lirongwu/gcml

Code for WACV 2022 paper "Generalized Clustering and Multi-Manifold Learning with Geometric Structure Preservation"

gcml's Issues

Training/Inference on custom dataset

Hi,

Are there any instructions for training and/or infering a custom data on your model?

RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation

There is a non-obvious inplace operation on the tensor loss_com. Consequently, I am unable to replicate your results. Suggestions to fix?

File "/home/eich467/AHA-ToDo/GCML/main.py", line 227, in
train(model=dec, num_epochs=param['dec_epochs'], path="./plots/{}/".format(param['name']))
File "/home/eich467/AHA-ToDo/GCML/main.py", line 115, in train
loss_com.backward(retain_graph=True)
File "/data/eich467/miniconda3/envs/GCML/lib/python3.10/site-packages/torch/_tensor.py", line 488, in backward
torch.autograd.backward(
File "/data/eich467/miniconda3/envs/GCML/lib/python3.10/site-packages/torch/autograd/init.py", line 197, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
RuntimeError: one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [2000, 10]], which is output 0 of AsStridedBackward0, is at version 3017; expected version 2743 instead. Hint: the backtrace further above shows the operation that failed to compute its gradient. The variable in question was changed in there or anywhere later.