Spaces:
Running
on
Zero
Running
on
Zero
Commit
·
26dc4f5
0
Parent(s):
Initial Gradio app for Dental VQA Model Comparison
Browse files- .gitignore +55 -0
- .idea/.gitignore +8 -0
- .idea/dental-vqa-comparison.iml +8 -0
- .idea/inspectionProfiles/profiles_settings.xml +6 -0
- .idea/material_theme_project_new.xml +12 -0
- .idea/modules.xml +8 -0
- .idea/vcs.xml +6 -0
- README.md +54 -0
- app.py +234 -0
- requirements.txt +6 -0
.gitignore
ADDED
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Python
|
2 |
+
__pycache__/
|
3 |
+
*.py[cod]
|
4 |
+
*$py.class
|
5 |
+
*.so
|
6 |
+
.Python
|
7 |
+
build/
|
8 |
+
develop-eggs/
|
9 |
+
dist/
|
10 |
+
downloads/
|
11 |
+
eggs/
|
12 |
+
.eggs/
|
13 |
+
lib/
|
14 |
+
lib64/
|
15 |
+
parts/
|
16 |
+
sdist/
|
17 |
+
var/
|
18 |
+
wheels/
|
19 |
+
*.egg-info/
|
20 |
+
.installed.cfg
|
21 |
+
*.egg
|
22 |
+
|
23 |
+
# Virtual Environment
|
24 |
+
venv/
|
25 |
+
ENV/
|
26 |
+
env/
|
27 |
+
.venv/
|
28 |
+
|
29 |
+
# IDE
|
30 |
+
.vscode/
|
31 |
+
*.swp
|
32 |
+
*.swo
|
33 |
+
*~
|
34 |
+
|
35 |
+
# OS
|
36 |
+
.DS_Store
|
37 |
+
Thumbs.db
|
38 |
+
|
39 |
+
# Hugging Face
|
40 |
+
.gradio/
|
41 |
+
gradio_cached_examples/
|
42 |
+
|
43 |
+
# Scratch files
|
44 |
+
.scratches/
|
45 |
+
|
46 |
+
# PyCharm
|
47 |
+
.idea/
|
48 |
+
|
49 |
+
# Local environment files
|
50 |
+
.env
|
51 |
+
.env.local
|
52 |
+
|
53 |
+
# Model cache
|
54 |
+
models/
|
55 |
+
.cache/
|
.idea/.gitignore
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Default ignored files
|
2 |
+
/shelf/
|
3 |
+
/workspace.xml
|
4 |
+
# Editor-based HTTP Client requests
|
5 |
+
/httpRequests/
|
6 |
+
# Datasource local storage ignored files
|
7 |
+
/dataSources/
|
8 |
+
/dataSources.local.xml
|
.idea/dental-vqa-comparison.iml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<module type="PYTHON_MODULE" version="4">
|
3 |
+
<component name="NewModuleRootManager">
|
4 |
+
<content url="file://$MODULE_DIR$" />
|
5 |
+
<orderEntry type="inheritedJdk" />
|
6 |
+
<orderEntry type="sourceFolder" forTests="false" />
|
7 |
+
</component>
|
8 |
+
</module>
|
.idea/inspectionProfiles/profiles_settings.xml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<component name="InspectionProjectProfileManager">
|
2 |
+
<settings>
|
3 |
+
<option name="USE_PROJECT_PROFILE" value="false" />
|
4 |
+
<version value="1.0" />
|
5 |
+
</settings>
|
6 |
+
</component>
|
.idea/material_theme_project_new.xml
ADDED
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="MaterialThemeProjectNewConfig">
|
4 |
+
<option name="metadata">
|
5 |
+
<MTProjectMetadataState>
|
6 |
+
<option name="migrated" value="true" />
|
7 |
+
<option name="pristineConfig" value="false" />
|
8 |
+
<option name="userId" value="-31f020fc:19626a96fac:-7ffa" />
|
9 |
+
</MTProjectMetadataState>
|
10 |
+
</option>
|
11 |
+
</component>
|
12 |
+
</project>
|
.idea/modules.xml
ADDED
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="ProjectModuleManager">
|
4 |
+
<modules>
|
5 |
+
<module fileurl="file://$PROJECT_DIR$/.idea/dental-vqa-comparison.iml" filepath="$PROJECT_DIR$/.idea/dental-vqa-comparison.iml" />
|
6 |
+
</modules>
|
7 |
+
</component>
|
8 |
+
</project>
|
.idea/vcs.xml
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<project version="4">
|
3 |
+
<component name="VcsDirectoryMappings">
|
4 |
+
<mapping directory="$PROJECT_DIR$" vcs="Git" />
|
5 |
+
</component>
|
6 |
+
</project>
|
README.md
ADDED
@@ -0,0 +1,54 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
title: Dental VQA Model Comparison
|
3 |
+
emoji: 🦷
|
4 |
+
colorFrom: blue
|
5 |
+
colorTo: green
|
6 |
+
sdk: gradio
|
7 |
+
sdk_version: 4.44.0
|
8 |
+
app_file: app.py
|
9 |
+
pinned: false
|
10 |
+
license: apache-2.0
|
11 |
+
models:
|
12 |
+
- yasserrmd/DentaInstruct-1.2B
|
13 |
+
---
|
14 |
+
|
15 |
+
# Dental VQA Model Comparison
|
16 |
+
|
17 |
+
An interactive Gradio interface for comparing dental visual question answering models, currently featuring the DentaInstruct-1.2B model for educational information about dental health and oral care.
