axinc-ai / ailia-models

The collection of pre-trained, state-of-the-art AI models for ailia SDK
2.04k stars 325 forks source link

ADD Japanese-CLIP #1246

Closed itsmeterada closed 1 year ago

itsmeterada commented 1 year ago

https://github.com/rinnakk/japanese-clip

kyakuno commented 1 year ago

@ooe1123 BARKが終わったら、こちらのモデルをお願いできればと考えています。

kyakuno commented 1 year ago

データセットはCC12M。 https://github.com/google-research-datasets/conceptual-12m

kyakuno commented 1 year ago
Training
The model was trained on [CC12M](https://github.com/google-research-datasets/conceptual-12m) translated the captions to Japanese.

https://huggingface.co/rinna/japanese-clip-vit-b-16

kyakuno commented 1 year ago

TokenizeにはT5Tokenizerを使用している。