Update README.md
Browse files
README.md
CHANGED
@@ -10,8 +10,15 @@ tags:
|
|
10 |
- intelligence-test
|
11 |
---
|
12 |
# 🧠 ReasonBench: Benchmark for Complex Visual Reasoning
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
-
<img src="https://huggingface.co/datasets/cistine/ReasonBench/resolve/main/image_1.jpg" alt="模型架构图" width="50%"/>
|
15 |
|
16 |
## 🌐 Overview
|
17 |
**ReasonBench** is a comprehensive benchmark designed to evaluate Visual Language Models (VLMs) on complex graphical reasoning tasks. It contains **1,613 problems** collected from real-world intelligence tests, covering **11 core cognitive dimensions** and **29 task types**. This benchmark provides a robust framework for assessing VLMs' spatial, relational, and abstract reasoning capabilities.
|
|
|
10 |
- intelligence-test
|
11 |
---
|
12 |
# 🧠 ReasonBench: Benchmark for Complex Visual Reasoning
|
13 |
+
<div style="text-align:center">
|
14 |
+
|
15 |
+
</div>
|
16 |
+
|
17 |
+
<img src="https://huggingface.co/datasets/cistine/ReasonBench/resolve/main/image_1.jpg"
|
18 |
+
alt="background"
|
19 |
+
width="50%"/>
|
20 |
+
<p style="font-style:italic">image:background</p>
|
21 |
|
|
|
22 |
|
23 |
## 🌐 Overview
|
24 |
**ReasonBench** is a comprehensive benchmark designed to evaluate Visual Language Models (VLMs) on complex graphical reasoning tasks. It contains **1,613 problems** collected from real-world intelligence tests, covering **11 core cognitive dimensions** and **29 task types**. This benchmark provides a robust framework for assessing VLMs' spatial, relational, and abstract reasoning capabilities.
|