maptr训练卡死

一、问题现象（附报错日志上下文）：
参考model_sample中的maptr训练步骤训练，程序卡死
日志如下：
[root@fedf3901b7b1 MapTR]# ./MapTR/tools/dist_train.sh ./MapTR/projects/configs/maptr/maptr_tiny_r50_24e_bevformer.py 1
Warning: please export TSAN_OPTIONS='ignore_noninstrumented_modules=1' to avoid false positive reports from the OpenMP runtime!
/usr/local/python3.8/lib/python3.8/site-packages/torch/distributed/launch.py:181: FutureWarning: The module torch.distributed.launch is deprecated
and will be removed in future. Use torchrun.
Note that --use-env is set by default in torchrun.
If your script expects `--local-rank` argument to be set, please
change it to read from `os.environ['LOCAL_RANK']` instead. See
https://pytorch.org/docs/stable/distributed.html#launch-utility for
further instructions

warnings.warn(
Warning: please export TSAN_OPTIONS='ignore_noninstrumented_modules=1' to avoid false positive reports from the OpenMP runtime!
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/__init__.py:20: UserWarning: On January 1, 2023, MMCV will release v2.0.0, in which it will remove components related to the training process and add a data transformation module. In addition, it will rename the package names mmcv to mmcv-lite and mmcv-full to mmcv. See https://github.com/open-mmlab/mmcv/blob/master/docs/en/compatibility.md for more details.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmdetection3d/mmdet3d/core/evaluation/kitti_utils/eval.py:10: NumbaDeprecationWarning: The 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.
  def get_thresholds(scores: np.ndarray, num_gt, num_sample_pts=41):
/usr/local/python3.8/lib/python3.8/site-packages/torch_npu/contrib/transfer_to_npu.py:293: ImportWarning:
    *************************************************************************************************************
    The torch.Tensor.cuda and torch.nn.Module.cuda are replaced with torch.Tensor.npu and torch.nn.Module.npu now..
    The torch.cuda.DoubleTensor is replaced with torch.npu.FloatTensor cause the double type is not supported now..
    The backend in torch.distributed.init_process_group set to hccl now..
    The torch.cuda.* and torch.cuda.amp.* are replaced with torch.npu.* and torch.npu.amp.* now..
    The device parameters have been replaced with npu in the function below:
    torch.logspace, torch.randint, torch.hann_window, torch.rand, torch.full_like, torch.ones_like, torch.rand_like, torch.randperm, torch.arange, torch.frombuffer, torch.normal, torch._empty_per_channel_affine_quantized, torch.empty_strided, torch.empty_like, torch.scalar_tensor, torch.tril_indices, torch.bartlett_window, torch.ones, torch.sparse_coo_tensor, torch.randn, torch.kaiser_window, torch.tensor, torch.triu_indices, torch.as_tensor, torch.zeros, torch.randint_like, torch.full, torch.eye, torch._sparse_csr_tensor_unsafe, torch.empty, torch._sparse_coo_tensor_unsafe, torch.blackman_window, torch.zeros_like, torch.range, torch.sparse_csr_tensor, torch.randn_like, torch.from_file, torch._cudnn_init_dropout_state, torch._empty_affine_quantized, torch.linspace, torch.hamming_window, torch.empty_quantized, torch._pin_memory, torch.autocast, torch.load, torch.Generator, torch.set_default_device, torch.Tensor.new_empty, torch.Tensor.new_empty_strided, torch.Tensor.new_full, torch.Tensor.new_ones, torch.Tensor.new_tensor, torch.Tensor.new_zeros, torch.Tensor.to, torch.Tensor.pin_memory, torch.nn.Module.to, torch.nn.Module.to_empty
    *************************************************************************************************************

warnings.warn(msg, ImportWarning)
/usr/local/python3.8/lib/python3.8/site-packages/torch_npu/contrib/transfer_to_npu.py:250: RuntimeWarning: torch.jit.script and torch.jit.script_method will be disabled by transfer_to_npu, which currently does not support them, if you need to enable them, please do not use transfer_to_npu.
  warnings.warn(msg, RuntimeWarning)
projects.mmdet3d_plugin
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/MapTR/projects/mmdet3d_plugin/bevformer/modules/custom_base_transformer_layer.py:23: ImportWarning: ``MultiScaleDeformableAttention`` has been moved to ``mmcv.ops.multi_scale_deform_attn``, please change original path ``from mmcv.cnn.bricks.transformer import MultiScaleDeformableAttention`` to ``from mmcv.ops.multi_scale_deform_attn import MultiScaleDeformableAttention``
  warnings.warn(
Warning: please export TSAN_OPTIONS='ignore_noninstrumented_modules=1' to avoid false positive reports from the OpenMP runtime!
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/__init__.py:20: UserWarning: On January 1, 2023, MMCV will release v2.0.0, in which it will remove components related to the training process and add a data transformation module. In addition, it will rename the package names mmcv to mmcv-lite and mmcv-full to mmcv. See https://github.com/open-mmlab/mmcv/blob/master/docs/en/compatibility.md for more details.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmdetection3d/mmdet3d/core/evaluation/kitti_utils/eval.py:10: NumbaDeprecationWarning: The 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.
  def get_thresholds(scores: np.ndarray, num_gt, num_sample_pts=41):
/usr/local/python3.8/lib/python3.8/site-packages/torch_npu/contrib/transfer_to_npu.py:293: ImportWarning:
    *************************************************************************************************************
    The torch.Tensor.cuda and torch.nn.Module.cuda are replaced with torch.Tensor.npu and torch.nn.Module.npu now..
    The torch.cuda.DoubleTensor is replaced with torch.npu.FloatTensor cause the double type is not supported now..
    The backend in torch.distributed.init_process_group set to hccl now..
    The torch.cuda.* and torch.cuda.amp.* are replaced with torch.npu.* and torch.npu.amp.* now..
