Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,67 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from llama_cpp import Llama
|
3 |
+
from huggingface_hub import hf_hub_download
|
4 |
+
|
5 |
+
# Download model from Hugging Face (if not already present)
|
6 |
+
model_file_path = hf_hub_download(
|
7 |
+
repo_id="TheBloke/Llama-2-7B-GGUF",
|
8 |
+
filename="llama-2-7b.Q4_0.gguf"
|
9 |
+
)
|
10 |
+
|
11 |
+
# Try to load the model from the downloaded file
|
12 |
+
try:
|
13 |
+
llm_llama_cpp = Llama(
|
14 |
+
model_path=model_file_path, # Path where the model is downloaded
|
15 |
+
verbose=False # Suppress llama.cpp's own informational prints
|
16 |
+
)
|
17 |
+
|
18 |
+
# Function that generates a response using the Llama model
|
19 |
+
def talk(prompt, history):
|
20 |
+
try:
|
21 |
+
# Generate text with streaming
|
22 |
+
response_stream = llm_llama_cpp.create_completion(
|
23 |
+
prompt,
|
24 |
+
max_tokens=200, # You can adjust the max tokens as needed
|
25 |
+
stream=True
|
26 |
+
)
|
27 |
+
|
28 |
+
response = ""
|
29 |
+
for chunk in response_stream:
|
30 |
+
# Extract the text from the current chunk
|
31 |
+
if 'choices' in chunk and len(chunk['choices']) > 0 and 'text' in chunk['choices'][0]:
|
32 |
+
response += chunk['choices'][0]['text']
|
33 |
+
|
34 |
+
# Return the full response after streaming is done
|
35 |
+
return response
|
36 |
+
|
37 |
+
except Exception as e:
|
38 |
+
return f"Error with llama-cpp-python: {e}"
|
39 |
+
|
40 |
+
except FileNotFoundError:
|
41 |
+
print(f"Error: Model file not found at {model_file_path}")
|
42 |
+
except Exception as e:
|
43 |
+
print(f"Error with llama-cpp-python: {e}")
|
44 |
+
|
45 |
+
# Gradio interface setup
|
46 |
+
TITLE = "AI Copilot for Diabetes Patients"
|
47 |
+
DESCRIPTION = "I provide answers to concerns related to Diabetes"
|
48 |
+
|
49 |
+
# Design chatbot interface
|
50 |
+
demo = gr.ChatInterface(
|
51 |
+
fn=talk, # The function that processes user input and returns the response
|
52 |
+
chatbot=gr.Chatbot(
|
53 |
+
show_label=True,
|
54 |
+
show_share_button=True,
|
55 |
+
show_copy_button=True,
|
56 |
+
likeable=True,
|
57 |
+
layout="bubble", # Display messages in bubble format
|
58 |
+
bubble_full_width=False,
|
59 |
+
),
|
60 |
+
theme="Soft", # Soft theme for the UI
|
61 |
+
examples=[["what is Diabetes?"]], # Example query to get started
|
62 |
+
title=TITLE, # Title of the interface
|
63 |
+
description=DESCRIPTION, # Description for context
|
64 |
+
)
|
65 |
+
|
66 |
+
# Launch the chatbot interface
|
67 |
+
demo.launch()
|