Missing download sources for required T2M evaluation models mentioned in Quick Start

[Raincxt]

Hi authors,
Thank you for releasing the FRoM-W1 / H-GPT codebase.
While following the Quick Start section in the README, I noticed that the evaluation pipeline requires several T2M-related models to be placed under H-GPT/deps/t2m/, with the following directory structure explicitly specified:

t2m/
├── Comp_v6_KLD005
│ ├── meta
│ └── opt.txt
├── Comp_v6_KLD01
│ ├── meta
│ ├── model
│ └── opt.txt
├── VQVAEV3_CB1024_CMT_H1024_NRES3
│ ├── meta
│ └── model
└── text_mot_match
├── eval
└── model

However, I could not find any download links or references for these required models in the README, repository, or supplementary materials.
Could you please clarify:

1. Where these T2M evaluation models can be downloaded (e.g., Google Drive, HuggingFace, Baidu, etc.), or
2. Whether these models are reused from an existing public benchmark or repository, and if so, which one and which checkpoints should be used?
   This information would greatly help in reproducing the evaluation results described in the paper.
   Thanks again for sharing your work and for your time.
   Best regards,

[artpli]

Hi @Raincxt , thank you for your interest!

Our datasets and checkpoints are maintained in this huggingface repo. But due to the Spring Festival, we have not orginized/documented them very well. You can have a look at them first and we will update them this week, then I will remind you when ended.

Best,
Peng