Quantitation Aware Fine Tuning?

[ramkumarkoppu]

Hi,

I used to do Quantitation aware training on CNN models for improved accuracy on converted TFlite models for int8 inference using tensorflow_model_optimization. Will this same approach useful on keras_nlp LLM fine-tuning for int8 TFlite inference?

[jbischof]

Yes @ramkumarkoppu all our models are compatible with TFMOT. See https://www.tensorflow.org/model_optimization