Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
@@ -1,10 +1,12 @@
|
|
1 |
-
import
|
2 |
import torch
|
3 |
from transformers import AutoModel, AutoTokenizer, BitsAndBytesConfig
|
4 |
from PIL import Image
|
5 |
-
import
|
|
|
|
|
6 |
|
7 |
-
# Get API token from environment
|
8 |
api_token = os.getenv("HF_TOKEN").strip()
|
9 |
|
10 |
# Quantization configuration
|
@@ -15,7 +17,7 @@ bnb_config = BitsAndBytesConfig(
|
|
15 |
bnb_4bit_compute_dtype=torch.float16
|
16 |
)
|
17 |
|
18 |
-
# Load
|
19 |
model = AutoModel.from_pretrained(
|
20 |
"ContactDoctor/Bio-Medical-MultiModal-Llama-3-8B-V1",
|
21 |
quantization_config=bnb_config,
|
@@ -31,42 +33,54 @@ tokenizer = AutoTokenizer.from_pretrained(
|
|
31 |
token=api_token
|
32 |
)
|
33 |
|
34 |
-
|
35 |
-
def process_query(image, question):
|
36 |
try:
|
37 |
-
if
|
38 |
-
|
39 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
40 |
inputs = model.prepare_inputs_for_generation(
|
41 |
input_ids=tokenizer(question, return_tensors="pt").input_ids,
|
42 |
images=[image]
|
43 |
)
|
44 |
-
outputs = model.generate(**inputs, max_new_tokens=256)
|
45 |
else:
|
46 |
-
# Process text-only
|
47 |
inputs = tokenizer(question, return_tensors="pt")
|
48 |
-
outputs = model.generate(**inputs, max_new_tokens=256)
|
49 |
|
50 |
-
|
51 |
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
52 |
-
|
53 |
-
|
|
|
|
|
|
|
54 |
except Exception as e:
|
55 |
-
return
|
|
|
|
|
|
|
56 |
|
57 |
-
#
|
58 |
-
|
59 |
-
fn=
|
60 |
inputs=[
|
61 |
-
gr.Image(type="
|
62 |
-
gr.Textbox(label="Enter
|
63 |
],
|
64 |
-
outputs="
|
65 |
-
title="
|
66 |
-
description="
|
67 |
-
|
68 |
)
|
69 |
|
70 |
-
# Launch
|
71 |
-
|
72 |
-
interface.launch()
|
|
|
1 |
+
import os
|
2 |
import torch
|
3 |
from transformers import AutoModel, AutoTokenizer, BitsAndBytesConfig
|
4 |
from PIL import Image
|
5 |
+
import gradio as gr
|
6 |
+
import base64
|
7 |
+
import io
|
8 |
|
9 |
+
# Get API token from environment variable
|
10 |
api_token = os.getenv("HF_TOKEN").strip()
|
11 |
|
12 |
# Quantization configuration
|
|
|
17 |
bnb_4bit_compute_dtype=torch.float16
|
18 |
)
|
19 |
|
20 |
+
# Load model
|
21 |
model = AutoModel.from_pretrained(
|
22 |
"ContactDoctor/Bio-Medical-MultiModal-Llama-3-8B-V1",
|
23 |
quantization_config=bnb_config,
|
|
|
33 |
token=api_token
|
34 |
)
|
35 |
|
36 |
+
def analyze_input(image_data=None, question=""):
|
|
|
37 |
try:
|
38 |
+
# Handle base64 image if provided
|
39 |
+
if isinstance(image_data, str) and image_data.startswith('data:image'):
|
40 |
+
# Extract base64 data after the comma
|
41 |
+
base64_data = image_data.split(',')[1]
|
42 |
+
image_bytes = base64.b64decode(base64_data)
|
43 |
+
image = Image.open(io.BytesIO(image_bytes)).convert('RGB')
|
44 |
+
# Handle direct image input
|
45 |
+
elif image_data is not None:
|
46 |
+
image = Image.fromarray(image_data).convert('RGB')
|
47 |
+
else:
|
48 |
+
image = None
|
49 |
+
|
50 |
+
# Process with or without image
|
51 |
+
if image is not None:
|
52 |
inputs = model.prepare_inputs_for_generation(
|
53 |
input_ids=tokenizer(question, return_tensors="pt").input_ids,
|
54 |
images=[image]
|
55 |
)
|
|
|
56 |
else:
|
|
|
57 |
inputs = tokenizer(question, return_tensors="pt")
|
|
|
58 |
|
59 |
+
outputs = model.generate(**inputs, max_new_tokens=256)
|
60 |
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
61 |
+
|
62 |
+
return {
|
63 |
+
"status": "success",
|
64 |
+
"response": response
|
65 |
+
}
|
66 |
except Exception as e:
|
67 |
+
return {
|
68 |
+
"status": "error",
|
69 |
+
"message": str(e)
|
70 |
+
}
|
71 |
|
72 |
+
# Create Gradio interface
|
73 |
+
demo = gr.Interface(
|
74 |
+
fn=analyze_input,
|
75 |
inputs=[
|
76 |
+
gr.Image(type="numpy", label="Medical Image (Optional)", optional=True),
|
77 |
+
gr.Textbox(label="Question", placeholder="Enter your medical query...")
|
78 |
],
|
79 |
+
outputs=gr.JSON(label="Analysis"),
|
80 |
+
title="Bio-Medical MultiModal Analysis",
|
81 |
+
description="Ask questions with or without an image",
|
82 |
+
allow_flagging="never",
|
83 |
)
|
84 |
|
85 |
+
# Launch with API access enabled
|
86 |
+
demo.launch(share=True, server_name="0.0.0.0", server_port=7860, enable_queue=True)
|
|