graphnav-anonymous commited on
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,7 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
# GraphNav: Benchmarking Spatial Cognitive Graph Reasoning in Vision-Language Models
|
| 2 |
|
| 3 |
## Overview
|
| 4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
GraphNav is a controlled benchmark designed to evaluate spatial cognitive-graph reasoning in vision-language models (VLMs) by isolating high-level topological reasoning from execution-level confounds. The benchmark comprises **420 procedurally generated 3D maze environments** and **37,443 navigation paths** across three navigation tasks.
|
| 6 |
|
| 7 |
- **Paper**: Submitted to NeurIPS 2026 Evaluations and Datasets Track
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: cc-by-4.0
|
| 3 |
+
task_categories:
|
| 4 |
+
- visual-question-answering
|
| 5 |
+
- robotics
|
| 6 |
+
tags:
|
| 7 |
+
- spatial-reasoning
|
| 8 |
+
- navigation
|
| 9 |
+
- vision-language-models
|
| 10 |
+
- benchmark
|
| 11 |
+
- cognitive-graph
|
| 12 |
+
size_categories:
|
| 13 |
+
- 10K<n<100K
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
# GraphNav: Benchmarking Spatial Cognitive Graph Reasoning in Vision-Language Models
|
| 17 |
|
| 18 |
## Overview
|
| 19 |
|
| 20 |
+
GraphNav is a controlled benchmark designed to evaluate spatial cognitive-graph reasoning in vision-language models (VLMs) # GraphNav: Benchmarking Spatial Cognitive Graph Reasoning in Vision-Language Models
|
| 21 |
+
|
| 22 |
+
## Overview
|
| 23 |
+
|
| 24 |
GraphNav is a controlled benchmark designed to evaluate spatial cognitive-graph reasoning in vision-language models (VLMs) by isolating high-level topological reasoning from execution-level confounds. The benchmark comprises **420 procedurally generated 3D maze environments** and **37,443 navigation paths** across three navigation tasks.
|
| 25 |
|
| 26 |
- **Paper**: Submitted to NeurIPS 2026 Evaluations and Datasets Track
|