How to do quantization for pretrained model?

[EvelinaAleksiutenko]

I need to do quantization for the pretrained model. After reading the tutorialI try to run the code
snippet:
model = nemo_nlp.models.PunctuationCapitalizationModel.from_pretrained("punctuation_en_distilbert"). The model was loaded correctly.
quantized_model = torch.quantization.quantize_dynamic(model, {torch.nn.Linear}, dtype=torch.qint8)
Here I faced an error:
NotImplementedError: object proxy must define deepcopy()

How can I fix this? How can I actually do quantization for NLP models from NeMo?

[TheFunyBunky]

Having the same issue