|
18 |
+
|
19 |
+
## Features
|
20 |
+
|
21 |
+
- Interactive chat interface for dental health questions
|
22 |
+
- Adjustable generation parameters (temperature, max tokens, etc.)
|
23 |
+
- Example questions to get started
|
24 |
+
- Mobile-responsive design
|
25 |
+
- Clear disclaimers about educational use only
|
26 |
+
|
27 |
+
## Important Disclaimer
|
28 |
+
|
29 |
+
⚠️ **This model is for educational purposes only.** It is NOT a substitute for professional dental care. Do not use this model for clinical diagnosis or treatment advice. Always consult a qualified dental professional.
|
30 |
+
|
31 |
+
## Model Information
|
32 |
+
|
33 |
+
- **Base Model**: LFM2-1.2B
|
34 |
+
- **Parameters**: 1.17B
|
35 |
+
- **Training Data**: Dental subset of MIRIAD dataset
|
36 |
+
- **Purpose**: Educational dental information
|
37 |
+
|
38 |
+
## Usage
|
39 |
+
|
40 |
+
Ask questions about:
|
41 |
+
- Dental procedures and treatments
|
42 |
+
- Oral health and hygiene
|
43 |
+
- Common dental conditions
|
44 |
+
- Preventive dental care
|
45 |
+
- Dental anatomy and terminology
|
46 |
+
|
47 |
+
## Credits
|
48 |
+
|
49 |
+
- **Model**: Created by @yasserrmd
|
50 |
+
- **Interface**: Space by @chrisvoncsefalvay
|
51 |
+
|
52 |
+
## License
|
53 |
+
|
54 |
+
Apache-2.0
|
app.py
ADDED
@@ -0,0 +1,234 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
import torch
|
3 |
+
from transformers import AutoModelForCausalLM, AutoTokenizer
|
4 |
+
|
5 |
+
# Model configuration
|
6 |
+
MODEL_ID = "yasserrmd/DentaInstruct-1.2B"
|
7 |
+
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
|
8 |
+
|
9 |
+
# Initialize model and tokenizer
|
10 |
+
print(f"Loading model {MODEL_ID}...")
