Project Page: VLN-R1: Vision-Language Navigation via Reinforcement Fine-Tuning

This is the training dataset of VLN-R1. It is collected from R2R and RxR, and can be directly used for training with LLaMA-Factory.

We have also released the process of making VLN-Ego using VLN-CE, with the code available at VLN-Ego-making.

这是VLN-R1的训练数据集。它源自R2R和RxR，可直接用于LLaMA-Factory的训练。

我们也公布了使用VLN-CE制作VLN-Ego的过程，相关代码位于VLN-Ego-making。