Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -12,54 +12,42 @@ def respond(
|
|
| 12 |
hf_token: gr.OAuthToken,
|
| 13 |
):
|
| 14 |
"""
|
| 15 |
-
|
|
|
|
| 16 |
"""
|
| 17 |
-
|
|
|
|
| 18 |
|
| 19 |
messages = [{"role": "system", "content": system_message}]
|
| 20 |
-
|
| 21 |
messages.extend(history)
|
| 22 |
-
|
| 23 |
messages.append({"role": "user", "content": message})
|
| 24 |
|
| 25 |
response = ""
|
| 26 |
-
|
| 27 |
-
for message in client.chat_completion(
|
| 28 |
messages,
|
| 29 |
max_tokens=max_tokens,
|
| 30 |
-
stream=True,
|
| 31 |
temperature=temperature,
|
| 32 |
top_p=top_p,
|
|
|
|
| 33 |
):
|
| 34 |
-
choices
|
| 35 |
-
|
| 36 |
-
|
| 37 |
-
|
| 38 |
-
|
| 39 |
-
response += token
|
| 40 |
-
yield response
|
| 41 |
|
| 42 |
|
| 43 |
-
"""
|
| 44 |
-
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
|
| 45 |
-
"""
|
| 46 |
chatbot = gr.ChatInterface(
|
| 47 |
respond,
|
| 48 |
type="messages",
|
| 49 |
additional_inputs=[
|
| 50 |
gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
|
| 51 |
-
gr.Slider(minimum=1, maximum=
|
| 52 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
| 53 |
-
gr.Slider(
|
| 54 |
-
minimum=0.1,
|
| 55 |
-
maximum=1.0,
|
| 56 |
-
value=0.95,
|
| 57 |
-
step=0.05,
|
| 58 |
-
label="Top-p (nucleus sampling)",
|
| 59 |
-
),
|
| 60 |
],
|
| 61 |
)
|
| 62 |
|
|
|
|
| 63 |
with gr.Blocks() as demo:
|
| 64 |
with gr.Sidebar():
|
| 65 |
gr.LoginButton()
|
|
@@ -67,4 +55,4 @@ with gr.Blocks() as demo:
|
|
| 67 |
|
| 68 |
|
| 69 |
if __name__ == "__main__":
|
| 70 |
-
demo.launch()
|
|
|
|
| 12 |
hf_token: gr.OAuthToken,
|
| 13 |
):
|
| 14 |
"""
|
| 15 |
+
Chat with a base LLM hosted on Hugging Face Hub.
|
| 16 |
+
Uses streaming to show tokens as they arrive.
|
| 17 |
"""
|
| 18 |
+
# Replace with a model you have access to, e.g. "meta-llama/Llama-2-7b-chat-hf"
|
| 19 |
+
client = InferenceClient(model="openai/gpt-oss-20b", token=hf_token.token)
|
| 20 |
|
| 21 |
messages = [{"role": "system", "content": system_message}]
|
|
|
|
| 22 |
messages.extend(history)
|
|
|
|
| 23 |
messages.append({"role": "user", "content": message})
|
| 24 |
|
| 25 |
response = ""
|
| 26 |
+
for chunk in client.chat_completion(
|
|
|
|
| 27 |
messages,
|
| 28 |
max_tokens=max_tokens,
|
|
|
|
| 29 |
temperature=temperature,
|
| 30 |
top_p=top_p,
|
| 31 |
+
stream=True,
|
| 32 |
):
|
| 33 |
+
if len(chunk.choices) and chunk.choices[0].delta.content:
|
| 34 |
+
token = chunk.choices[0].delta.content
|
| 35 |
+
response += token
|
| 36 |
+
yield response
|
|
|
|
|
|
|
|
|
|
| 37 |
|
| 38 |
|
|
|
|
|
|
|
|
|
|
| 39 |
chatbot = gr.ChatInterface(
|
| 40 |
respond,
|
| 41 |
type="messages",
|
| 42 |
additional_inputs=[
|
| 43 |
gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
|
| 44 |
+
gr.Slider(minimum=1, maximum=1024, value=256, step=1, label="Max new tokens"),
|
| 45 |
gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
|
| 46 |
+
gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
],
|
| 48 |
)
|
| 49 |
|
| 50 |
+
|
| 51 |
with gr.Blocks() as demo:
|
| 52 |
with gr.Sidebar():
|
| 53 |
gr.LoginButton()
|
|
|
|
| 55 |
|
| 56 |
|
| 57 |
if __name__ == "__main__":
|
| 58 |
+
demo.launch()
|