |
messages.append({"role": "assistant", "content": val[1]})
|
47 |
|
48 |
+
# Append the new user message
|
49 |
messages.append({"role": "user", "content": message})
|
50 |
|
51 |
response = ""
|
52 |
|
53 |
+
# Generate the response in a streaming fashion
|
54 |
for message in client.chat_completion(
|
55 |
messages,
|
56 |
max_tokens=max_tokens,
|
|
|
59 |
top_p=top_p,
|
60 |
):
|
61 |
token = message.choices[0].delta.content
|
|
|
62 |
response += token
|
63 |
yield response
|
64 |
|
65 |
|
66 |
+
# Set up the Gradio ChatInterface with an initial history
|
|
|
|
|
67 |
with gr.Blocks() as demo:
|
68 |
+
gr.ChatInterface(respond, value=[("None",
|
69 |
+
"Hi there! I'm your Dietician Assistant, here to help with general advice on diet, "
|
70 |
+
"nutrition, and healthy eating habits. Let's explore your questions.")])
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
|
72 |
if __name__ == "__main__":
|
73 |
demo.launch()
|