Spaces:
Runtime error
Runtime error
Threatthriver
commited on
Commit
•
e7f3548
1
Parent(s):
09fa947
Update app.py
Browse files
app.py
CHANGED
@@ -4,6 +4,16 @@ from huggingface_hub import InferenceClient
|
|
4 |
# Initialize the InferenceClient with the model ID from Hugging Face
|
5 |
client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta")
|
6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
def respond(
|
8 |
message: str,
|
9 |
history: list[tuple[str, str]],
|
@@ -14,7 +24,7 @@ def respond(
|
|
14 |
):
|
15 |
"""
|
16 |
Generates a response from the AI model based on the user's message and chat history.
|
17 |
-
|
18 |
Args:
|
19 |
message (str): The user's input message.
|
20 |
history (list): A list of tuples representing the conversation history (user, assistant).
|
@@ -26,46 +36,48 @@ def respond(
|
|
26 |
Yields:
|
27 |
str: The AI's response as it is generated.
|
28 |
"""
|
29 |
-
|
30 |
# Prepare the conversation history for the API call
|
31 |
messages = [{"role": "system", "content": system_message}]
|
32 |
-
|
33 |
for user_input, assistant_response in history:
|
34 |
-
|
35 |
-
|
36 |
-
if assistant_response:
|
37 |
-
messages.append({"role": "assistant", "content": assistant_response})
|
38 |
|
39 |
# Add the latest user message to the conversation
|
40 |
messages.append({"role": "user", "content": message})
|
41 |
|
42 |
# Initialize an empty response
|
43 |
-
|
44 |
|
45 |
try:
|
46 |
# Generate a response from the model with streaming
|
47 |
-
for
|
48 |
messages=messages,
|
49 |
max_tokens=max_tokens,
|
50 |
stream=True,
|
51 |
temperature=temperature,
|
52 |
top_p=top_p,
|
53 |
):
|
54 |
-
|
55 |
-
|
56 |
-
yield
|
57 |
|
58 |
except Exception as e:
|
59 |
-
yield f"
|
60 |
|
61 |
|
62 |
-
# Define the ChatInterface with additional input components
|
63 |
demo = gr.ChatInterface(
|
64 |
fn=respond,
|
65 |
additional_inputs=[
|
66 |
-
gr.Textbox(
|
|
|
|
|
|
|
|
|
67 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
68 |
-
gr.Slider(minimum=0.1, maximum=
|
69 |
gr.Slider(
|
70 |
minimum=0.1,
|
71 |
maximum=1.0,
|
@@ -74,10 +86,22 @@ demo = gr.ChatInterface(
|
|
74 |
label="Top-p (nucleus sampling)",
|
75 |
),
|
76 |
],
|
77 |
-
title="Chatbot
|
78 |
-
description="A customizable chatbot interface using Hugging Face's Inference API.",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
79 |
)
|
80 |
|
81 |
# Launch the Gradio interface
|
82 |
if __name__ == "__main__":
|
83 |
-
demo.launch()
|
|
|
4 |
# Initialize the InferenceClient with the model ID from Hugging Face
|
5 |
client = InferenceClient(model="HuggingFaceH4/zephyr-7b-beta")
|
6 |
|
7 |
+
# Latest updates (you can replace this with actual update information)
|
8 |
+
latest_updates = """
|
9 |
+
**Zephyr 7B Beta Chatbot - Latest Updates:**
|
10 |
+
|
11 |
+
* **Improved Error Handling:** The chatbot now provides clearer error messages if something goes wrong.
|
12 |
+
* **Enhanced System Message Input:** You can now provide multi-line system messages to guide the AI's behavior.
|
13 |
+
* **Optimized Temperature Range:** The temperature slider's range has been adjusted for better control over randomness.
|
14 |
+
* **Robust Chunk Handling:** The chatbot now handles streamed responses more reliably, even if some chunks are missing content.
|
15 |
+
"""
|
16 |
+
|
17 |
def respond(
|
18 |
message: str,
|
19 |
history: list[tuple[str, str]],
|
|
|
24 |
):
|
25 |
"""
|
26 |
Generates a response from the AI model based on the user's message and chat history.
|
27 |
+
|
28 |
Args:
|
29 |
message (str): The user's input message.
|
30 |
history (list): A list of tuples representing the conversation history (user, assistant).
|
|
|
36 |
Yields:
|
37 |
str: The AI's response as it is generated.
|
38 |
"""
|
39 |
+
|
40 |
# Prepare the conversation history for the API call
|
41 |
messages = [{"role": "system", "content": system_message}]
|
42 |
+
|
43 |
for user_input, assistant_response in history:
|
44 |
+
messages.append({"role": "user", "content": user_input})
|
45 |
+
messages.append({"role": "assistant", "content": assistant_response})
|
|
|
|
|
46 |
|
47 |
# Add the latest user message to the conversation
|
48 |
messages.append({"role": "user", "content": message})
|
49 |
|
50 |
# Initialize an empty response
|
51 |
+
streamed_response = ""
|
52 |
|
53 |
try:
|
54 |
# Generate a response from the model with streaming
|
55 |
+
for response in client.chat_completion(
|
56 |
messages=messages,
|
57 |
max_tokens=max_tokens,
|
58 |
stream=True,
|
59 |
temperature=temperature,
|
60 |
top_p=top_p,
|
61 |
):
|
62 |
+
chunk = response.choices[0].delta.get("content", "")
|
63 |
+
streamed_response += chunk
|
64 |
+
yield streamed_response
|
65 |
|
66 |
except Exception as e:
|
67 |
+
yield f"**Error:** {str(e)}"
|
68 |
|
69 |
|
70 |
+
# Define the ChatInterface with additional input components
|
71 |
demo = gr.ChatInterface(
|
72 |
fn=respond,
|
73 |
additional_inputs=[
|
74 |
+
gr.Textbox(
|
75 |
+
value="You are a friendly and helpful assistant.",
|
76 |
+
label="System message",
|
77 |
+
lines=2
|
78 |
+
),
|
79 |
gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
|
80 |
+
gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.1, label="Temperature"),
|
81 |
gr.Slider(
|
82 |
minimum=0.1,
|
83 |
maximum=1.0,
|
|
|
86 |
label="Top-p (nucleus sampling)",
|
87 |
),
|
88 |
],
|
89 |
+
title="Zephyr 7B Beta Chatbot",
|
90 |
+
description="A customizable chatbot interface using Hugging Face's Zephyr 7B Beta model and Inference API.",
|
91 |
+
)
|
92 |
+
|
93 |
+
# Add the "Show Updates" button and output area
|
94 |
+
with gr.Row():
|
95 |
+
updates_button = gr.Button("Show Latest Updates")
|
96 |
+
updates_output = gr.Markdown(visible=False) # Initially hidden
|
97 |
+
|
98 |
+
# Define the button's click event
|
99 |
+
updates_button.click(
|
100 |
+
fn=lambda: latest_updates,
|
101 |
+
outputs=updates_output,
|
102 |
+
show_progress=False
|
103 |
)
|
104 |
|
105 |
# Launch the Gradio interface
|
106 |
if __name__ == "__main__":
|
107 |
+
demo.launch()
|