Quantize Callback
Quantize your network during training
QuantizeCallback
QuantizeCallback (quantizer=None, backend='x86', use_per_tensor=False, verbose=False)
Simple callback for Quantization-Aware Training (QAT) in fastai. Uses the Quantizer class for configuration and conversion.
| Type | Default | Details | |
|---|---|---|---|
| quantizer | NoneType | None | Provide custom quantizer |
| backend | str | x86 | Target backend for quantization: ‘x86’, ‘qnnpack’ |
| use_per_tensor | bool | False | Force per-tensor quantization |
| verbose | bool | False | Enable verbose output |