# Long-tail Planning with Language

Our evaluation dataset shares the same format as the DriveLM dataset and can be downloaded here: [v1_1_val_nus_q_only_with_long_tail](https://mega.nz/file/AS8USBCY#6Oqnnwz7E1z1pRuqWlqD4YGC1uM6yCaKuRwUO8dTG8I).  
Note that the evaluation data file is the same as the `DriveLM-nuScenes version-1.1 val` (the data file used in the `driving with language` track). We directly added our long-tail planning QAs into each evaluation keyframe (if the scenario is a long-tail event).

<!-- - <a href="https://github.com/OpenDriveLab/DriveLM" target="_blank">Github</a> | <a href="https://arxiv.org/abs/2312.14150" target="_blank">Paper</a>

- Point of contact: [Chonghao (司马崇昊)](mailto:chonghaosima@gmail.com) -->

<!-- ## Dataset Description

Please visit <a href="https://github.com/OpenDriveLab/DriveLM" target="_blank">DriveLM: Driving with Graph Visual Question Answering</a> for details. -->

<!-- ## Dataset Download

<a href="https://github.com/OpenDriveLab/DriveLM/tree/main/challenge" target="_blank">The baseline code</a> can run on both full-scale and demo train data. All the code in the challenge repo (including train / infer the baseline, and evaluation) supports demo train data (which is in the same format of the full-scale trian data).

For dataset download, you can visit the following pages.

- <a href="https://github.com/OpenDriveLab/DriveLM/tree/main/challenge#drivelm" target="_blank">DriveLM-nuScenes version-1.1 dataset download</a>.  -->