fn
layer_norm_op
→TensorImplPtrint layer_norm_op(const int & x, const int & gamma, const int & beta, double eps)Python wrappers
Public Python APIs implemented by this engine symbol.Public free-function entry point for Layer Normalization.
Thin wrapper that delegates to LayerNormBackward::forward.
This is the symbol the pybind11 binding layer forwards to from
lucid.nn.functional.layer_norm.
See Also
LayerNormBackward : Autograd node implementing the forward + backward. rms_norm_op : Variant that omits mean subtraction.
Parameters
xTensorImplPtrInput tensor of any shape.
gammaTensorImplPtrPer-element scale; must match the trailing dims of
x.betaTensorImplPtrPer-element shift; same shape as
gamma.epsdoubleNumerical-stability constant.
Returns
TensorImplPtrNormalised output, same shape as x.