I am not able to put my the model I’ve written in LibTorch on my CUDA device. I’m sure the CUDA device is working because I can put tensors on it.
I tried putting the copies of the model I stored on the CUDA device like so,
auto actor = Act_Net(low, high);
auto target_actor = Act_Net(low, high);
auto critic = Critic_Net();
auto target_critic = Critic_Net();
actor->to(device);
target_actor->to(device);
critic->to(device);
target_critic->to(device);
This returns the error…
/home/iii/tor/m_gym/multiv_normal.cpp:183:1: error: ‘actor’ does not name a type
183 | actor->to(device);
| ^~~~~
/home/iii/tor/m_gym/multiv_normal.cpp:184:1: error: ‘target_actor’ does not name a type
184 | target_actor->to(device);
| ^~~~~~~~~~~~
/home/iii/tor/m_gym/multiv_normal.cpp:185:1: error: ‘critic’ does not name a type
185 | critic->to(device);
| ^~~~~~
/home/iii/tor/m_gym/multiv_normal.cpp:186:1: error: ‘target_critic’ does not name a type
186 | target_critic->to(device);
| ^~~~~~~~~~~~~
This is what it looks like in VS Code.
The LibTorch documentation says…
“and also our model parameters should be moved to the correct device:”
generator->to(device);
discriminator->to(device);
Maybe I don’t know where in the script to put actor->to(device);
?