TCNN backend does not work
Opened this issue · 2 comments
evl-synth commented
I get the following error, when trying to train with the tcnn backend. I have PyTorch 2.0.1+cu118, Python 3.10, tinycudann 1.7. I've tried all versions of tinycudann down to 1.4:
Traceback (most recent call last):
File "/scratch/instant-nsr/train_nerf.py", line 116, in <module>
trainer.train(train_loader, valid_loader, 200)
File "/scratch/instant-nsr/nerf/utils.py", line 434, in train
self.train_one_epoch(train_loader)
File "/scratch/instant-nsr/nerf/utils.py", line 635, in train_one_epoch
self.scaler.scale(loss).backward()
File "/usr/local/lib/python3.10/dist-packages/torch/_tensor.py", line 487, in backward
torch.autograd.backward(
File "/usr/local/lib/python3.10/dist-packages/torch/autograd/__init__.py", line 200, in backward
Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass
File "/usr/local/lib/python3.10/dist-packages/torch/autograd/function.py", line 274, in apply
return user_fn(self, *args)
File "/usr/local/lib/python3.10/dist-packages/tinycudann/modules.py", line 145, in backward
doutput_grad, params_grad, input_grad = ctx.ctx_fwd.native_tcnn_module.bwd_bwd_input(
RuntimeError: DifferentiableObject::backward_backward_input_impl: not implemented error
qiuyuenheng commented
I also encountered the same error. Have you resolved this error?
HelloRicky123 commented
Similar problem.