Get projected gravity

Encode text using CLIP model.

Apply gradient clipping to model parameters. Args: model: PyTorch model to clip gradients for

Create a linear module.

Copy model parameters into a (differently-shaped) list of full-precision parameters.

w x y z Convert rotations given as rotation matrices to quaternions. Args: matrix: Rotation matrices as tensor of shape (..., 3,

Future N-step motion anchor orientation in body frame - matching Isaac Lab order

Future N-step motion anchor position in body frame - matching Isaac Lab order

Compute the KL divergence between two gaussians. Shapes are automatically broadcasted, so batches can be compared to scalars, among othe

Apply an inverse quaternion rotation to a vector. Args: quat: The quaternion in (w, x, y, z). Shape is (..., 4). vec:

Construct 3D identity rotation given shape

Construct 3D identity rotation given shape

Combine two set of 3D rotations together using \**\* operator. The shape needs to be broadcastable

Construct 3D rotation from quaternion (the quaternion needs not to be normalized).

Construct 3D rotation from quaternion (the quaternion needs not to be normalized).

WXYZ Convert rotations given as quaternions to rotation matrices. Args: quaternions: quaternions with real part first,

Collect gradients from all GPUs and average them. This function is called after the backward pass to synchronize the gradients across all GPU

Register a model for EMA (Exponential Moving Average) tracking. Args: name: Unique identifier for the model

Runs the simulation loop.

是否需要评估

Perform a single inference step and return timing information.

Take the sum over all non-batch dimensions.

Get translation from transform

Get translation from transform

Update EMA model parameters using exponential moving average. Args: name: Model identifier model: Current mo

Args: δx: desired position - current position dx: current velocity

Compute suitable font scale.

returns a list of [pad, conv] => should be += to some list, then apply sequential

copy current camera to clipboard as MJCF specification

copy qpos to clipboard as key

init profiler figures

init sensor figure

------------------------------------- load mjb or xml model --------------------------------------

load state from history buffer

make group section of UI

make physics section of UI

make rendering section of UI

make model-dependent UI sections

make visualization section of UI

simulate in background thread (while rendering in main thread)

show profiler figures

show sensor figure

set window layout

prepare info text

update watch