    The device parameters have been replaced with npu in the function below:
    torch.logspace, torch.randint, torch.hann_window, torch.rand, torch.full_like, torch.ones_like, torch.rand_like, torch.randperm, torch.arange, torch.frombuffer, torch.normal, torch._empty_per_channel_affine_quantized, torch.empty_strided, torch.empty_like, torch.scalar_tensor, torch.tril_indices, torch.bartlett_window, torch.ones, torch.sparse_coo_tensor, torch.randn, torch.kaiser_window, torch.tensor, torch.triu_indices, torch.as_tensor, torch.zeros, torch.randint_like, torch.full, torch.eye, torch._sparse_csr_tensor_unsafe, torch.empty, torch._sparse_coo_tensor_unsafe, torch.blackman_window, torch.zeros_like, torch.range, torch.sparse_csr_tensor, torch.randn_like, torch.from_file, torch._cudnn_init_dropout_state, torch._empty_affine_quantized, torch.linspace, torch.hamming_window, torch.empty_quantized, torch._pin_memory, torch.autocast, torch.load, torch.Generator, torch.set_default_device, torch.Tensor.new_empty, torch.Tensor.new_empty_strided, torch.Tensor.new_full, torch.Tensor.new_ones, torch.Tensor.new_tensor, torch.Tensor.new_zeros, torch.Tensor.to, torch.Tensor.pin_memory, torch.nn.Module.to, torch.nn.Module.to_empty
    *************************************************************************************************************

warnings.warn(msg, ImportWarning)
/usr/local/python3.8/lib/python3.8/site-packages/torch_npu/contrib/transfer_to_npu.py:250: RuntimeWarning: torch.jit.script and torch.jit.script_method will be disabled by transfer_to_npu, which currently does not support them, if you need to enable them, please do not use transfer_to_npu.
  warnings.warn(msg, RuntimeWarning)
2025-05-06 02:29:12,801 - mmdet - INFO - Environment info:
------------------------------------------------------------
sys.platform: linux
Python: 3.8.17 (default, Apr 29 2025, 07:31:52) [Clang 12.0.1 (openEuler 12.0.1-6.oe2203sp4 f4a7df2c51fa9eb3679555ef2de92c638ba
CUDA available: True
GPU 0: Ascend910B3
CUDA_HOME: None
GCC: clang version 12.0.1 (openEuler 12.0.1-6.oe2203sp4 f4a7df2c51fa9eb3679555ef2de92c638ba9f880)
PyTorch: 2.1.0a0+git7bcf7da
PyTorch compiling details: PyTorch built with:
  - GCC 4.2
  - C++ Version: 201703
  - clang 17.0.6
  - OpenMP 202011
  - NNPACK is enabled
  - CPU capability usage: NO AVX
  - Build settings: BUILD_TYPE=Release, CXX_COMPILER=/home/xinyi/tool/BiShengCompiler-4.1.0-aarch64-linux/bin/clang++, CXX_FLAGS=-flto=thin -fuse-ld=lld -D_GLIBCXX_USE_CXX11_ABI=0 -fvisibility-inlines-hidden -DUSE_PTHREADPOOL -DNDEBUG -DUSE_KINETO -DLIBKINETO_NOCUPTI -DLIBKINETO_NOROCTRACER -DUSE_QNNPACK -DUSE_PYTORCH_QNNPACK -DSYMBOLICATE_MOBILE_DEBUG_HANDLE -O2 -fPIC -Wall -Wextra -Werror=return-type -Werror=non-virtual-dtor -Werror=braced-scalar-init -Werror=range-loop-construct -Werror=bool-operation -Wnarrowing -Wno-missing-field-initializers -Wno-type-limits -Wno-array-bounds -Wno-unknown-pragmas -Wno-unused-parameter -Wno-unused-function -Wno-unused-result -Wno-strict-overflow -Wno-strict-aliasing -Wvla-extension -Wnewline-eof -Winconsistent-missing-override -Winconsistent-missing-destructor-override -Wno-range-loop-analysis -Wno-pass-failed -Wsuggest-override -Wno-error=pedantic -Wno-error=old-style-cast -Wno-error=inconsistent-missing-override -Wno-error=inconsistent-missing-destructor-override -Wconstant-conversion -Wno-invalid-partial-specialization -Wno-unused-private-field -Wno-missing-braces -Wunused-lambda-capture -Qunused-arguments -fcolor-diagnostics -faligned-new -Wno-unused-but-set-variable -fno-math-errno -fno-trapping-math -Werror=format, TORCH_DISABLE_GPU_ASSERTS=ON, TORCH_VERSION=2.1.0, USE_CUDA=OFF, USE_CUDNN=OFF, USE_EIGEN_FOR_BLAS=ON, USE_EXCEPTION_PTR=1, USE_GFLAGS=OFF, USE_GLOG=OFF, USE_MKL=OFF, USE_MKLDNN=OFF, USE_MPI=OFF, USE_NCCL=OFF, USE_NNPACK=ON, USE_OPENMP=ON, USE_ROCM=OFF,

TorchVision: 0.16.0
OpenCV: 4.11.0
MMCV: 1.7.2
MMCV Compiler: clang 12.0.1
MMCV CUDA Compiler: not available
MMDetection: 2.14.0
MMSegmentation: 0.14.1
MMDetection3D: 1.0.0rc4+0ff02cd
spconv2.0: False
------------------------------------------------------------

2025-05-06 02:29:14,649 - mmdet - INFO - Distributed training: True
2025-05-06 02:29:16,207 - mmdet - INFO - Config:
point_cloud_range = [-15.0, -30.0, -2.0, 15.0, 30.0, 2.0]
class_names = [
    'car', 'truck', 'construction_vehicle', 'bus', 'trailer', 'barrier',
    'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
]
dataset_type = 'CustomNuScenesLocalMapDataset'
data_root = 'data/nuscenes/'
input_modality = dict(
    use_lidar=False,
    use_camera=True,
    use_radar=False,
    use_map=False,
    use_external=True)
file_client_args = dict(backend='disk')
train_pipeline = [
    dict(type='LoadMultiViewImageFromFiles', to_float32=True),
    dict(type='PhotoMetricDistortionMultiViewImage'),
    dict(
        type='LoadAnnotations3D',
        with_bbox_3d=True,
        with_label_3d=True,
        with_attr_label=False),
    dict(
        type='ObjectRangeFilter',
        point_cloud_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0]),
    dict(
        type='ObjectNameFilter',
        classes=[
            'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
            'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
        ]),
    dict(
        type='NormalizeMultiviewImage',
        mean=[123.675, 116.28, 103.53],
        std=[58.395, 57.12, 57.375],
        to_rgb=True),
    dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
    dict(type='PadMultiViewImage', size_divisor=32),
    dict(
        type='DefaultFormatBundle3D',
        class_names=[
            'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
            'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
        ]),
    dict(type='CustomCollect3D', keys=['gt_bboxes_3d', 'gt_labels_3d', 'img'])
]
test_pipeline = [
    dict(type='LoadMultiViewImageFromFiles', to_float32=True),
    dict(
        type='NormalizeMultiviewImage',
        mean=[123.675, 116.28, 103.53],
        std=[58.395, 57.12, 57.375],
        to_rgb=True),
    dict(
        type='MultiScaleFlipAug3D',
        img_scale=(1600, 900),
        pts_scale_ratio=1,
        flip=False,
        transforms=[
            dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
            dict(type='PadMultiViewImage', size_divisor=32),
            dict(
                type='DefaultFormatBundle3D',
                class_names=[
                    'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
