search

cxh0519 / threestudio-gaussiandreamer

GaussianDreamer extension of threestudio.
MIT License
44 stars 3 forks source link

how to accumulate the gradient ? #3

Closed 1986727548 closed 7 months ago

1986727548 commented 7 months ago

since automatic_optimization is set false, how to accumulate the gradient of several batches and achieve a large effective batch size?

cxh0519 commented 7 months ago

The training is inherited from threestudio-3dgs, please check their codes for more details.