where is .yaml file?

[AaddX-ai]

I cannot find the yaml when I run the gen the template:
python run.py --mode test --ttype ALL --dataset /mnt/sunxiaogang/RealNet/data/MVTec-AD/mvtec
/mnt/sunxiaogang/miniconda3/envs/hetmm/lib/python3.9/site-packages/torch/utils/cpp_extension.py:1964: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.
If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
warnings.warn(
Traceback (most recent call last):
File "/mnt/sunxiaogang/HETMM/run.py", line 80, in
cfg = Cfg(args)
File "/mnt/sunxiaogang/HETMM/src/config.py", line 19, in init
with open(os.path.join('configs', f'{args.dataset}.yaml'), 'r') as dfile:
FileNotFoundError: [Errno 2] No such file or directory: '/mnt/sunxiaogang/RealNet/data/MVTec-AD/mvtec.yaml'

[NarcissusEx]

Hello, thanks for your attentions. The previous version needs to replace the datapath in the "configs/MVTec_AD.yaml" file and then run the command: python run.py --mode test --ttype ALL --dataset MVTec_AD, which is so inconvenient for evaluation. Thus, we optimize the implementation and add an argument "--datapath" in "run.py". Please update the code and see the scripts in "run.sh" file in details.


[AaddX-ai]

thank u for your patience! but I get one new error:
python run.py --mode test --ttype PTS --tsize 60 --dataset MVTec_AD --datapath=/mnt/sunxiaogang/RealNet/data/MVTec-AD/mvtec
/mnt/sunxiaogang/miniconda3/envs/hetmm/lib/python3.9/site-packages/torch/utils/cpp_extension.py:1964: UserWarning: TORCH_CUDA_ARCH_LIST is not set, all archs for visible cards are included for compilation.
If this is not desired, please set os.environ['TORCH_CUDA_ARCH_LIST'].
warnings.warn(
carpet: 0%| | 0/1 [00:00<?, ?it/s]/mnt/sunxiaogang/miniconda3/envs/hetmm/lib/python3.9/site-packages/torch/functional.py:534: UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at /opt/conda/conda-bld/pytorch_1729647350383/work/aten/src/ATen/native/TensorShape.cpp:3595.)
return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined]
carpet: 0%| | 0/1 [00:01<?, ?it/s]
Traceback (most recent call last):
File "/mnt/sunxiaogang/HETMM/run.py", line 84, in
globals()args.mode
File "/mnt/sunxiaogang/HETMM/run.py", line 37, in test
score = cfg.model.impl(query, cfg.temp)
File "/mnt/sunxiaogang/HETMM/src/models/atmm.py", line 105, in impl
fores = self.match(super().impl(querys, temps, self.fore_patches), self.fore_patches, forward_ATM)
File "/mnt/sunxiaogang/HETMM/src/models/atmm.py", line 79, in match
return [self.upsample(1 - func(q, t, r).amax(dim=1).reshape(q.shape[0], 1, *q.shape[-2:])) for (q, t), r in zip(feas, patches)]
File "/mnt/sunxiaogang/HETMM/src/models/atmm.py", line 79, in
return [self.upsample(1 - func(q, t, r).amax(dim=1).reshape(q.shape[0], 1, *q.shape[-2:])) for (q, t), r in zip(feas, patches)]
IndexError: Dimension out of range (expected to be in range of [-3, 2], but got 3)

[NarcissusEx]

I have never seen this bug before. The code in my machine works well. I guess the problem may be caused by the invalid/unexpected dimension of input features, i.e., q and t. You can print their shape to locate the reason.