Datasets and annotated method

[5Jiahe]

Whether the N-ImageNet-Chat, N-ImageNet-Instruction and Event-Chat datasets will be released? How were the instructions and dialogue data included in it annotated? Were they generated?

[XduSyL]

We have released a subset of the EventGPT dataset, comprising nearly 90,000 samples. You can access it on Hugging Face.
Please stay tuned for our future open-source updates.