Questions about GPU Configuration and Memory Requirements for BuildingGNN Model Training

[kg571852741]

Hi @buildingnet,

Could you please provide information on the GPU configuration and memory requirements to train the BuildingGNN model successfully with my current training settings? I am using 2x RTX 4090Ti (48GB each) GPUs, but often run out of memory during training.
Train setting:

python3 train.py --loadfromckpt 1 --datadir="data" --epoch 200 --batch_size 1 --outputdir 'Output' --ckpt_dir 'checkpoint' --normalization 'GN' --modeltype 'Edge' --edgetype 'all' --lr 1e-4 --nodetype 'node+minkownormal' --pretrainedtype 'fc3_avg'

Additionally, do you have any scripts or tutorials to visualize the inference and predicted labels for point cloud models (e.g., ply, rcs)? It would be great to evaluate the model's performance.

Thank you for your time and the excellent work on BuildingNet. I appreciate your help.