Batch_normalize

[pyqpyqpyqpyq789]

 File "train.py", line 363, in main
    res = train(opt=options,
  File "train.py", line 175, in train
    model_output = model(x)
  File "/home/user/anaconda3/envs/pyq38/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/devdata/pyq/DASTM-main/protonet.py", line 196, in forward
    x = self.model(x)
  File "/home/user/anaconda3/envs/pyq38/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/devdata/pyq/DASTM-main/mmskl/st_gcn_aaai18.py", line 124, in forward
    x = self.data_bn(x)
  File "/home/user/anaconda3/envs/pyq38/lib/python3.8/site-packages/torch/nn/modules/module.py", line 727, in _call_impl
    result = self.forward(*input, **kwargs)
  File "/home/user/anaconda3/envs/pyq38/lib/python3.8/site-packages/torch/nn/modules/batchnorm.py", line 131, in forward
    return F.batch_norm(
  File "/home/user/anaconda3/envs/pyq38/lib/python3.8/site-packages/torch/nn/functional.py", line 2056, in batch_norm
    return torch.batch_norm(
RuntimeError: running_mean should contain 54 elements not 75

[NingMa-AI]

Hi， thanks for your attention to our work. I can figure out what happens for your reproducing. Could you please provide more details? For example, do you check again for the environment and the command?

[pyqpyqpyqpyq789]

Hi， thanks for your attention to our work. I can figure out what happens for your reproducing. Could you please provide more details? For example, do you check again for the environment and the command?

It is possible that this problem is caused by the misuse of the 18-keypoint dataset and the 25-keypoint dataset.

[pyqpyqpyqpyq789]

尊敬的马老师， 您好！ 十分感谢您杰出的、富有启发性的工作。 先前的问题已解决。 关于DASTM，我还有两个新的问题想请教您， 问题1：nturgbd_dataset.py文件中，第37行segment = 30代表什么？是NTU-T数据集有30个类别的意思吗？还是说从300帧中抽样了30帧？ 问题2：原文“The explainability of spatial activation.”中讲到，通过绘制公式4中Softmax的输出获得可解释性。我发现cross_attention.py文件第68行Softmax输出的是一个大小为 (查询数150, 1, 关节数, 关节数) 的张量，请问在绘制热图过程中，应当怎样处理 “查询数150” 这个维度？ 还望您不吝赐教！ 祝您工作顺利，身体健康！

庞逸群 敬上

------------------ 原始邮件 ------------------ 发件人: "NingMa-AI/DASTM" @.>; 发送时间: 2022年10月4日(星期二) 上午10:45 @.>; @.@.>; 主题: Re: [NingMa-AI/DASTM] Batch_normalize (Issue #3)

Hi， thanks for your attention to our work. I can figure out what happens for your reproducing. Could you please provide more details? For example, do you check again for the environment and the command?

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