                    'barrier', 'motorcycle', 'bicycle', 'pedestrian',
                    'traffic_cone'
                ],
                with_label=False),
            dict(type='CustomCollect3D', keys=['img'])
        ])
]
eval_pipeline = [
    dict(
        type='LoadPointsFromFile',
        coord_type='LIDAR',
        load_dim=5,
        use_dim=5,
        file_client_args=dict(backend='disk')),
    dict(
        type='LoadPointsFromMultiSweeps',
        sweeps_num=10,
        file_client_args=dict(backend='disk')),
    dict(
        type='DefaultFormatBundle3D',
        class_names=[
            'car', 'truck', 'trailer', 'bus', 'construction_vehicle',
            'bicycle', 'motorcycle', 'pedestrian', 'traffic_cone', 'barrier'
        ],
        with_label=False),
    dict(type='Collect3D', keys=['points'])
]
data = dict(
    samples_per_gpu=4,
    workers_per_gpu=8,
    train=dict(
        type='CustomNuScenesLocalMapDataset',
        data_root='data/nuscenes/',
        ann_file='data/nuscenes/nuscenes_infos_temporal_train.pkl',
        pipeline=[
            dict(type='LoadMultiViewImageFromFiles', to_float32=True),
            dict(type='PhotoMetricDistortionMultiViewImage'),
            dict(
                type='LoadAnnotations3D',
                with_bbox_3d=True,
                with_label_3d=True,
                with_attr_label=False),
            dict(
                type='ObjectRangeFilter',
                point_cloud_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0]),
            dict(
                type='ObjectNameFilter',
                classes=[
                    'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
                    'barrier', 'motorcycle', 'bicycle', 'pedestrian',
                    'traffic_cone'
                ]),
            dict(
                type='NormalizeMultiviewImage',
                mean=[123.675, 116.28, 103.53],
                std=[58.395, 57.12, 57.375],
                to_rgb=True),
            dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
            dict(type='PadMultiViewImage', size_divisor=32),
            dict(
                type='DefaultFormatBundle3D',
                class_names=[
                    'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
                    'barrier', 'motorcycle', 'bicycle', 'pedestrian',
                    'traffic_cone'
                ]),
            dict(
                type='CustomCollect3D',
                keys=['gt_bboxes_3d', 'gt_labels_3d', 'img'])
        ],
        classes=[
            'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
            'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
        ],
        modality=dict(
            use_lidar=False,
            use_camera=True,
            use_radar=False,
            use_map=False,
            use_external=True),
        test_mode=False,
        box_type_3d='LiDAR',
        use_valid_flag=True,
        bev_size=(200, 100),
        pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
        fixed_ptsnum_per_line=20,
        eval_use_same_gt_sample_num_flag=True,
        padding_value=-10000,
        map_classes=['divider', 'ped_crossing', 'boundary'],
        queue_length=1,
        gt_shift_pts_pattern='v5'),
    val=dict(
        type='CustomNuScenesLocalMapDataset',
        ann_file='data/nuscenes/nuscenes_infos_temporal_val.pkl',
        pipeline=[
            dict(type='LoadMultiViewImageFromFiles', to_float32=True),
            dict(
                type='NormalizeMultiviewImage',
                mean=[123.675, 116.28, 103.53],
                std=[58.395, 57.12, 57.375],
                to_rgb=True),
            dict(
                type='MultiScaleFlipAug3D',
                img_scale=(1600, 900),
                pts_scale_ratio=1,
                flip=False,
                transforms=[
                    dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
                    dict(type='PadMultiViewImage', size_divisor=32),
                    dict(
                        type='DefaultFormatBundle3D',
                        class_names=[
                            'car', 'truck', 'construction_vehicle', 'bus',
                            'trailer', 'barrier', 'motorcycle', 'bicycle',
                            'pedestrian', 'traffic_cone'
                        ],
                        with_label=False),
                    dict(type='CustomCollect3D', keys=['img'])
                ])
        ],
        classes=[
            'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
            'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
        ],
        modality=dict(
            use_lidar=False,
            use_camera=True,
            use_radar=False,
            use_map=False,
            use_external=True),
        test_mode=True,
        box_type_3d='LiDAR',
        data_root='data/nuscenes/',
        map_ann_file='data/nuscenes/nuscenes_map_anns_val.json',
        bev_size=(200, 100),
        pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
        fixed_ptsnum_per_line=20,
        eval_use_same_gt_sample_num_flag=True,
        padding_value=-10000,
        gt_shift_pts_pattern='v5',
        map_classes=['divider', 'ped_crossing', 'boundary'],
        samples_per_gpu=1),
    test=dict(
        type='CustomNuScenesLocalMapDataset',
        data_root='data/nuscenes/',
        ann_file='data/nuscenes/nuscenes_infos_temporal_val.pkl',
        pipeline=[
            dict(type='LoadMultiViewImageFromFiles', to_float32=True),
            dict(
                type='NormalizeMultiviewImage',
                mean=[123.675, 116.28, 103.53],
                std=[58.395, 57.12, 57.375],
                to_rgb=True),
            dict(
                type='MultiScaleFlipAug3D',
                img_scale=(1600, 900),
                pts_scale_ratio=1,
                flip=False,
                transforms=[
                    dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
                    dict(type='PadMultiViewImage', size_divisor=32),
                    dict(
                        type='DefaultFormatBundle3D',
                        class_names=[
                            'car', 'truck', 'construction_vehicle', 'bus',
                            'trailer', 'barrier', 'motorcycle', 'bicycle',
                            'pedestrian', 'traffic_cone'
