Spaces:
Running
on
Zero
Running
on
Zero
Update Gradio app with multiple files
Browse files
app.py
CHANGED
|
@@ -16,21 +16,9 @@ print("Model loaded successfully!")
|
|
| 16 |
|
| 17 |
|
| 18 |
@spaces.GPU
|
| 19 |
-
def generate_response(messages):
|
| 20 |
-
"""Generate response using the pipeline."""
|
| 21 |
-
response = pipe(
|
| 22 |
-
messages,
|
| 23 |
-
max_new_tokens=4096,
|
| 24 |
-
do_sample=True,
|
| 25 |
-
temperature=0.6,
|
| 26 |
-
top_p=0.95
|
| 27 |
-
)
|
| 28 |
-
return response[0]["generated_text"][-1]["content"]
|
| 29 |
-
|
| 30 |
-
|
| 31 |
def respond(message, history):
|
| 32 |
"""
|
| 33 |
-
Generate response for the chatbot.
|
| 34 |
|
| 35 |
Args:
|
| 36 |
message: The user's current message
|
|
@@ -45,10 +33,30 @@ def respond(message, history):
|
|
| 45 |
# Add current message
|
| 46 |
messages.append({"role": "user", "content": message})
|
| 47 |
|
| 48 |
-
# Generate response
|
| 49 |
-
|
| 50 |
-
|
| 51 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 52 |
|
| 53 |
|
| 54 |
# Create the Gradio interface
|
|
@@ -74,8 +82,10 @@ with gr.Blocks(
|
|
| 74 |
title="",
|
| 75 |
description="Ask me anything! I'm powered by VibeThinker with ZeroGPU acceleration.",
|
| 76 |
examples=[
|
|
|
|
| 77 |
"Explain quantum computing in simple terms",
|
| 78 |
"Write a short poem about artificial intelligence",
|
|
|
|
| 79 |
],
|
| 80 |
cache_examples=False,
|
| 81 |
)
|
|
|
|
| 16 |
|
| 17 |
|
| 18 |
@spaces.GPU
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 19 |
def respond(message, history):
|
| 20 |
"""
|
| 21 |
+
Generate streaming response for the chatbot.
|
| 22 |
|
| 23 |
Args:
|
| 24 |
message: The user's current message
|
|
|
|
| 33 |
# Add current message
|
| 34 |
messages.append({"role": "user", "content": message})
|
| 35 |
|
| 36 |
+
# Generate response with streaming
|
| 37 |
+
full_response = ""
|
| 38 |
+
for output in pipe(
|
| 39 |
+
messages,
|
| 40 |
+
max_new_tokens=4096,
|
| 41 |
+
do_sample=True,
|
| 42 |
+
temperature=0.6,
|
| 43 |
+
top_p=0.95,
|
| 44 |
+
return_full_text=False,
|
| 45 |
+
streamer=None
|
| 46 |
+
):
|
| 47 |
+
# Get the generated text
|
| 48 |
+
generated_text = output[0]["generated_text"]
|
| 49 |
+
|
| 50 |
+
# Extract only the assistant's response
|
| 51 |
+
if isinstance(generated_text, list):
|
| 52 |
+
assistant_response = generated_text[-1]["content"]
|
| 53 |
+
else:
|
| 54 |
+
assistant_response = generated_text
|
| 55 |
+
|
| 56 |
+
# Stream character by character
|
| 57 |
+
for char in assistant_response[len(full_response):]:
|
| 58 |
+
full_response += char
|
| 59 |
+
yield full_response
|
| 60 |
|
| 61 |
|
| 62 |
# Create the Gradio interface
|
|
|
|
| 82 |
title="",
|
| 83 |
description="Ask me anything! I'm powered by VibeThinker with ZeroGPU acceleration.",
|
| 84 |
examples=[
|
| 85 |
+
"What is the meaning of life?",
|
| 86 |
"Explain quantum computing in simple terms",
|
| 87 |
"Write a short poem about artificial intelligence",
|
| 88 |
+
"How can I improve my productivity?",
|
| 89 |
],
|
| 90 |
cache_examples=False,
|
| 91 |
)
|