Quantization techniques that aren't supported in Transformers can be added with the [HfQuantizer] class.