                        ],
                        with_label=False),
                    dict(type='CustomCollect3D', keys=['img'])
                ])
        ],
        classes=[
            'car', 'truck', 'construction_vehicle', 'bus', 'trailer',
            'barrier', 'motorcycle', 'bicycle', 'pedestrian', 'traffic_cone'
        ],
        modality=dict(
            use_lidar=False,
            use_camera=True,
            use_radar=False,
            use_map=False,
            use_external=True),
        test_mode=True,
        box_type_3d='LiDAR',
        map_ann_file='data/nuscenes/nuscenes_map_anns_val.json',
        bev_size=(200, 100),
        pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
        fixed_ptsnum_per_line=20,
        eval_use_same_gt_sample_num_flag=True,
        padding_value=-10000,
        gt_shift_pts_pattern='v5',
        map_classes=['divider', 'ped_crossing', 'boundary']),
    pin_memory=True,
    shuffler_sampler=dict(type='DistributedGroupSampler'),
    nonshuffler_sampler=dict(type='DistributedSampler'))
evaluation = dict(
    interval=2,
    pipeline=[
        dict(type='LoadMultiViewImageFromFiles', to_float32=True),
        dict(
            type='NormalizeMultiviewImage',
            mean=[123.675, 116.28, 103.53],
            std=[58.395, 57.12, 57.375],
            to_rgb=True),
        dict(
            type='MultiScaleFlipAug3D',
            img_scale=(1600, 900),
            pts_scale_ratio=1,
            flip=False,
            transforms=[
                dict(type='RandomScaleImageMultiViewImage', scales=[0.5]),
                dict(type='PadMultiViewImage', size_divisor=32),
                dict(
                    type='DefaultFormatBundle3D',
                    class_names=[
                        'car', 'truck', 'construction_vehicle', 'bus',
                        'trailer', 'barrier', 'motorcycle', 'bicycle',
                        'pedestrian', 'traffic_cone'
                    ],
                    with_label=False),
                dict(type='CustomCollect3D', keys=['img'])
            ])
    ],
    metric='chamfer')
checkpoint_config = dict(interval=1)
log_config = dict(
    interval=50,
    hooks=[dict(type='TextLoggerHook'),
           dict(type='TensorboardLoggerHook')])
dist_params = dict(backend='nccl')
log_level = 'INFO'
work_dir = './work_dirs/maptr_tiny_r50_24e_bevformer'
load_from = None
resume_from = None
workflow = [('train', 1)]
plugin = True
plugin_dir = 'projects/mmdet3d_plugin/'
voxel_size = [0.15, 0.15, 4]
img_norm_cfg = dict(
    mean=[123.675, 116.28, 103.53], std=[58.395, 57.12, 57.375], to_rgb=True)
map_classes = ['divider', 'ped_crossing', 'boundary']
fixed_ptsnum_per_gt_line = 20
fixed_ptsnum_per_pred_line = 20
eval_use_same_gt_sample_num_flag = True
num_map_classes = 3
_dim_ = 256
_pos_dim_ = 128
_ffn_dim_ = 512
_num_levels_ = 1
bev_h_ = 200
bev_w_ = 100
queue_length = 1
model = dict(
    type='MapTR',
    use_grid_mask=True,
    video_test_mode=False,
    pretrained=dict(img='ckpts/resnet50-19c8e357.pth'),
    img_backbone=dict(
        type='ResNet',
        depth=50,
        num_stages=4,
        out_indices=(3, ),
        frozen_stages=1,
        norm_cfg=dict(type='BN', requires_grad=False),
        norm_eval=True,
        style='pytorch'),
    img_neck=dict(
        type='FPN',
        in_channels=[2048],
        out_channels=256,
        start_level=0,
        add_extra_convs='on_output',
        num_outs=1,
        relu_before_extra_convs=True),
    pts_bbox_head=dict(
        type='MapTRHead',
        bev_h=200,
        bev_w=100,
        num_query=900,
        num_vec=50,
        num_pts_per_vec=20,
        num_pts_per_gt_vec=20,
        dir_interval=1,
        query_embed_type='instance_pts',
        transform_method='minmax',
        gt_shift_pts_pattern='v5',
        num_classes=3,
        in_channels=256,
        sync_cls_avg_factor=True,
        with_box_refine=True,
        as_two_stage=False,
        code_size=2,
        code_weights=[1.0, 1.0, 1.0, 1.0],
        transformer=dict(
            type='MapTRPerceptionTransformer',
            rotate_prev_bev=True,
            use_shift=True,
            use_can_bus=True,
            embed_dims=256,
            encoder=dict(
                type='BEVFormerEncoder',
                num_layers=1,
                pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
                num_points_in_pillar=4,
                return_intermediate=False,
                transformerlayers=dict(
                    type='BEVFormerLayer',
                    attn_cfgs=[
                        dict(
                            type='TemporalSelfAttention',
                            embed_dims=256,
                            num_levels=1),
                        dict(
                            type='SpatialCrossAttention',
                            pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
                            deformable_attention=dict(
                                type='MSDeformableAttention3D',
                                embed_dims=256,
                                num_points=8,
                                num_levels=1),
                            embed_dims=256)
                    ],
                    feedforward_channels=512,
                    ffn_dropout=0.1,
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm'))),
            decoder=dict(
                type='MapTRDecoder',
                num_layers=6,
                return_intermediate=True,
                transformerlayers=dict(
                    type='DetrTransformerDecoderLayer',
                    attn_cfgs=[
                        dict(
                            type='MultiheadAttention',
                            embed_dims=256,
                            num_heads=8,
                            dropout=0.1),
                        dict(
                            type='CustomMSDeformableAttention',
                            embed_dims=256,
                            num_levels=1)
                    ],
                    feedforward_channels=512,
                    ffn_dropout=0.1,
                    operation_order=('self_attn', 'norm', 'cross_attn', 'norm',
                                     'ffn', 'norm')))),
        bbox_coder=dict(
            type='MapTRNMSFreeCoder',
            post_center_range=[-20, -35, -20, -35, 20, 35, 20, 35],
            pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
