Function quantize_named_parameters 

pub fn quantize_named_parameters<B: Backend, M: Module<B>>(
    module: &M,
    config: &QuantConfig,
) -> Result<Vec<(String, QuantizedTensor)>>

Quantize all Linear layers in a model’s named_parameters.

Returns a vector of (name, QuantizedTensor) for each weight parameter. Biases are kept in FP32.

Example

let model = Sequential::new(vec![...]);
let quantized = quantize_named_parameters(&model, &QuantConfig::int8())?;
for (name, qt) in &quantized {
    println!("{}: {} → {} bytes ({:.1}x compression)",
        name, qt.numel() * 4, qt.size_bytes(), qt.compression_ratio());
}