Cuda Memory Error

[ahm3draxa]

I am running Stable Diffusion / Automatic1111 WebUI Locally on RTX 3060 (12 GB VRAM). Whenever I try to train the model whether its 2.1 or 1.5 (512 or 768) every time I run in to problem of "OOM" Error and "Cuda Out of Memory".

Current Batch Size = 1
Tried reducing image size and still same.

Any help would be appreciated, please