RMSNorm

class torch.nn.RMSNorm(normalized_shape, eps=None, elementwise_affine=True, device=None, dtype=None)

Applies Root Mean Square Layer Normalization over a mini-batch of inputs.

This layer implements the operation as described in the paper Root Mean Square Layer Normalization

$$y_i = \frac{x_i}{\mathrm{RMS}(x)} * \gamma_i, \quad \text{where} \quad \text{RMS}(x) = \sqrt{\epsilon + \frac{1}{n} \sum_{i=1}^{n} x_i^2}$$

The RMS is taken over the last D dimensions, where D is the dimension of normalized_shape. For example, if normalized_shape is (3, 5) (a 2-dimensional shape), the RMS is computed over the last 2 dimensions of the input.

Parameters

• normalized_shape (int or list or torch.Size) –

  input shape from an expected input of size

  $$[* \times \text{normalized\_shape}[0] \times \text{normalized\_shape}[1] \times \ldots \times \text{normalized\_shape}[-1]]$$

  If a single integer is used, it is treated as a singleton list, and this module will normalize over the last dimension which is expected to be of that specific size.

• eps (Optional[float]) – a value added to the denominator for numerical stability. Default: torch.finfo(x.dtype).eps()

• elementwise_affine (bool) – a boolean value that when set to True, this module has learnable per-element affine parameters initialized to ones (for weights). Default: True.

Shape:

• Input: $(N, *)$

• Output: $(N, *)$ (same shape as input)

Examples:

>>> rms_norm = nn.RMSNorm([2, 3])
>>> input = torch.randn(2, 2, 3)
>>> rms_norm(input)

extra_repr()

Extra information about the module.

Return type

str

forward(x)

Runs forward pass.

Return type

Tensor

reset_parameters()

Resets parameters based on their initialization used in __init__.