Skip to main content

nn_tensor_set_scales

Function nn_tensor_set_scales 

Source
pub unsafe extern "C" fn nn_tensor_set_scales(
    tensor: *mut NNTensor,
    n_scales: usize,
    scales: *const f32,
    own: c_int,
)
Expand description

Sets the quantization scales for the tensor. If n_scales>1 it should match the channel dimension (axis) of the tensor.

If own=1 then the tensor will take ownership of the buffer and free it when the tensor is released. Otherwise the buffer must outlive the tensor.

@public @memberof NNTensor @since 2.4