ahmedheakl
commited on
Commit
β’
f2e3361
1
Parent(s):
b0ee7b4
Update app.py
Browse files
app.py
CHANGED
@@ -1,75 +1,26 @@
|
|
1 |
import pandas as pd
|
2 |
import gradio as gr
|
3 |
-
import plotly.graph_objects as go
|
4 |
|
5 |
-
# Create the DataFrame
|
6 |
data = {
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
16 |
}
|
17 |
|
18 |
df = pd.DataFrame(data)
|
19 |
|
20 |
-
def
|
21 |
-
categories = df.columns[1:]
|
22 |
-
fig = go.Figure()
|
23 |
-
|
24 |
-
for method in df['Method']:
|
25 |
-
values = df[df['Method'] == method].iloc[0, 1:].tolist()
|
26 |
-
fig.add_trace(go.Scatterpolar(
|
27 |
-
r=values,
|
28 |
-
theta=categories,
|
29 |
-
fill='toself',
|
30 |
-
name=method
|
31 |
-
))
|
32 |
-
|
33 |
-
fig.update_layout(
|
34 |
-
polar=dict(
|
35 |
-
radialaxis=dict(
|
36 |
-
visible=True,
|
37 |
-
range=[0, 100]
|
38 |
-
)),
|
39 |
-
showlegend=True,
|
40 |
-
title="Performance Comparison across Categories"
|
41 |
-
)
|
42 |
-
return fig
|
43 |
-
|
44 |
-
def create_leaderboard():
|
45 |
return df
|
46 |
|
47 |
-
# Define the Gradio interface
|
48 |
with gr.Blocks() as demo:
|
49 |
-
gr.Markdown("#
|
50 |
-
|
51 |
-
|
52 |
-
with gr.TabItem("π Performance Plot"):
|
53 |
-
gr.Plot(plot_performance)
|
54 |
-
|
55 |
-
with gr.TabItem("π Leaderboard Table"):
|
56 |
-
gr.DataFrame(create_leaderboard)
|
57 |
-
|
58 |
-
with gr.TabItem("π About"):
|
59 |
-
gr.Markdown("""
|
60 |
-
This leaderboard compares the performance of various models across different categories of multimodal understanding tasks. The scores represent the accuracy or performance metric for each model in the respective category.
|
61 |
-
|
62 |
-
**Categories:**
|
63 |
-
- MM Understanding & Reasoning
|
64 |
-
- OCR & Document Understanding
|
65 |
-
- Charts & Diagram Understanding
|
66 |
-
- Video Understanding
|
67 |
-
- Cultural Specific Understanding
|
68 |
-
- Medical Imaging
|
69 |
-
- Agro Specific
|
70 |
-
- Remote Sensing Understanding
|
71 |
-
|
72 |
-
The data is presented both as a radar chart for visual comparison and as a table for detailed viewing.
|
73 |
-
""")
|
74 |
|
75 |
-
demo.launch()
|
|
|
1 |
import pandas as pd
|
2 |
import gradio as gr
|
|
|
3 |
|
|
|
4 |
data = {
|
5 |
+
"Method": ["GPT-4o", "GPT-4o-mini", "Gemini-1.5-Pro", "Gemini-1.5-Flash", "Qwen2-VL-2B"],
|
6 |
+
"MM Understanding & Reasoning": [57.90, 48.82, 46.67, 45.58, 40.59],
|
7 |
+
"OCR & Document Understanding": [59.11, 42.89, 36.59, 33.59, 25.68],
|
8 |
+
"Charts & Diagram Understanding": [73.57, 64.98, 47.06, 48.25, 27.83],
|
9 |
+
"Video Understanding": [74.27, 68.11, 42.94, 53.31, 38.90],
|
10 |
+
"Cultural Specific Understanding": [80.86, 65.92, 56.24, 46.54, 34.27],
|
11 |
+
"Medical Imaging": [49.90, 47.37, 33.77, 42.86, 29.12],
|
12 |
+
"Agro Specific": [80.75, 79.58, 72.12, 76.06, 52.02],
|
13 |
+
"Remote Sensing Understanding": [22.85, 16.93, 17.07, 14.95, 12.56]
|
14 |
}
|
15 |
|
16 |
df = pd.DataFrame(data)
|
17 |
|
18 |
+
def display_data():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
return df
|
20 |
|
|
|
21 |
with gr.Blocks() as demo:
|
22 |
+
gr.Markdown("# Model Performance Across Various Understanding Tasks")
|
23 |
+
gr.Markdown("This table shows the performance of different models across various tasks including OCR, chart understanding, video, medical imaging, and more.")
|
24 |
+
gr.Dataframe(value=df, label="Model Performance", interactive=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
25 |
|
26 |
+
demo.launch()
|