Google I/O returns May 18-20! Reserve space and build your schedule Register now


View source on GitHub

Computes Relu(x * weight + biases).

x a 2D tensor. Dimensions typically: batch, in_units
weights a 2D tensor. Dimensions typically: in_units, out_units
biases a 1D tensor. Dimensions: out_units
name A name for the operation (optional). If not specified "nn_relu_layer" is used.

A 2-D Tensor computing relu(matmul(x, weights) + biases). Dimensions typically: batch, out_units.