            max_num=50,
            voxel_size=[0.15, 0.15, 4],
            num_classes=3),
        positional_encoding=dict(
            type='LearnedPositionalEncoding',
            num_feats=128,
            row_num_embed=200,
            col_num_embed=100),
        loss_cls=dict(
            type='FocalLoss',
            use_sigmoid=True,
            gamma=2.0,
            alpha=0.25,
            loss_weight=2.0),
        loss_bbox=dict(type='L1Loss', loss_weight=0.0),
        loss_iou=dict(type='GIoULoss', loss_weight=0.0),
        loss_pts=dict(type='PtsL1Loss', loss_weight=5.0),
        loss_dir=dict(type='PtsDirCosLoss', loss_weight=0.005)),
    train_cfg=dict(
        pts=dict(
            grid_size=[512, 512, 1],
            voxel_size=[0.15, 0.15, 4],
            point_cloud_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0],
            out_size_factor=4,
            assigner=dict(
                type='MapTRAssigner',
                cls_cost=dict(type='FocalLossCost', weight=2.0),
                reg_cost=dict(
                    type='BBoxL1Cost', weight=0.0, box_format='xywh'),
                iou_cost=dict(type='IoUCost', iou_mode='giou', weight=0.0),
                pts_cost=dict(type='OrderedPtsL1Cost', weight=5),
                pc_range=[-15.0, -30.0, -2.0, 15.0, 30.0, 2.0]))))
optimizer = dict(
    type='AdamW',
    lr=0.0006,
    paramwise_cfg=dict(custom_keys=dict(img_backbone=dict(lr_mult=0.1))),
    weight_decay=0.01)
optimizer_config = dict(grad_clip=dict(max_norm=35, norm_type=2))
lr_config = dict(
    policy='CosineAnnealing',
    warmup='linear',
    warmup_iters=500,
    warmup_ratio=0.3333333333333333,
    min_lr_ratio=0.001)
total_epochs = 24
runner = dict(type='EpochBasedRunner', max_epochs=24)
fp16 = dict(loss_scale=512.0)
gpu_ids = range(0, 1)

2025-05-06 02:29:16,207 - mmdet - INFO - Set random seed to 0, deterministic: True
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/MapTR/projects/mmdet3d_plugin/bevformer/modules/custom_base_transformer_layer.py:94: UserWarning: The arguments `feedforward_channels` in BaseTransformerLayer has been deprecated, now you should set `feedforward_channels` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/MapTR/projects/mmdet3d_plugin/bevformer/modules/custom_base_transformer_layer.py:94: UserWarning: The arguments `ffn_dropout` in BaseTransformerLayer has been deprecated, now you should set `ffn_drop` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/MapTR/projects/mmdet3d_plugin/bevformer/modules/custom_base_transformer_layer.py:94: UserWarning: The arguments `ffn_num_fcs` in BaseTransformerLayer has been deprecated, now you should set `num_fcs` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/cnn/bricks/transformer.py:690: DeprecationWarning: The arguments `feedforward_channels` in BaseTransformerLayer has been deprecated, now you should set `feedforward_channels` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/cnn/bricks/transformer.py:690: DeprecationWarning: The arguments `ffn_dropout` in BaseTransformerLayer has been deprecated, now you should set `ffn_drop` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/cnn/bricks/transformer.py:690: DeprecationWarning: The arguments `ffn_num_fcs` in BaseTransformerLayer has been deprecated, now you should set `num_fcs` and other FFN related arguments to a dict named `ffn_cfgs`.
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/cnn/bricks/transformer.py:440: DeprecationWarning: The arguments `dropout` in MultiheadAttention has been deprecated, now you can separately set `attn_drop`(float), proj_drop(float), and `dropout_layer`(dict)
  warnings.warn(
/home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmdetection3d/mmdet3d/models/detectors/mvx_two_stage.py:88: UserWarning: DeprecationWarning: pretrained is a deprecated key, please consider using init_cfg.
  warnings.warn('DeprecationWarning: pretrained is a deprecated '
> /home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/runner/base_module.py(79)init_weights()
-> for name, param in self.named_parameters():
(Pdb) c
!!!! pts_bbox_head.code_weights
!!!! pts_bbox_head.positional_encoding.row_embed.weight
!!!! pts_bbox_head.positional_encoding.col_embed.weight
!!!! pts_bbox_head.transformer.level_embeds
!!!! pts_bbox_head.transformer.cams_embeds
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.sampling_offsets.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.sampling_offsets.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.attention_weights.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.attention_weights.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.value_proj.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.value_proj.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.output_proj.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.0.output_proj.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.sampling_offsets.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.sampling_offsets.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.attention_weights.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.attention_weights.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.value_proj.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.deformable_attention.value_proj.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.0.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.0.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.1.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.1.bias
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.2.weight
!!!! pts_bbox_head.transformer.encoder.layers.0.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.0.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.1.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.2.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.3.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.4.norms.2.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.0.attn.in_proj_weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.0.attn.in_proj_bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.0.attn.out_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.0.attn.out_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.sampling_offsets.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.sampling_offsets.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.attention_weights.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.attention_weights.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.value_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.value_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.output_proj.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.attentions.1.output_proj.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.ffns.0.layers.0.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.ffns.0.layers.0.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.ffns.0.layers.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.ffns.0.layers.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.0.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.0.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.1.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.1.bias
