| PYTHONWARNINGS="ignore" CUDA_VISIBLE_DEVICES=4 python -m torch.distributed.launch --nproc_per_node=1 \ | |
| --master_port=$(( RANDOM % 1000 + 50000 )) \ | |
| eval_sfrustum.py --config configs/sfrustum_scannet.txt \ | |
| --ckpt out/frustum-scannet-generalized_trainning_4layer/model_379999.pth \ | |
| --expname frustum-scannet-eval --val_set_list configs/scannetv2_test_split.txt --no_load_opt --no_load_scheduler |