r"""Move internal buffers of the ExponentialMovingAverage to `device`. Args: device: like `device` argument to `torch.Tensor.to`

r""" Instantiate a pretrained PyTorch model from a pretrained model configuration. The model is set in evaluation mode - `model.eval(

r""" Args: Loads the ExponentialMovingAverage state. This method is used by accelerate during checkpointing to save the ema st

Return a logger with the specified name. This function is not supposed to be directly accessed unless you are writing a custom muse module.

returns a sampling function with given ODE settings Args: - sampling_method: type of sampler used in solving the ODE; default to be Do

Performs a single optimization step. Args: closure (callable, optional): A closure that reevaluates the model and return

Compute the noise coefficient along the path

`torch.device`: The device on which the module is (assuming that all the module parameters are on the same device).

Function to reshape time t to broadcastable dimension of x Args: t: [batch_dim,], time vector x: [batch_dim,...], data point

function for creating Transport object **Note**: model prediction defaults to velocity Args: - path_type: type of path to use; default to

r""" Returns the state of the ExponentialMovingAverage as a dict. This method is used by accelerate during checkpointing to save the e

This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, he

Save a model and its configuration file to a directory so that it can be reloaded using the [`~models.ModelMixin.from_pretrained`] cl

Generate 1:1 similar to the original MaskGit repo https://github.com/google-research/maskgit/blob/main/maskgit/libml/parallel_decode.

Compute the data coefficient along the path

We always output sde according to score parametrization;

This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, he

This is the equivalent of torch.repeat_interleave(x, dim=1, repeats=n_rep). The hidden states go from (batch, num_key_value_heads, seqlen, he

Set the verbosity level for the 🤗 muse' root logger. Args: verbosity (`int`): Logging level, e.g., one of:

Set the verbosity level for the 🤗 muse' root logger. Args: verbosity (`int`): Logging level, e.g., one of:

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T

`torch.device`: The device on which the module is (assuming that all the module parameters are on the same device).

Return a logger with the specified name. This function is not supposed to be directly accessed unless you are writing a custom muse module.

Unified API to get any scheduler from its name. Args: name (`str` or `SchedulerType`): The name of the scheduler to use.

Sampling x0 & t based on shape of x1 (if needed) Args: x1 - data point; [batch, *dim]

r""" Set the switch for the npu flash attention.

r""" Set the switch for the npu flash attention.

Set the verbosity to the `ERROR` level.

Set the verbosity to the `INFO` level.

Applies Rotary Position Embedding to the query and key tensors. Args: q (`torch.Tensor`): The query tensor. k (`torch.Tensor`): T