How cloud I convert a quantized model into a lossy model of float32?

quantization
4

maybe you can try [quantization] Frequently Asked Questions to see if this works for you or not