Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,49 +1,24 @@
|
|
| 1 |
import re
|
| 2 |
import base64
|
| 3 |
-
|
|
|
|
| 4 |
import gradio as gr
|
| 5 |
from PIL import Image
|
| 6 |
-
import
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
model =
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
""
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
Parameters:
|
| 17 |
-
- image_file: Image file uploaded via Gradio
|
| 18 |
-
|
| 19 |
-
Returns:
|
| 20 |
-
- encoded_image (str): Base64 encoded string of the image data
|
| 21 |
-
"""
|
| 22 |
-
if image_file is not None:
|
| 23 |
-
# Convert the PIL Image object to bytes and encode in Base64
|
| 24 |
-
buffered = io.BytesIO()
|
| 25 |
-
image_file.save(buffered, format="JPEG")
|
| 26 |
-
img_bytes = buffered.getvalue()
|
| 27 |
-
encoded_image = base64.b64encode(img_bytes).decode("utf-8")
|
| 28 |
-
return encoded_image
|
| 29 |
-
else:
|
| 30 |
-
raise FileNotFoundError("No file uploaded")
|
| 31 |
-
|
| 32 |
-
def format_response(response_text):
|
| 33 |
-
"""
|
| 34 |
-
Formats the model response to display each item as HTML elements.
|
| 35 |
-
"""
|
| 36 |
-
response_text = re.sub(r"\*\*(.*?)\*\*", r"<p><strong>\1</strong></p>", response_text)
|
| 37 |
-
response_text = re.sub(r"(?m)^\s*\*\s(.*)", r"<li>\1</li>", response_text)
|
| 38 |
-
response_text = re.sub(r"(<li>.*?</li>)+", lambda match: f"<ul>{match.group(0)}</ul>", response_text, flags=re.DOTALL)
|
| 39 |
-
response_text = re.sub(r"</p>(?=<p>)", r"</p><br>", response_text)
|
| 40 |
-
response_text = re.sub(r"(\n|\\n)+", r"<br>", response_text)
|
| 41 |
-
return response_text
|
| 42 |
|
| 43 |
def generate_model_response(image_file, user_query):
|
| 44 |
"""
|
| 45 |
Processes the uploaded image and user query to generate a response from the model.
|
| 46 |
-
|
| 47 |
Parameters:
|
| 48 |
- image_file: The uploaded image file.
|
| 49 |
- user_query: The user's question about the image.
|
|
@@ -52,32 +27,34 @@ def generate_model_response(image_file, user_query):
|
|
| 52 |
- str: The generated response from the model, formatted as HTML.
|
| 53 |
"""
|
| 54 |
try:
|
| 55 |
-
|
| 56 |
-
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
|
| 60 |
-
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
|
| 68 |
-
|
| 69 |
-
|
| 70 |
-
|
| 71 |
-
|
| 72 |
-
|
| 73 |
-
|
| 74 |
-
|
| 75 |
-
|
| 76 |
-
try:
|
| 77 |
outputs = model.generate(**inputs)
|
| 78 |
-
|
| 79 |
-
|
| 80 |
-
|
|
|
|
|
|
|
|
|
|
| 81 |
except Exception as e:
|
| 82 |
print(f"Error in generating response: {e}")
|
| 83 |
return f"<p>An error occurred: {str(e)}</p>"
|
|
@@ -86,10 +63,10 @@ def generate_model_response(image_file, user_query):
|
|
| 86 |
iface = gr.Interface(
|
| 87 |
fn=generate_model_response,
|
| 88 |
inputs=[
|
| 89 |
-
gr.Image(type="
|
| 90 |
gr.Textbox(label="Enter your question", placeholder="How many calories are in this food?")
|
| 91 |
],
|
| 92 |
-
outputs=gr.HTML(label="
|
| 93 |
)
|
| 94 |
|
| 95 |
-
iface.launch(share=True)
|
|
|
|
| 1 |
import re
|
| 2 |
import base64
|
| 3 |
+
import io
|
| 4 |
+
import torch
|
| 5 |
import gradio as gr
|
| 6 |
from PIL import Image
|
| 7 |
+
from transformers import MllamaForConditionalGeneration, AutoProcessor
|
| 8 |
+
|
| 9 |
+
# Load the model and processor
|
| 10 |
+
model_id = "meta-llama/Llama-3.2-11B-Vision-Instruct"
|
| 11 |
+
model = MllamaForConditionalGeneration.from_pretrained(
|
| 12 |
+
model_id,
|
| 13 |
+
torch_dtype=torch.bfloat16,
|
| 14 |
+
device_map="auto",
|
| 15 |
+
)
|
| 16 |
+
processor = AutoProcessor.from_pretrained(model_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
def generate_model_response(image_file, user_query):
|
| 19 |
"""
|
| 20 |
Processes the uploaded image and user query to generate a response from the model.
|
| 21 |
+
|
| 22 |
Parameters:
|
| 23 |
- image_file: The uploaded image file.
|
| 24 |
- user_query: The user's question about the image.
|
|
|
|
| 27 |
- str: The generated response from the model, formatted as HTML.
|
| 28 |
"""
|
| 29 |
try:
|
| 30 |
+
# Load and prepare the image
|
| 31 |
+
raw_image = Image.open(image_file).convert("RGB")
|
| 32 |
+
|
| 33 |
+
# Prepare input for the model using the processor
|
| 34 |
+
conversation = [
|
| 35 |
+
{
|
| 36 |
+
"role": "user",
|
| 37 |
+
"content": [
|
| 38 |
+
{"type": "image", "url": "<|image|>"}, # Placeholder for image
|
| 39 |
+
{"type": "text", "text": user_query}
|
| 40 |
+
]
|
| 41 |
+
}
|
| 42 |
+
]
|
| 43 |
+
|
| 44 |
+
# Apply chat template to prepare inputs for the model
|
| 45 |
+
inputs = processor.apply_chat_template(conversation, add_generation_prompt=True, tokenize=False)
|
| 46 |
+
|
| 47 |
+
# Process the image and text inputs together
|
| 48 |
+
inputs = processor(inputs, raw_image, return_tensors="pt").to(model.device)
|
| 49 |
+
|
| 50 |
+
# Generate response from the model
|
|
|
|
| 51 |
outputs = model.generate(**inputs)
|
| 52 |
+
|
| 53 |
+
# Decode and format the response
|
| 54 |
+
generated_text = processor.decode(outputs[0], skip_special_tokens=True)
|
| 55 |
+
|
| 56 |
+
return generated_text
|
| 57 |
+
|
| 58 |
except Exception as e:
|
| 59 |
print(f"Error in generating response: {e}")
|
| 60 |
return f"<p>An error occurred: {str(e)}</p>"
|
|
|
|
| 63 |
iface = gr.Interface(
|
| 64 |
fn=generate_model_response,
|
| 65 |
inputs=[
|
| 66 |
+
gr.Image(type="file", label="Upload Image"),
|
| 67 |
gr.Textbox(label="Enter your question", placeholder="How many calories are in this food?")
|
| 68 |
],
|
| 69 |
+
outputs=gr.HTML(label="Response from Model"),
|
| 70 |
)
|
| 71 |
|
| 72 |
+
iface.launch(share=True)
|