a problem with the default learning configuration

The-Learning-And-Vision-Atelier-LAVA / PAM

[TPAMI 2020] Parallax Attention for Unsupervised Stereo Correspondence Learning

95 stars 16 forks source link

Closed HydrogenSulfate closed 3 years ago

HydrogenSulfate commented 3 years ago

lr = cfg.lr * (cfg.gamma ** -(epoch // cfg.n_steps)) cfg.gamma = 0.1, so the lr might become larger during training?

LongguangWang commented 3 years ago

Hi @HydrogenSulfate , this bug has been fixed.