How to judge the convergence of the pre-training model？

ChenRocks / UNITER

Research code for ECCV 2020 paper "UNITER: UNiversal Image-TExt Representation Learning"

https://arxiv.org/abs/1909.11740

777 stars 109 forks source link

Open PamelaDDD opened 2 years ago

PamelaDDD commented 2 years ago

How to measure the loss weight of different pre-training tasks? Which task's loss determines the model training convergence?