search

ruotianluo / ImageCaptioning.pytorch

I decide to sync up this repo and self-critical.pytorch. (The old master is in old master branch for archive)
MIT License
1.44k stars 412 forks source link

在模型中有structure_after的参数，这是指在structure_after之前模型训练用cross-entropy，之后用的是CIDEr optimization一类的吗 #194

Open yeeeeee0 opened 11 months ago

ruotianluo commented 11 months ago

是的。Original self critical 用的是self critical after. 如果想用我改进的new self critical, 用的是structure afterRuotian LuoOn Oct 29, 2023, at 8:50 PM, yeeeeee0 @.***> wrote:

—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: @.***>