nerfbaselines.communication

class nerfbaselines.communication.ConnectionParams(port: int = <factory>, authkey: bytes = <factory>)

Bases: object

authkey: bytes

port: int

class nerfbaselines.communication.RemoteCallable(i)

Bases: object

class nerfbaselines.communication.RemoteMethod(*args, checkpoint: Path | None = None, connection_params: ConnectionParams | None = None, **kwargs)

Bases: Method

call(function: str, *args, **kwargs)

close()

static decode_args(encoded_args)

property encoded_args

get_info() → MethodInfo

Get method defaults for the trainer.

Returns:

Method info.

render(*args, **kwargs)

Render images.

Parameters:

• cameras – Cameras.

• progress_callback – Callback for progress.

save(path: Path, **kwargs)

Save model.

Parameters:

path – Path to save.

setup_train(*args, **kwargs)

Setup training data, model, optimizer, etc.

Parameters:

• train_dataset – Training dataset.

• num_iterations – Number of iterations to train.

property shared_path: Tuple[str, str] | None

train_iteration(*args, **kwargs)

Train one iteration.

Parameters:

step – Current step.

class nerfbaselines.communication.RemoteProcessMethod(*args, build_code: str | None = None, python_path: str | None = None, **kwargs)

Bases: RemoteMethod

build_code: str | None = None

close()

classmethod install()

Install the method.

python_path: str = 'python'

classmethod wrap(method: Type[Method], **remote_kwargs)

nerfbaselines.communication.inject_callables(obj, output_queue, my_id)

nerfbaselines.communication.replace_callables(obj, callables, depth=0)

nerfbaselines.communication.start_backend(method: Method, params: ConnectionParams, address: str = 'localhost')