How to train it on custom datasets?

この issue には説明がありません。

The issue asks about training on custom datasets but provides no details. Look at the repository's README and existing documentation for any training scripts or guides. Check if there are example datasets, configuration files, or a training pipeline in the codebase. The goal is to document the steps needed to prepare data and run training.

索引モデルが issue の本文から書いたものです。