We have set the parameters of the Toronto 3D dataset based on the S3DIS dataset, but during training, we have already set the batch_ The size has been reduced to the minimum, but there is still an error reported that the GPU is out of memory.

RuntimeError: CUDA out of memory. Tried to allocate 16.32 GiB (GPU 0; 47.35 GiB total capacity; 37.84 GiB already allocated; 8.27 GiB free; 37.98 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF The configs file is as follows: `base = [“…/base/default_runtime.py”]

misc custom setting

batch_size = 1 #12 # bs: total bs in all gpus mix_prob = 0.8 empty_cache = False enable_amp = True

model settings

model = dict( type=“DefaultSegmentor”, backbone=dict( type=“PT-v2m2”, in_channels=6, num_classes=8, patch_embed_depth=2, patch_embed_channels=48, patch_embed_groups=6, patch_embed_neighbours=16, enc_depths=(2, 6, 2), enc_channels=(96, 192, 384), enc_groups=(12, 24, 48), enc_neighbours=(16, 16, 16), dec_depths=(1, 1, 1), dec_channels=(48, 96, 192), dec_groups=(6, 12, 24), dec_neighbours=(16, 16, 16), grid_sizes=(0.1, 0.2, 0.4), attn_qkv_bias=True, pe_multiplier=False, pe_bias=True, attn_drop_rate=0.0, drop_path_rate=0.3, enable_checkpoint=False, unpool_backend=“interp”, # map / interp ), criteria=[dict(type=“CrossEntropyLoss”, loss_weight=1.0, ignore_index=-1)], )

scheduler settings

epoch = 3000 optimizer = dict(type=“AdamW”, lr=0.006, weight_decay=0.05) scheduler = dict(type=“MultiStepLR”, milestones=[0.6, 0.8], gamma=0.1)

dataset settings

dataset_type = “TorontoDataset” data_root = “/root/autodl-tmp/toronto1”

data = dict( num_classes=8, ignore_index=-1, names=[ “road”, “road_marking”, “natural”, “building”, “utility_line”, “pole”, “car”, “fence”, ], train=dict( type=dataset_type, split=(“L001”, “L002”, “L003”, “L004”), data_root=data_root, transform=[ dict(type=“CenterShift”, apply_z=True), # dict(type=“RandomDropout”, dropout_ratio=0.2, dropout_application_ratio=0.2), # dict(type=“RandomRotateTargetAngle”, angle=(1/2, 1, 3/2), center=[0, 0, 0], axis=“z”, p=0.75), # dict(type=“RandomRotate”, angle=[-1, 1], axis=“z”, center=[0, 0, 0], p=0.5), # dict(type=“RandomRotate”, angle=[-1 / 64, 1 / 64], axis=“x”, p=0.5), # dict(type=“RandomRotate”, angle=[-1 / 64, 1 / 64], axis=“y”, p=0.5), dict(type=“RandomScale”, scale=[0.9, 1.1]), # dict(type=“RandomShift”, shift=[0.2, 0.2, 0.2]), dict(type=“RandomFlip”, p=0.5), dict(type=“RandomJitter”, sigma=0.005, clip=0.02), # dict(type=“ElasticDistortion”, distortion_params=[[0.2, 0.4], [0.8, 1.6]]), dict(type=“ChromaticAutoContrast”, p=0.2, blend_factor=None), dict(type=“ChromaticTranslation”, p=0.95, ratio=0.05), dict(type=“ChromaticJitter”, p=0.95, std=0.05), # dict(type=“HueSaturationTranslation”, hue_max=0.2, saturation_max=0.2), # dict(type=“RandomColorDrop”, p=0.2, color_augment=0.0), dict( type=“GridSample”, grid_size=0.04, #0.04 hash_type=“fnv”, mode=“train”, keys=(“coord”, “color”, “segment”), return_discrete_coord=True, ), dict(type=“SphereCrop”, point_max=80000, mode=“random”), dict(type=“CenterShift”, apply_z=False), dict(type=“NormalizeColor”), # dict(type=“ShufflePoint”), dict(type=“ToTensor”), dict( type=“Collect”, keys=(“coord”, “discrete_coord”, “segment”), feat_keys=[“coord”, “color”], ), ], test_mode=False, ), val=dict( type=dataset_type, split=“L002”, data_root=data_root, transform=[ dict(type=“CenterShift”, apply_z=True), dict( type=“Copy”, keys_dict={“coord”: “origin_coord”, “segment”: “origin_segment”}, ), dict( type=“GridSample”, grid_size=0.04, #0.04 hash_type=“fnv”, mode=“train”, keys=(“coord”, “color”, “segment”), return_discrete_coord=True, ), dict(type=“CenterShift”, apply_z=False), dict(type=“NormalizeColor”), dict(type=“ToTensor”), dict( type=“Collect”, keys=(“coord”, “discrete_coord”, “segment”), offset_keys_dict=dict(offset=“coord”), feat_keys=[“coord”, “color”], ), ], test_mode=False, ), test=dict( type=dataset_type, split=“L002”, data_root=data_root, transform=[dict(type=“CenterShift”, apply_z=True), dict(type=“NormalizeColor”)], test_mode=True, test_cfg=dict( voxelize=dict( type=“GridSample”, grid_size=0.04, #0.04 hash_type=“fnv”, mode=“test”, keys=(“coord”, “color”), return_discrete_coord=True, ), crop=None, post_transform=[ dict(type=“CenterShift”, apply_z=False), dict(type=“ToTensor”), dict( type=“Collect”, keys=(“coord”, “discrete_coord”, “index”), feat_keys=(“coord”, “color”), ), ], aug_transform=[ [dict(type=“RandomScale”, scale=[0.9, 0.9])], [dict(type=“RandomScale”, scale=[0.95, 0.95])], [dict(type=“RandomScale”, scale=[1, 1])], [dict(type=“RandomScale”, scale=[1.05, 1.05])], [dict(type=“RandomScale”, scale=[1.1, 1.1])], [ dict(type=“RandomScale”, scale=[0.9, 0.9]), dict(type=“RandomFlip”, p=1), ], [ dict(type=“RandomScale”, scale=[0.95, 0.95]), dict(type=“RandomFlip”, p=1), ], [dict(type=“RandomScale”, scale=[1, 1]), dict(type=“RandomFlip”, p=1)], [ dict(type=“RandomScale”, scale=[1.05, 1.05]), dict(type=“RandomFlip”, p=1), ], [ dict(type=“RandomScale”, scale=[1.1, 1.1]), dict(type=“RandomFlip”, p=1), ], ], ), ), ) ` What parameters should we adjust?