Rate this Page
★ ★ ★ ★ ★

torch.Tensor.q_per_channel_zero_points#

Tensor.q_per_channel_zero_points() → Tensor#

Given a Tensor quantized by linear (affine) per-channel quantization, returns a tensor of zero_points of the underlying quantizer. It has the number of elements that matches the corresponding dimensions (from q_per_channel_axis) of the tensor.