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.2.weight
!!!! pts_bbox_head.transformer.decoder.layers.5.norms.2.bias
!!!! pts_bbox_head.transformer.reference_points.weight
!!!! pts_bbox_head.transformer.reference_points.bias
!!!! pts_bbox_head.transformer.can_bus_mlp.0.weight
!!!! pts_bbox_head.transformer.can_bus_mlp.0.bias
!!!! pts_bbox_head.transformer.can_bus_mlp.2.weight
!!!! pts_bbox_head.transformer.can_bus_mlp.2.bias
!!!! pts_bbox_head.transformer.can_bus_mlp.norm.weight
!!!! pts_bbox_head.transformer.can_bus_mlp.norm.bias
!!!! pts_bbox_head.cls_branches.0.0.weight
!!!! pts_bbox_head.cls_branches.0.0.bias
!!!! pts_bbox_head.cls_branches.0.1.weight
!!!! pts_bbox_head.cls_branches.0.1.bias
!!!! pts_bbox_head.cls_branches.0.3.weight
!!!! pts_bbox_head.cls_branches.0.3.bias
!!!! pts_bbox_head.cls_branches.0.4.weight
!!!! pts_bbox_head.cls_branches.0.4.bias
!!!! pts_bbox_head.cls_branches.0.6.weight
!!!! pts_bbox_head.cls_branches.0.6.bias
!!!! pts_bbox_head.cls_branches.1.0.weight
!!!! pts_bbox_head.cls_branches.1.0.bias
!!!! pts_bbox_head.cls_branches.1.1.weight
!!!! pts_bbox_head.cls_branches.1.1.bias
!!!! pts_bbox_head.cls_branches.1.3.weight
!!!! pts_bbox_head.cls_branches.1.3.bias
!!!! pts_bbox_head.cls_branches.1.4.weight
!!!! pts_bbox_head.cls_branches.1.4.bias
!!!! pts_bbox_head.cls_branches.1.6.weight
!!!! pts_bbox_head.cls_branches.1.6.bias
!!!! pts_bbox_head.cls_branches.2.0.weight
!!!! pts_bbox_head.cls_branches.2.0.bias
!!!! pts_bbox_head.cls_branches.2.1.weight
!!!! pts_bbox_head.cls_branches.2.1.bias
!!!! pts_bbox_head.cls_branches.2.3.weight
!!!! pts_bbox_head.cls_branches.2.3.bias
!!!! pts_bbox_head.cls_branches.2.4.weight
!!!! pts_bbox_head.cls_branches.2.4.bias
!!!! pts_bbox_head.cls_branches.2.6.weight
!!!! pts_bbox_head.cls_branches.2.6.bias
!!!! pts_bbox_head.cls_branches.3.0.weight
!!!! pts_bbox_head.cls_branches.3.0.bias
!!!! pts_bbox_head.cls_branches.3.1.weight
!!!! pts_bbox_head.cls_branches.3.1.bias
!!!! pts_bbox_head.cls_branches.3.3.weight
!!!! pts_bbox_head.cls_branches.3.3.bias
!!!! pts_bbox_head.cls_branches.3.4.weight
!!!! pts_bbox_head.cls_branches.3.4.bias
!!!! pts_bbox_head.cls_branches.3.6.weight
!!!! pts_bbox_head.cls_branches.3.6.bias
!!!! pts_bbox_head.cls_branches.4.0.weight
!!!! pts_bbox_head.cls_branches.4.0.bias
!!!! pts_bbox_head.cls_branches.4.1.weight
!!!! pts_bbox_head.cls_branches.4.1.bias
!!!! pts_bbox_head.cls_branches.4.3.weight
!!!! pts_bbox_head.cls_branches.4.3.bias
!!!! pts_bbox_head.cls_branches.4.4.weight
!!!! pts_bbox_head.cls_branches.4.4.bias
!!!! pts_bbox_head.cls_branches.4.6.weight
!!!! pts_bbox_head.cls_branches.4.6.bias
!!!! pts_bbox_head.cls_branches.5.0.weight
!!!! pts_bbox_head.cls_branches.5.0.bias
!!!! pts_bbox_head.cls_branches.5.1.weight
!!!! pts_bbox_head.cls_branches.5.1.bias
!!!! pts_bbox_head.cls_branches.5.3.weight
!!!! pts_bbox_head.cls_branches.5.3.bias
!!!! pts_bbox_head.cls_branches.5.4.weight
!!!! pts_bbox_head.cls_branches.5.4.bias
!!!! pts_bbox_head.cls_branches.5.6.weight
!!!! pts_bbox_head.cls_branches.5.6.bias
!!!! pts_bbox_head.reg_branches.0.0.weight
!!!! pts_bbox_head.reg_branches.0.0.bias
!!!! pts_bbox_head.reg_branches.0.2.weight
!!!! pts_bbox_head.reg_branches.0.2.bias
!!!! pts_bbox_head.reg_branches.0.4.weight
!!!! pts_bbox_head.reg_branches.0.4.bias
!!!! pts_bbox_head.reg_branches.1.0.weight
!!!! pts_bbox_head.reg_branches.1.0.bias
!!!! pts_bbox_head.reg_branches.1.2.weight
!!!! pts_bbox_head.reg_branches.1.2.bias
!!!! pts_bbox_head.reg_branches.1.4.weight
!!!! pts_bbox_head.reg_branches.1.4.bias
!!!! pts_bbox_head.reg_branches.2.0.weight
!!!! pts_bbox_head.reg_branches.2.0.bias
!!!! pts_bbox_head.reg_branches.2.2.weight
!!!! pts_bbox_head.reg_branches.2.2.bias
!!!! pts_bbox_head.reg_branches.2.4.weight
!!!! pts_bbox_head.reg_branches.2.4.bias
!!!! pts_bbox_head.reg_branches.3.0.weight
!!!! pts_bbox_head.reg_branches.3.0.bias
!!!! pts_bbox_head.reg_branches.3.2.weight
!!!! pts_bbox_head.reg_branches.3.2.bias
!!!! pts_bbox_head.reg_branches.3.4.weight
!!!! pts_bbox_head.reg_branches.3.4.bias
!!!! pts_bbox_head.reg_branches.4.0.weight
!!!! pts_bbox_head.reg_branches.4.0.bias
!!!! pts_bbox_head.reg_branches.4.2.weight
!!!! pts_bbox_head.reg_branches.4.2.bias
!!!! pts_bbox_head.reg_branches.4.4.weight
!!!! pts_bbox_head.reg_branches.4.4.bias
!!!! pts_bbox_head.reg_branches.5.0.weight
!!!! pts_bbox_head.reg_branches.5.0.bias
!!!! pts_bbox_head.reg_branches.5.2.weight
!!!! pts_bbox_head.reg_branches.5.2.bias
!!!! pts_bbox_head.reg_branches.5.4.weight
!!!! pts_bbox_head.reg_branches.5.4.bias
!!!! pts_bbox_head.bev_embedding.weight
> /home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv/mmcv/runner/base_module.py(84)init_weights()
-> 'init_info'] = f'The value is the same before and ' \
(Pdb) ll
 56         def init_weights(self) -> None:
 57             """Initialize the weights."""
 58
 59             is_top_level_module = False
 60             # check if it is top-level module
 61             if not hasattr(self, '_params_init_info'):
 62                 # The `_params_init_info` is used to record the initialization
 63                 # information of the parameters
 64                 # the key should be the obj:`nn.Parameter` of model and the value
 65                 # should be a dict containing
 66                 # - init_info (str): The string that describes the initialization.
 67                 # - tmp_mean_value (FloatTensor): The mean of the parameter,
 68                 #       which indicates whether the parameter has been modified.
 69                 # this attribute would be deleted after all parameters
 70                 # is initialized.