|
11 |
+
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
|
12 |
+
model = AutoModelForCausalLM.from_pretrained(
|
13 |
+
MODEL_ID,
|
14 |
+
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
|
15 |
+
device_map="auto" if torch.cuda.is_available() else None
|
16 |
+
)
|
17 |
+
|
18 |
+
if not torch.cuda.is_available():
|
19 |
+
model = model.to(DEVICE)
|
20 |
+
|
21 |
+
# Set padding token if not set
|
22 |
+
if tokenizer.pad_token is None:
|
23 |
+
tokenizer.pad_token = tokenizer.eos_token
|
24 |
+
|
25 |
+
def format_prompt(message, history):
|
26 |
+
"""Format the prompt for the model"""
|
27 |
+
messages = []
|
28 |
+
|
29 |
+
# Add conversation history
|
30 |
+
for user_msg, assistant_msg in history:
|
31 |
+
messages.append({"role": "user", "content": user_msg})
|
32 |
+
if assistant_msg:
|
33 |
+
messages.append({"role": "assistant", "content": assistant_msg})
|
34 |
+
|
35 |
+
# Add current message
|
36 |
+
messages.append({"role": "user", "content": message})
|
37 |
+
|
38 |
+
# Apply chat template
|
39 |
+
if hasattr(tokenizer, 'apply_chat_template'):
|
40 |
+
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
41 |
+
else:
|
42 |
+
# Fallback formatting
|
43 |
+
prompt = ""
|
44 |
+
for msg in messages:
|
45 |
+
if msg["role"] == "user":
|
46 |
+
prompt += f"User: {msg['content']}\n"
|
47 |
+
else:
|
48 |
+
prompt += f"Assistant: {msg['content']}\n"
|
49 |
+
prompt += "Assistant: "
|
50 |
+
|
51 |
+
return prompt
|
52 |
+
|
53 |
+
def generate_response(
|
54 |
+
message,
|
55 |
+
history,
|
56 |
+
temperature=0.7,
|
57 |
+
max_new_tokens=512,
|
58 |
+
top_p=0.95,
|
59 |
+
repetition_penalty=1.1,
|
60 |
+
):
|
61 |
+
"""Generate response from the model"""
|
62 |
+
|
63 |
+
# Format the prompt
|
64 |
+
prompt = format_prompt(message, history)
|
65 |
+
|
66 |
+
# Tokenize input
|
67 |
+
inputs = tokenizer(prompt, return_tensors="pt", truncation=True, max_length=2048)
|
68 |
+
inputs = {k: v.to(model.device) for k, v in inputs.items()}
|
69 |
+
|
70 |
+
# Generate response
|
71 |
+
with torch.no_grad():
|
72 |
+
outputs = model.generate(
|
73 |
+
**inputs,
|
74 |
+
max_new_tokens=max_new_tokens,
|
75 |
+
temperature=temperature,
|
76 |
+
top_p=top_p,
|
77 |
+
repetition_penalty=repetition_penalty,
|
78 |
+
do_sample=True,
|
79 |
+
pad_token_id=tokenizer.pad_token_id,
|
80 |
+
eos_token_id=tokenizer.eos_token_id,
|
81 |
+
)
|
82 |
+
|
83 |
+
# Decode response
|
84 |
+
response = tokenizer.decode(outputs[0][inputs['input_ids'].shape[1]:], skip_special_tokens=True)
|
85 |
+
|
86 |
+
return response
|
87 |
+
|
88 |
+
# Example questions
|
89 |
+
EXAMPLES = [
|
90 |
+
["What are the main types of dental cavities?"],
|
91 |
+
["Explain the process of root canal treatment"],
|
92 |
+
["What is the difference between gingivitis and periodontitis?"],
|
93 |
+
["How should I care for my teeth after a dental extraction?"],
|
94 |
+
["What are the benefits of fluoride in dental care?"],
|
95 |
+
["Explain the stages of tooth development in children"],
|
96 |
+
["What causes tooth sensitivity and how can it be treated?"],
|
97 |
+
["Describe the different types of dental fillings available"],
|
98 |
+
]
|
99 |
+
|
100 |
+
# Custom CSS for styling
|
101 |
+
custom_css = """
|
102 |
+
.disclaimer {
|
103 |
+
background-color: #fff3cd;
|
104 |
+
border: 1px solid #ffc107;
|
105 |
+
border-radius: 5px;
|
106 |
+
padding: 10px;
|
107 |
+
margin-bottom: 15px;
|
108 |
+
}
|
109 |
+
"""
|
110 |
+
|
111 |
+
# Create Gradio interface
|
112 |
+
with gr.Blocks(theme=gr.themes.Soft(), css=custom_css) as demo:
|
113 |
+
gr.Markdown(
|
114 |
+
"""
|
115 |
+
# Dental VQA Model Comparison
|
116 |
+
|
117 |
+
Interactive comparison of dental visual question answering models. Currently featuring DentaInstruct-1.2B for dental education and oral health information.
|
118 |
+
"""
|
119 |
+
)
|
120 |
+
|
121 |
+
gr.HTML(
|
122 |
+
"""
|
123 |
+
<div class="disclaimer">
|
124 |
+
<strong>⚠️ Important Disclaimer:</strong><br>
|
125 |
+
This model is for educational purposes only. It is NOT a substitute for professional dental care.
|
126 |
+
Do not use this model for clinical diagnosis or treatment advice. Always consult a qualified dental professional.
