search

Deci-AI / super-gradients

Easily train or fine-tune SOTA computer vision models with one open source training library. The home of Yolo-NAS.
https://www.supergradients.com
Apache License 2.0
4.53k stars 490 forks source link

Change Quantization Precision #1991

Closed Halafz7 closed 3 months ago

Halafz7 commented 4 months ago

💡 Your Question

Dear Author @BloodAxe

So I've read the PTQ and QAT for Classification README and I found about QuantDescriptor image

I want to use PTQ_and_QAT_for_classification.ipynb file to produce vary quantized model with custom precision (4-bit, 8-bit, 16-bit). On what file I should declare the num_bits of QuantDescriptor's args so I can produce the quantized model?

Thank you

Versions

No response

BloodAxe commented 3 months ago

At this moment, changing precision is not supported and is not anywhere on the roadmap.