 71                 self._params_init_info: defaultdict = defaultdict(dict)
 72                 is_top_level_module = True
 73
 74                 # Initialize the `_params_init_info`,
 75                 # When detecting the `tmp_mean_value` of
 76                 # the corresponding parameter is changed, update related
 77                 # initialization information
 78                 import pdb;pdb.set_trace()
 79                 for name, param in self.named_parameters():
 80                     print("!!!!", name)
 81                     if name == "pts_bbox_head.bev_embedding.weight":
 82                         import pdb;pdb.set_trace()
 83                     self._params_init_info[param][
 84  ->                     'init_info'] = f'The value is the same before and ' \
 85                                        f'after calling `init_weights` ' \
 86                                        f'of {self.__class__.__name__} '
 87                     self._params_init_info[param][
 88                         'tmp_mean_value'] = param.data.mean()
 89
 90                 # pass `params_init_info` to all submodules
 91                 # All submodules share the same `params_init_info`,
 92                 # so it will be updated when parameters are
 93                 # modified at any level of the model.
 94                 for sub_module in self.modules():
 95                     sub_module._params_init_info = self._params_init_info
 96
 97             # Get the initialized logger, if not exist,
 98             # create a logger named `mmcv`
 99             logger_names = list(logger_initialized.keys())
100             logger_name = logger_names[0] if logger_names else 'mmcv'
101
102             from ..cnn import initialize
103             from ..cnn.utils.weight_init import update_init_info
104             module_name = self.__class__.__name__
105             if not self._is_init:
106                 if self.init_cfg:
107                     print_log(
108                         f'initialize {module_name} with init_cfg {self.init_cfg}',
109                         logger=logger_name)
110                     initialize(self, self.init_cfg)
111                     if isinstance(self.init_cfg, dict):
112                         # prevent the parameters of
113                         # the pre-trained model
114                         # from being overwritten by
115                         # the `init_weights`
116                         if self.init_cfg['type'] == 'Pretrained':
117                             return
118
119                 for m in self.children():
120                     if hasattr(m, 'init_weights'):
121                         m.init_weights()
122                         # users may overload the `init_weights`
123                         update_init_info(
124                             m,
125                             init_info=f'Initialized by '
126                             f'user-defined `init_weights`'
127                             f' in {m.__class__.__name__} ')
128
129                 self._is_init = True
130             else:
131                 warnings.warn(f'init_weights of {self.__class__.__name__} has '
132                               f'been called more than once.')
133
134             if is_top_level_module:
135                 self._dump_init_info(logger_name)
136
137                 for sub_module in self.modules():
138                     del sub_module._params_init_info
(Pdb) param.data
tensor([[ 0.3346, -0.8828, -0.1136,  ...,  0.8826, -1.8123,  0.8151],
        [-0.5951, -1.4626,  0.8475,  ...,  1.4519,  2.5579, -0.2043],
        [-0.6746,  0.0264,  0.7649,  ...,  1.3184, -0.7278,  0.0154],
        ...,
        [ 0.5182,  1.1863,  0.4237,  ...,  2.0670,  0.0473,  1.8927],
        [-0.1231, -1.1684,  1.9254,  ..., -0.5681, -0.6239, -0.5316],
        [ 1.1549,  0.5097, -0.6489,  ..., -1.0827,  0.0209, -1.4871]])
(Pdb) param.data[0]
tensor([ 0.3346, -0.8828, -0.1136,  0.3244, -0.3209,  0.7281,  0.2600,  1.0407,
         0.3501, -0.7894, -0.2642,  0.0256,  0.4568, -0.7666,  0.5501,  0.6672,
         0.9547, -1.5918, -0.5885,  0.6217, -1.7893,  0.5015,  1.3663,  0.5167,
        -0.8397,  1.2157, -0.2112, -0.1900, -1.1100, -0.1835,  0.1395, -0.1162,
        -0.0241, -0.2620, -1.4265, -3.5514, -1.0383,  0.5434,  0.6944, -0.4704,
        -0.6737, -0.3197,  2.0386,  0.9962,  0.3623,  0.2161,  0.1009,  0.4102,
        -0.5804,  0.9949, -0.9284, -1.5280, -0.6673,  1.0168,  0.2570, -0.0159,
         0.0816, -0.0709,  1.8976,  0.8071, -2.1896, -0.2846,  0.7351, -0.0122,
         0.4719, -1.0442,  1.3061, -0.6122, -1.7751,  0.0140, -0.1835, -1.3341,
        -0.4490, -0.8574, -0.5970, -0.6065, -0.8981, -0.4639, -0.3914,  0.9162,
         0.5757, -0.2449,  0.2926,  0.3605, -0.8455,  0.1431,  1.4882,  1.0701,
         0.4523, -0.5384,  0.7856,  0.5376,  0.5866,  0.3177,  0.3368,  0.5757,
        -0.4450, -0.4648, -0.2667, -0.4380, -0.6291,  1.0922,  0.8174,  1.6219,
         0.1084, -1.0030,  1.0718,  0.0233,  0.3976, -0.2483,  0.1274,  1.2697,
         1.7519,  0.5526, -0.0835, -0.2765,  1.5941,  0.1815, -0.9194, -0.3283,
         0.7386,  1.2244, -1.5840,  0.5081, -0.4533, -0.3685, -0.7840, -2.2048,
        -1.2149, -0.1261,  0.4236, -0.5035,  0.7673,  3.0645,  1.5516, -1.0794,
        -0.9831, -0.1259, -0.0342, -0.9783, -0.9835,  0.4927,  0.2750,  0.0925,
         0.9347, -1.0328,  0.2586, -0.1690,  1.6564, -1.0191, -0.2295, -0.4965,
        -1.5436, -0.4743, -0.3224, -0.6014, -1.1098, -0.4344,  0.6232,  0.2970,
        -0.6383,  0.2032,  1.4624, -0.1955, -0.9351,  0.8184, -1.6618, -0.3535,
         0.5543,  0.6959, -0.0045,  0.5455,  0.2708,  0.3808, -0.8948, -1.1259,
         0.2557, -0.2562,  1.0773,  0.0700,  0.9273, -0.2175, -0.6563, -0.1454,
         1.2647, -0.9731,  0.5140,  0.2004, -1.2946, -0.9143,  0.2443, -0.6664,
        -1.4068,  1.9337, -1.9506, -0.0645,  1.2034, -0.9366, -1.4137,  0.5090,
        -1.4954,  0.3777,  0.1271,  1.4837, -2.3799, -0.7307,  0.8735,  1.1964,
         0.3695, -0.0382,  0.1699, -1.2928,  0.7482, -0.4217,  0.4145, -0.2290,
         0.8246,  0.6280, -0.9238,  0.9698,  2.5129, -0.3167, -2.5102, -0.5279,
