about terra_incognita dataset
shuangliumax opened this issue · comments
Hello, when I downloaded the "terra_incognita" dataset according to the download.py , I tried to test it , but such an error occurred. RuntimeError: CUDA error: device-side assert triggered
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [0,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [2,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [3,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [4,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [5,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [6,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [7,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [8,0,0] Assertion t >= 0 && t < n_classes failed.
/opt/conda/conda-bld/pytorch_1607370131125/work/aten/src/THCUNN/ClassNLLCriterion.cu:108: cunn_ClassNLLCriterion_updateOutput_kernel: block: [0,0,0], thread: [9,`
this error I checked may be the problem of label. But I don't know how to correct it. Do you have any good suggestions? Thank you!
The code works well in my environment. I agree with the error is raised related to the label, so I recommend to check the labels are loaded properly (in TerraIncognita, # of classes = 10) and the model output is also shaped properly.
Ok, thanks for your reply. I'll check again