Update app.py
Browse files
app.py
CHANGED
@@ -4,13 +4,12 @@ import gradio as gr
|
|
4 |
import tempfile
|
5 |
import secrets
|
6 |
from pathlib import Path
|
7 |
-
from transformers import AutoModelForCausalLM, AutoTokenizer,
|
8 |
-
from qwen_vl_utils import process_vision_info
|
9 |
from PIL import Image
|
10 |
|
11 |
# Load Vision-Language Model
|
12 |
vl_model = Qwen2VLForConditionalGeneration.from_pretrained(
|
13 |
-
"Qwen/Qwen2-VL-2B-Instruct", torch_dtype=
|
14 |
)
|
15 |
vl_processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
16 |
|
@@ -34,18 +33,17 @@ def process_image(image, shouldConvert=False):
|
|
34 |
new_img.paste(image, (0, 0), mask=image)
|
35 |
image = new_img
|
36 |
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
|
45 |
-
|
46 |
-
|
47 |
-
|
48 |
-
return f"Math-related content detected: {description}"
|
49 |
|
50 |
def get_math_response(image_description, user_question):
|
51 |
global math_messages
|
@@ -55,6 +53,7 @@ def get_math_response(image_description, user_question):
|
|
55 |
content = f'Image description: {image_description}\n\n' if image_description else ''
|
56 |
query = f"{content}User question: {user_question}"
|
57 |
math_messages.append({'role': 'user', 'content': query})
|
|
|
58 |
model_inputs = tokenizer(query, return_tensors="pt").to(device)
|
59 |
output = model.generate(**model_inputs, max_new_tokens=512)
|
60 |
answer = tokenizer.decode(output[0], skip_special_tokens=True)
|
@@ -62,14 +61,12 @@ def get_math_response(image_description, user_question):
|
|
62 |
math_messages.append({'role': 'assistant', 'content': answer})
|
63 |
|
64 |
def math_chat_bot(image, sketchpad, question, state):
|
65 |
-
current_tab_index = state
|
66 |
image_description = None
|
67 |
-
if current_tab_index == 0:
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
if sketchpad and sketchpad.get("composite"):
|
72 |
-
image_description = process_image(sketchpad["composite"], True)
|
73 |
yield from get_math_response(image_description, question)
|
74 |
|
75 |
css = """
|
@@ -81,13 +78,12 @@ css = """
|
|
81 |
def tabs_select(e: gr.SelectData, _state):
|
82 |
_state["tab_index"] = e.index
|
83 |
|
84 |
-
# 创建Gradio接口
|
85 |
with gr.Blocks(css=css) as demo:
|
86 |
gr.HTML("""
|
87 |
-
<p align="center"><img src="https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png" style="height: 60px"
|
88 |
-
|
89 |
-
|
90 |
-
|
91 |
state = gr.State({"tab_index": 0})
|
92 |
with gr.Row():
|
93 |
with gr.Column():
|
@@ -97,18 +93,18 @@ with gr.Blocks(css=css) as demo:
|
|
97 |
with gr.Tab("Sketch"):
|
98 |
input_sketchpad = gr.Sketchpad(type="pil", label="Sketch", layers=False)
|
99 |
input_tabs.select(fn=tabs_select, inputs=[state])
|
100 |
-
input_text = gr.Textbox(label="
|
101 |
with gr.Row():
|
102 |
with gr.Column():
|
103 |
clear_btn = gr.ClearButton([input_image, input_sketchpad, input_text])
|
104 |
with gr.Column():
|
105 |
submit_btn = gr.Button("Submit", variant="primary")
|
106 |
with gr.Column():
|
107 |
-
output_md = gr.Markdown(label="
|
108 |
submit_btn.click(
|
109 |
fn=math_chat_bot,
|
110 |
inputs=[input_image, input_sketchpad, input_text, state],
|
111 |
outputs=output_md)
|
112 |
|
113 |
if __name__ == "__main__":
|
114 |
-
demo.launch()
|
|
|
4 |
import tempfile
|
5 |
import secrets
|
6 |
from pathlib import Path
|
7 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer, AutoProcessor, Qwen2VLForConditionalGeneration
|
|
|
8 |
from PIL import Image
|
9 |
|
10 |
# Load Vision-Language Model
|
11 |
vl_model = Qwen2VLForConditionalGeneration.from_pretrained(
|
12 |
+
"Qwen/Qwen2-VL-2B-Instruct", torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32, device_map="auto"
|
13 |
)
|
14 |
vl_processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
15 |
|
|
|
33 |
new_img.paste(image, (0, 0), mask=image)
|
34 |
image = new_img
|
35 |
|
36 |
+
try:
|
37 |
+
inputs = vl_processor(images=image, return_tensors="pt").to(device)
|
38 |
+
if inputs is None:
|
39 |
+
return "Error processing image."
