When training to the second epoch, encounter "AssertionError: scalar should be 0D"

[d-serafly]

Hi Yisheng,

   There are somethings wrong when I train the Linemod dataset. Specifically，When training to the second epoch，the program will stop training and encounter the following error code. I hope to get your help，thanks a lot. Best wish.

At the first epoch, everything looks normal. ####################################################### train_dataset_size: 186 cls_id in lm_dataset.py 1 cls_id in lm_dataset.py 1 test_dataset_size: 1050 test_dataset_size: 1050 loading resnet34 pretrained mdl. loading resnet34 pretrained mdl. local_rank: 0 local_rank: 1 Selected optimization level O0: Pure FP32 training.

Defaults for this optimization level are: enabled : True opt_level : O0 cast_model_type : torch.float32 patch_torch_functions : False keep_batchnorm_fp32 : None master_weights : False loss_scale : 1.0 Processing user overrides (additional kwargs that are not None)... After processing overrides, optimization options are: enabled : True opt_level : O0 cast_model_type : torch.float32 patch_torch_functions : False keep_batchnorm_fp32 : None master_weights : False loss_scale : 1.0 /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/_reduction.py:43: UserWarning: size_average and reduce args will be deprecated, please use reduction='mean' instead. warnings.warn(warning.format(ret)) Totally train 775 iters per gpu. ape_epochs: 0%| | 0/25 [00:00<?, ?it/s]/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/_reduction.py:43: UserWarning: size_average and reduce args will be deprecated, please use reduction='mean' instead. warnings.warn(warning.format(ret)) Totally train 775 iters per gpu. ape_epochs: 0%| | 0/25 [00:00<?, ?it/s]kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt [00:00<?, ?it/s] kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/functional.py:2416: UserWarning: nn.functional.upsample is deprecated. Use nn.functional.interpolate instead. warnings.warn("nn.functional.upsample is deprecated. Use nn.functional.interpolate instead.") /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/functional.py:2506: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details. "See the documentation of nn.Upsample for details.".format(mode)) /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/functional.py:2416: UserWarning: nn.functional.upsample is deprecated. Use nn.functional.interpolate instead. warnings.warn("nn.functional.upsample is deprecated. Use nn.functional.interpolate instead.") /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/functional.py:2506: UserWarning: Default upsampling behavior when mode=bilinear is changed to align_corners=False since 0.4.0. Please specify align_corners=True if the old behavior is desired. See the documentation of nn.Upsample for details. "See the documentation of nn.Upsample for details.".format(mode)) /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/modules/container.py:100: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument. input = module(input) /home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/nn/modules/container.py:100: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument. input = module(input) /home/extend/gy/pyproject/FFB6D/ffb6d/models/loss.py:30: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument. logpt = F.log_softmax(input) /home/extend/gy/pyproject/FFB6D/ffb6d/models/loss.py:30: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument. logpt = F.log_softmax(input) ape_epochs: 4%|\u2588\u258e | 1/25 [00:52<20:48, 52.02s/it]kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt/it, total_it=31] kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt/it, total_it=31] kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt ########################################################### But complete the second training epoch, it will stop training. Like this, ########################################################### ape_epochs: 8%|\u2588\u2588\u258c | 2/25 [01:30<16:57, 44.25s/it]kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt/it, total_it=62] kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt/it, total_it=61] kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt ape_epochs: 8%|\u2588\u2588\u258c kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txtit [00:00, ?it/s]
kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txtkps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt kps_pth in get_kps: datasets/linemod/kps_orb9_fps/ape_8_kps.txt loss_rgbd_seg 0.11550371497869491 loss_kp_of 11.04855725969587 loss_ctr_of 1.3788805430276054 loss_all 12.658445249285016 loss_target 12.658445249285016 acc_rgbd 84.9538414137704 ape_epochs: 8%|\u2588\u2588\u258c | 2/25 [02:42<31:09, 81.28s/it] Traceback (most recent call last): File "train_lm.py", line 697, in train() File "train_lm.py", line 688, in train clr_div=clr_div File "train_lm.py", line 518, in train val_loss, res = self.eval_epoch(test_loader, it=it) File "train_lm.py", line 420, in eval_epoch writer.add_scalars('val_acc', acc_dict, it) File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/tensorboardX/writer.py", line 490, in add_scalars fw.add_summary(scalar(main_tag, scalar_value), File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/tensorboardX/summary.py", line 153, in scalar assert(scalar.squeeze().ndim == 0), 'scalar should be 0D' AssertionError: scalar should be 0D Traceback (most recent call last): File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/runpy.py", line 193, in _run_module_as_main "main", mod_spec) File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/runpy.py", line 85, in _run_code exec(code, run_globals) File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/distributed/launch.py", line 263, in main() File "/home/extend/gy/miniconda3/envs/FFB/lib/python3.7/site-packages/torch/distributed/launch.py", line 259, in main cmd=cmd) subprocess.CalledProcessError: Command '['/home/extend/gy/miniconda3/envs/FFB/bin/python3', '-u', 'train_lm.py', '--local_rank=1', '--gpus=2', '--cls=ape']' returned non-zero exit status 1. ######################################################################## #48 The error seems to be due to the version of tensorboardX? or something else. Please give me some advises.

[YanhuiDua]

Hi, I have meet the same question as you. If you solve this problem, please give me some advices.

[d-serafly]

Hi, I have meet the same question as you. If you solve this problem, please give me some advices.

I have tried to lower the version of python and tensorboard, but it doesn't work. On the other way, since "write_scalar" only plays the role of saving data for display on the tensorboard. It does not affect training, so I commenting that. There is everything normal until the training stage，I hope it can help you.

[lin-fangzhou]

Have you solved this problem? Requesting help, I have also encountered this issue

[YanhuiDua]

看之前的回答应该是不影响训练精度，可以不用管

At 2023-05-06 23:03:36, "lin-fangzhou" @.***> wrote:

Have you solved this problem? Requesting help, I have also encountered this issue

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[hs475]

你把acc_dict输出，其中acc_rgbd不是零维，我不太确定是不是可以取平均，然后这样就是零维可以正常运行