Computer scienceData scienceMachine learningML deployment

Quantization

Static and dynamic

Report a typo

What is the difference between the static and dynamic calibration in the post-training quantization?

Select one option from the list

Dynamic calibration is performed during the quantization, static calibration is done at the runtime

Static calibration requires retraining the model after quantization, while dynamic calibration does not require any retraining or fine-tuning

Dynamic calibration is performed at runtime, static calibration is done at the quantization time

Static calibration quantizes the entire model using a single set of quantization parameters, while dynamic calibration quantizes each tensor independently with their quantization parameters

Create a free account to access the full topic