dequantize_tensor

Function dequantize_tensor 

Source
pub fn dequantize_tensor<B: Backend>(
    qtensor: &QuantizedTensor,
    device: &B::Device,
) -> Result<Tensor<B>>
Expand description

Dequantize a QuantizedTensor back to a float tensor.

The dequantized values are approximate: float ≈ (int - zero_point) * scale.