|
import pandas as pd |
|
import gradio as gr |
|
|
|
data = { |
|
"Method": [ |
|
"GPT-4o", "GPT-4o-mini", "Gemini-1.5-Pro", "Gemini-1.5-Flash", "Qwen2-VL-2B", "Pangea-7B", "InternVL2-8B", "LLaVa-NeXt-7B" |
|
], |
|
"MM Understanding & Reasoning": [ |
|
57.90, 48.82, 46.67, 45.58, 40.59, 40.09, 30.41, 26.33 |
|
], |
|
"OCR & Document Understanding": [ |
|
59.11, 42.89, 36.59, 33.59, 25.68, 26.47, 15.91, 19.12 |
|
], |
|
"Charts & Diagram Understanding": [ |
|
73.57, 64.98, 47.06, 48.25, 27.83, 38.87, 30.27, 27.56 |
|
], |
|
"Video Understanding": [ |
|
74.27, 68.11, 42.94, 53.31, 38.90, 49.01, 51.42, 44.90 |
|
], |
|
"Cultural Specific Understanding": [ |
|
80.86, 65.92, 56.24, 46.54, 34.27, 20.34, 20.88, 28.30 |
|
], |
|
"Medical Imaging": [ |
|
49.90, 47.37, 33.77, 42.86, 29.12, 31.99, 29.48, 22.54 |
|
], |
|
"Agro Specific": [ |
|
80.75, 79.58, 72.12, 76.06, 52.02, 74.51, 44.47, 42.00 |
|
], |
|
"Remote Sensing Understanding": [ |
|
22.85, 16.93, 17.07, 14.95, 12.56, 6.67, 5.36, 8.33 |
|
] |
|
} |
|
|
|
df = pd.DataFrame(data) |
|
df['Average Score'] = df.iloc[:, 1:].mean(axis=1) |
|
|
|
def display_data(): |
|
return df |
|
|
|
with gr.Blocks() as demo: |
|
gr.Markdown("", elem_id="camel-icon") |
|
gr.Markdown("# **CAMEL-Bench: Model Performance Across Vision Understanding Tasks**") |
|
gr.Markdown(""" |
|
This table shows the performance of different models across various tasks including OCR, chart understanding, video, medical imaging, and more. |
|
""") |
|
gr.Dataframe(value=df, label="CAMEL-Bench Model Performance", interactive=False) |
|
|
|
demo.launch() |
|
|