|
40 |
+
|
41 |
+
generated_ids = vl_model.generate(**inputs)
|
42 |
+
output = vl_processor.batch_decode(generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)
|
43 |
+
description = output[0] if output else ""
|
44 |
+
return f"Math-related content detected: {description}"
|
45 |
+
except Exception as e:
|
46 |
+
return f"Error in processing image: {str(e)}"
|
|
|
47 |
|
48 |
def get_math_response(image_description, user_question):
|
49 |
global math_messages
|
|
|
53 |
content = f'Image description: {image_description}\n\n' if image_description else ''
|
54 |
query = f"{content}User question: {user_question}"
|
55 |
math_messages.append({'role': 'user', 'content': query})
|
56 |
+
|
57 |
model_inputs = tokenizer(query, return_tensors="pt").to(device)
|
58 |
output = model.generate(**model_inputs, max_new_tokens=512)
|
59 |
answer = tokenizer.decode(output[0], skip_special_tokens=True)
|
|
|
61 |
math_messages.append({'role': 'assistant', 'content': answer})
|
62 |
|
63 |
def math_chat_bot(image, sketchpad, question, state):
|
64 |
+
current_tab_index = state.get("tab_index", 0)
|
65 |
image_description = None
|
66 |
+
if current_tab_index == 0 and image is not None:
|
67 |
+
image_description = process_image(image)
|
68 |
+
elif current_tab_index == 1 and sketchpad and sketchpad.get("composite"):
|
69 |
+
image_description = process_image(sketchpad["composite"], True)
|
|
|
|
|
70 |
yield from get_math_response(image_description, question)
|
71 |
|
72 |
css = """
|
|
|
78 |
def tabs_select(e: gr.SelectData, _state):
|
79 |
_state["tab_index"] = e.index
|
80 |
|
|
|
81 |
with gr.Blocks(css=css) as demo:
|
82 |
gr.HTML("""
|
83 |
+
<p align="center"><img src="https://modelscope.oss-cn-beijing.aliyuncs.com/resource/qwen.png" style="height: 60px"/></p>
|
84 |
+
<center><font size=8>📖 Qwen2-Math Demo</font></center>
|
85 |
+
<center><font size=3>This WebUI uses Qwen2-VL for OCR and Qwen2-Math for mathematical reasoning.</font></center>
|
86 |
+
""")
|
87 |
state = gr.State({"tab_index": 0})
|
88 |
with gr.Row():
|
89 |
with gr.Column():
|
|
|
93 |
with gr.Tab("Sketch"):
|
94 |
input_sketchpad = gr.Sketchpad(type="pil", label="Sketch", layers=False)
|
95 |
input_tabs.select(fn=tabs_select, inputs=[state])
|
96 |
+
input_text = gr.Textbox(label="Enter your question")
|
97 |
with gr.Row():
|
98 |
with gr.Column():
|
99 |
clear_btn = gr.ClearButton([input_image, input_sketchpad, input_text])
|
100 |
with gr.Column():
|
101 |
submit_btn = gr.Button("Submit", variant="primary")
|
102 |
with gr.Column():
|
103 |
+
output_md = gr.Markdown(label="Answer", elem_id="qwen-md")
|
104 |
submit_btn.click(
|
105 |
fn=math_chat_bot,
|
106 |
inputs=[input_image, input_sketchpad, input_text, state],
|
107 |
outputs=output_md)
|
108 |
|
109 |
if __name__ == "__main__":
|
110 |
+
demo.launch()
|