When using the parameter fp16=True, the training error is as follows：
fp16 training is a commonly used method to save video memory overhead. How do I modify the code?
warnings.warn(warning.format(ret))
0%| | 0/300000 [00:06<?, ?it/s]
Traceback (most recent call last):
File "/home/zzg/workspace/pycharm/HR-VITON/train_condition.py", line 497, in
main()
File "workspace/pycharm/HR-VITON/train_condition.py", line 488, in main
train(opt, train_loader, val_loader, test_loader, board, tocg, D)
File "workspace/pycharm/HR-VITON/train_condition.py", line 280, in train
loss_G.backward()
File "miniconda3/envs/py310_DL_cu118/lib/python3.10/site-packages/torch/_tensor.py", line 487, in backward
torch.autograd.backward(
File "miniconda3/envs/py310_DL_cu118/lib/python3.10/site-packages/torch/autograd/init.py", line 200, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
RuntimeError: Found dtype Half but expected Float
When using the parameter fp16=True, the training error is as follows： fp16 training is a commonly used method to save video memory overhead. How do I modify the code?
warnings.warn(warning.format(ret)) 0%| | 0/300000 [00:06<?, ?it/s] Traceback (most recent call last): File "/home/zzg/workspace/pycharm/HR-VITON/train_condition.py", line 497, in
main()
File "workspace/pycharm/HR-VITON/train_condition.py", line 488, in main
train(opt, train_loader, val_loader, test_loader, board, tocg, D)
File "workspace/pycharm/HR-VITON/train_condition.py", line 280, in train
loss_G.backward()
File "miniconda3/envs/py310_DL_cu118/lib/python3.10/site-packages/torch/_tensor.py", line 487, in backward
torch.autograd.backward(
File "miniconda3/envs/py310_DL_cu118/lib/python3.10/site-packages/torch/autograd/init.py", line 200, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
RuntimeError: Found dtype Half but expected Float
Process finished with exit code 1