Update app.py
Browse files
app.py
CHANGED
|
@@ -1,26 +1,24 @@
|
|
| 1 |
import gradio as gr
|
| 2 |
from huggingface_hub import InferenceClient
|
| 3 |
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
history,
|
| 7 |
-
system_message,
|
| 8 |
-
max_tokens,
|
| 9 |
-
temperature,
|
| 10 |
-
top_p,
|
| 11 |
-
hf_token
|
| 12 |
-
):
|
| 13 |
"""
|
| 14 |
-
Handles streaming responses from Hugging Face Inference API
|
|
|
|
| 15 |
"""
|
| 16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
messages = [{"role": "system", "content": system_message}]
|
| 19 |
messages.extend(history)
|
| 20 |
messages.append({"role": "user", "content": message})
|
| 21 |
|
| 22 |
response = ""
|
| 23 |
-
|
| 24 |
for message_chunk in client.chat_completion(
|
| 25 |
messages,
|
| 26 |
max_tokens=max_tokens,
|
|
@@ -32,15 +30,14 @@ def respond(
|
|
| 32 |
token = ""
|
| 33 |
if len(choices) and choices[0].delta.content:
|
| 34 |
token = choices[0].delta.content
|
| 35 |
-
|
| 36 |
response += token
|
| 37 |
yield response
|
| 38 |
|
| 39 |
-
# Gradio
|
| 40 |
with gr.Blocks() as demo:
|
| 41 |
with gr.Sidebar():
|
| 42 |
-
|
| 43 |
-
gr.
|
| 44 |
system_message = gr.Textbox(
|
| 45 |
value="You are a friendly Chatbot.",
|
| 46 |
label="System message"
|
|
@@ -55,14 +52,14 @@ with gr.Blocks() as demo:
|
|
| 55 |
minimum=0.1, maximum=1.0, value=0.95, step=0.05,
|
| 56 |
label="Top-p (nucleus sampling)"
|
| 57 |
)
|
| 58 |
-
|
| 59 |
chatbot = gr.Chatbot()
|
| 60 |
-
state = gr.State([]) #
|
| 61 |
msg = gr.Textbox(label="Your message")
|
| 62 |
|
| 63 |
msg.submit(
|
| 64 |
respond,
|
| 65 |
-
inputs=[msg, state, system_message, max_tokens, temperature, top_p,
|
| 66 |
outputs=[chatbot],
|
| 67 |
show_progress=True
|
| 68 |
)
|
|
|
|
| 1 |
import gradio as gr
|
| 2 |
from huggingface_hub import InferenceClient
|
| 3 |
|
| 4 |
+
# --- Response function ---
|
| 5 |
+
def respond(message, history, system_message, max_tokens, temperature, top_p, hf_token_info):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
"""
|
| 7 |
+
Handles streaming responses from Hugging Face Inference API.
|
| 8 |
+
`hf_token_info` comes from LoginButton: hf_token_info['access_token']
|
| 9 |
"""
|
| 10 |
+
if not hf_token_info or "access_token" not in hf_token_info:
|
| 11 |
+
yield "Error: You must log in with your Hugging Face token."
|
| 12 |
+
return
|
| 13 |
+
|
| 14 |
+
hf_token = hf_token_info["access_token"]
|
| 15 |
+
client = InferenceClient(token=hf_token, model="openai/gpt-oss-20b")
|
| 16 |
|
| 17 |
messages = [{"role": "system", "content": system_message}]
|
| 18 |
messages.extend(history)
|
| 19 |
messages.append({"role": "user", "content": message})
|
| 20 |
|
| 21 |
response = ""
|
|
|
|
| 22 |
for message_chunk in client.chat_completion(
|
| 23 |
messages,
|
| 24 |
max_tokens=max_tokens,
|
|
|
|
| 30 |
token = ""
|
| 31 |
if len(choices) and choices[0].delta.content:
|
| 32 |
token = choices[0].delta.content
|
|
|
|
| 33 |
response += token
|
| 34 |
yield response
|
| 35 |
|
| 36 |
+
# --- Gradio UI ---
|
| 37 |
with gr.Blocks() as demo:
|
| 38 |
with gr.Sidebar():
|
| 39 |
+
gr.Markdown("## Hugging Face Login & Chatbot Settings")
|
| 40 |
+
login_btn = gr.LoginButton("Log in with Hugging Face", provider="huggingface")
|
| 41 |
system_message = gr.Textbox(
|
| 42 |
value="You are a friendly Chatbot.",
|
| 43 |
label="System message"
|
|
|
|
| 52 |
minimum=0.1, maximum=1.0, value=0.95, step=0.05,
|
| 53 |
label="Top-p (nucleus sampling)"
|
| 54 |
)
|
| 55 |
+
|
| 56 |
chatbot = gr.Chatbot()
|
| 57 |
+
state = gr.State([]) # conversation history
|
| 58 |
msg = gr.Textbox(label="Your message")
|
| 59 |
|
| 60 |
msg.submit(
|
| 61 |
respond,
|
| 62 |
+
inputs=[msg, state, system_message, max_tokens, temperature, top_p, login_btn],
|
| 63 |
outputs=[chatbot],
|
| 64 |
show_progress=True
|
| 65 |
)
|