launch_ref

cubecl_quant::dequantize

Function launch_ref 

Source 
pub fn launch_ref<R: Runtime>(
    client: &ComputeClient<R>,
    values: &TensorHandleRef<'_, R>,
    output: &TensorHandleRef<'_, R>,
    params: &TensorHandleRef<'_, R>,
    scheme: &QuantScheme,
    input_dtype: StorageType,
) -> Result<(), LaunchError>
Expand description

Convert the tensor back to a higher precision data type.