|
127 |
+
</div>
|
128 |
+
"""
|
129 |
+
)
|
130 |
+
|
131 |
+
chatbot = gr.Chatbot(
|
132 |
+
height=400,
|
133 |
+
label="Conversation"
|
134 |
+
)
|
135 |
+
|
136 |
+
msg = gr.Textbox(
|
137 |
+
label="Your dental question",
|
138 |
+
placeholder="Ask a question about dental health, procedures, or oral care...",
|
139 |
+
lines=2
|
140 |
+
)
|
141 |
+
|
142 |
+
with gr.Row():
|
143 |
+
submit = gr.Button("Send", variant="primary")
|
144 |
+
clear = gr.Button("Clear")
|
145 |
+
|
146 |
+
with gr.Accordion("Advanced Settings", open=False):
|
147 |
+
temperature = gr.Slider(
|
148 |
+
minimum=0.1,
|
149 |
+
maximum=1.0,
|
150 |
+
value=0.7,
|
151 |
+
step=0.1,
|
152 |
+
label="Temperature",
|
153 |
+
info="Controls randomness in responses"
|
154 |
+
)
|
155 |
+
|
156 |
+
max_new_tokens = gr.Slider(
|
157 |
+
minimum=64,
|
158 |
+
maximum=1024,
|
159 |
+
value=512,
|
160 |
+
step=64,
|
161 |
+
label="Max New Tokens",
|
162 |
+
info="Maximum length of the response"
|
163 |
+
)
|
164 |
+
|
165 |
+
top_p = gr.Slider(
|
166 |
+
minimum=0.1,
|
167 |
+
maximum=1.0,
|
168 |
+
value=0.95,
|
169 |
+
step=0.05,
|
170 |
+
label="Top-p",
|
171 |
+
info="Nucleus sampling parameter"
|
172 |
+
)
|
173 |
+
|
174 |
+
repetition_penalty = gr.Slider(
|
175 |
+
minimum=1.0,
|
176 |
+
maximum=1.5,
|
177 |
+
value=1.1,
|
178 |
+
step=0.05,
|
179 |
+
label="Repetition Penalty",
|
180 |
+
info="Reduces repetition in responses"
|
181 |
+
)
|
182 |
+
|
183 |
+
gr.Examples(
|
184 |
+
examples=EXAMPLES,
|
185 |
+
inputs=msg,
|
186 |
+
label="Example Questions"
|
187 |
+
)
|
188 |
+
|
189 |
+
gr.Markdown(
|
190 |
+
"""
|
191 |
+
## About This Model
|
192 |
+
|
193 |
+
DentaInstruct-1.2B is a specialised language model fine-tuned on dental educational content.
|
194 |
+
It's designed to provide educational information about dental health, procedures, and oral care.
|
195 |
+
|
196 |
+
**Model Details:**
|
197 |
+
- Base Model: LFM2-1.2B
|
198 |
+
- Parameters: 1.17B
|
199 |
+
- Training Data: Dental subset of MIRIAD dataset
|
200 |
+
- Purpose: Educational dental information
|
201 |
+
|
202 |
+
**Created by:** @yasserrmd | **Space by:** @chrisvoncsefalvay
|
203 |
+
"""
|
204 |
+
)
|
205 |
+
|
206 |
+
# Event handlers
|
207 |
+
def respond(message, chat_history, temperature, max_new_tokens, top_p, repetition_penalty):
|
208 |
+
response = generate_response(
|
209 |
+
message,
|
210 |
+
chat_history,
|
211 |
+
temperature,
|
212 |
+
max_new_tokens,
|
213 |
+
top_p,
|
214 |
+
repetition_penalty
|
215 |
+
)
|
216 |
+
chat_history.append((message, response))
|
217 |
+
return "", chat_history
|
218 |
+
|
219 |
+
msg.submit(
|
220 |
+
respond,
|
221 |
+
[msg, chatbot, temperature, max_new_tokens, top_p, repetition_penalty],
|
222 |
+
[msg, chatbot]
|
223 |
+
)
|
224 |
+
|
225 |
+
submit.click(
|
226 |
+
respond,
|
227 |
+
[msg, chatbot, temperature, max_new_tokens, top_p, repetition_penalty],
|
228 |
+
[msg, chatbot]
|
229 |
+
)
|
230 |
+
|
231 |
+
clear.click(lambda: None, None, chatbot, queue=False)
|
232 |
+
|
233 |
+
if __name__ == "__main__":
|
234 |
+
demo.launch()
|
requirements.txt
ADDED
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
gradio==4.44.0
|
2 |
+
transformers==4.44.2
|
3 |
+
torch>=2.0.0
|
4 |
+
accelerate==0.33.0
|
5 |
+
sentencepiece==0.2.0
|
6 |
+
protobuf==5.27.3
|