         1.8864,  0.6001,  0.8151,  1.2361,  0.1404, -0.0159,  0.7300,  0.5005,
         0.9352,  2.0131, -1.9003, -0.3568, -1.5779, -0.1438,  0.4502, -0.3725,
        -0.8348, -1.3513, -2.3635,  1.6701,  0.4200,  1.7095, -0.5624,  0.4635,
         0.5312,  0.5178, -0.8639,  1.3752, -0.6619,  0.8826, -1.8123,  0.8151])
(Pdb) param.data[0].max()
tensor(3.0645)
(Pdb) param.data[0].mean()
tensor(-0.0098)
(Pdb) param.data[1].mean()
tensor(-0.0044)
(Pdb) param.data.mean()

二、软件版本:
-- CANN 版本 CANN8.1 使用的官方镜像
--Tensorflow/Pytorch/MindSpore 版本:torch-2.1.0使用毕昇编译器编译
--Python 版本 (e.g., Python 3.7.5):python3.8使用毕昇编译器编译
--操作系统版本 (e.g., Ubuntu 18.04):Linux version 5.10.0-60.18.0.50.oe2203.aarch64 (abuild@obs-worker-002) (gcc_old (GCC) 10.3.1, GNU ld (GNU Binutils) 2.37) #1 SMP Wed Mar 30 02:43:08 UTC 2022

[root@fedf3901b7b1 MapTR]# pip list
Package                 Version                 Editable project location
----------------------- ----------------------- -----------------------------------------------------------------------
absl-py                 2.2.2
addict                  2.4.0
argcomplete             3.6.2
asttokens               3.0.0
attrs                   25.3.0
auto_tune               0.1.0
av                      12.3.0
av2                     0.2.1
backcall                0.2.0
black                   24.8.0
cachetools              5.5.2
certifi                 2025.4.26
charset-normalizer      3.4.1
click                   8.1.8
cloudpickle             3.1.1
colorlog                6.9.0
cycler                  0.12.1
dataflow                0.0.1
decorator               5.2.1
dependency-groups       1.3.0
descartes               1.1.0
distlib                 0.3.9
exceptiongroup          1.2.2
executing               2.2.0
filelock                3.16.1
fire                    0.7.0
flake8                  7.1.2
fonttools               4.57.0
fsspec                  2023.6.0
google-auth             2.39.0
google-auth-oauthlib    1.0.0
grpcio                  1.70.0
hccl                    0.1.0
hccl_parser             0.1
huggingface-hub         0.30.2
idna                    3.10
imageio                 2.35.1
importlib_metadata      8.5.0
iniconfig               2.1.0
ipython                 8.12.3
jedi                    0.19.2
Jinja2                  3.1.6
joblib                  1.4.2
kiwisolver              1.4.7
kornia                  0.7.3
kornia_rs               0.1.8
llm_datadist            0.0.1
llvmlite                0.41.1
lyft-dataset-sdk        0.0.8
Markdown                3.7
markdown-it-py          3.0.0
MarkupSafe              2.1.5
matplotlib              3.5.3
matplotlib-inline       0.1.7
mccabe                  0.7.0
mdurl                   0.1.2
ml-dtypes               0.2.0
mmcv-full               1.7.2                   /home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmcv
mmdet                   2.14.0
mmdet3d                 1.0.0rc4                /home/xinyi/2025/MapTR-V2/DrivingSDK/model_examples/MapTR/mmdetection3d
mmsegmentation          0.14.1
mock                    5.2.0
mpmath                  1.3.0
msobjdump               0.1.0
mx_driving              1.0.0+git0ff02cd
mypy_extensions         1.1.0
narwhals                1.37.0
networkx                2.2
ninja                   1.11.1.4
nox                     2025.2.9
numba                   0.58.1
numpy                   1.23.0
nuscenes-devkit         1.1.11
oauthlib                3.2.2
op_compile_tool         0.1.0
op_gen                  0.1
op_test_frame           0.1
opc_tool                0.1.0
opencv-python           4.11.0.86
packaging               25.0
pandas                  2.0.3
parso                   0.8.4
pathspec                0.12.1
pexpect                 4.9.0
pickleshare             0.7.5
pillow                  10.4.0
pip                     22.3.1
platformdirs            4.3.6
plotly                  6.0.1
pluggy                  1.5.0
plyfile                 1.0.3
polars                  1.8.2
prettytable             3.11.0
projects                1.0.9
prompt_toolkit          3.0.51
protobuf                4.21.6
psutil                  7.0.0
ptyprocess              0.7.0
pure_eval               0.2.3
pyarrow                 17.0.0
pyasn1                  0.6.1
pyasn1_modules          0.4.2
pycocotools             2.0.7
pycodestyle             2.12.1
pyflakes                3.2.0
Pygments                2.19.1
pyparsing               3.1.4
pyproj                  3.5.0
pyquaternion            0.9.9
pytest                  8.3.5
python-dateutil         2.9.0.post0
pytz                    2025.2
PyWavelets              1.4.1
PyYAML                  6.0.2
requests                2.32.3
requests-oauthlib       2.0.0
rich                    14.0.0
rsa                     4.9.1
safetensors             0.5.3
schedule_search         0.0.1
scikit-image            0.19.3
scikit-learn            1.3.2
scipy                   1.10.1
setuptools              75.1.0
Shapely                 1.8.5.post1
show_kernel_debug_data  0.1.0
six                     1.17.0
stack-data              0.6.3
sympy                   1.13.3
synr                    0.5.0
te                      0.4.0
tensorboard             2.14.0
tensorboard-data-server 0.7.2
termcolor               2.4.0
terminaltables          3.1.10
threadpoolctl           3.5.0
tifffile                2023.7.10
timm                    1.0.15
tomli                   2.2.1
torch                   2.1.0a0+git7bcf7da
torch-npu               2.1.0.post13+gitcce8403
torchaudio              2.1.0
torchvision             0.16.0
tornado                 6.4.2
tqdm                    4.67.1
traitlets               5.14.3
trimesh                 2.35.39
typing_extensions       4.13.2
tzdata                  2025.2
universal_pathlib       0.2.6
urllib3                 2.2.3
urwid                   2.6.16
virtualenv              20.30.0
wcwidth                 0.2.13
Werkzeug                3.0.6
wheel                   0.45.1
yapf                    0.43.0
zipp                    3.20.2

GVP Ascend/DrivingSDK
暂停

内容风险标识

评论 (17)

GVPAscend/DrivingSDK暂停

内容风险标识