|
bitrl & cuberl Documentation
Simulation engine for reinforcement learning agents
|
Variables | |
| const std::string | CONFIG = "config.json" |
This example utilises the TensorboardServer class to log values of interest when running an experiment. We can monitor the experimet using tensorboard. The TensorboardServer class is a simple wrapper that exposes three functions
- ```add_scalar``` - ```add_scalars``` - ```add_text``` We will use ```add_scalar``` and ```add_text```. In order to run this example, fire up the
server using the torchboard_server/start_uvicorn.sh. The server listens at port 8002. You can change this however you want just make sure that the port is not used and also update the variable TORCH_SERVER_HOST in the code below accordingly. Note that the implementation uses SummaryWriter class. Thus you will need to have PyTorch installed on the machine that you run the server.
| const std::string example_12::CONFIG = "config.json" |