THUDM / CogView

Text-to-Image generation. The repo for NeurIPS 2021 paper "CogView: Mastering Text-to-Image Generation via Transformers".
Apache License 2.0
1.72k stars 174 forks source link

the Alibaba item-title image tokens dataset from our link at Tianchi(TODO) #24

Closed WXLL579 closed 3 years ago

WXLL579 commented 3 years ago

the link to the Alibaba item-title image tokens dataset at Tianchi(TODO) seems to be unavailable

Sleepychord commented 3 years ago

Hi, we have already update another the example dataset, which is only about 80,000 images. The training performance should not be as good as the released CogView. It is only an example dataset to try.