Dataset · maghzal
Route Planning Evaluation Benchmark for VLM
maghzal/PathEval
A dataset for testing vision-language models (VLM) on evaluating plan and route quality.
What's inside
Gathers between 10,000 and 100,000 image-text examples designed to evaluate vision-language models in the role of plan critics or evaluators. Created to study whether large language and multimodal models can effectively verify ready-made routes even if they struggle with end-to-end planning.
What you can do with it
- Evaluating and comparing VLMs in plan evaluation and critique tasks
- Testing multimodal models for navigation and spatial reasoning
- Size
- 10K–100K rows
- Downloads this month
- 27.4K
- Likes
- 0
How to load it
from datasets import load_dataset
ds = load_dataset("maghzal/PathEval")