rsoohyun/SpatialBlock-4B-direct

SpatialBlock-4B-direct

This repository contains the SpatialBlock-4B-direct checkpoint from the paper SpatialBlock: Enhancing Spatial Intelligence in LVLMs via Synthetic Block-Stacking Problem.

It is a fine-tuned version of Qwen3-VL-4B-Instruct on the synthetic SpatialBlock-15k dataset. The model directly predicts answers to spatial reasoning tasks such as 3D-to-2D projection, viewpoint transformation, and structural combination.

For training details, evaluation results, and the companion “reason” model, please refer to the GitHub repository: https://github.com/rsoohyun/SpatialBlock.