How to get quantized weights from QAT model?

Hi all. I've recently trained a keras implementation of ssd-keras. I've managed to run QAT training on the model and got desired the accuracy. I wanted to get the quantised weights from the QAT model saved as a H5 model. There's no support or documentation regarding this in the tensorflow website. How can I get the quantised weights from the saved model after QAT?
I tried converting it to TFLite but it is not converting due to a custom layer in model definition. So it would be helpful if I can get the quantised weights alone from the saved model. Here's the code snippet for my QAT training. I am using TF 2.3.

`quantize_scope = tfmot.quantization.keras.quantize_scope

def apply_quantization_to_dense(layer):
if 'priorbox' in layer.name:
return layer

if isinstance(layer,tf.keras.layers.Concatenate) or isinstance(layer, tf.keras.layers.Reshape) or isinstance(layer,tf.keras.layers.Lambda):
return layer

return tfmot.quantization.keras.quantize_annotate_layer(layer)

annotated_model = tf.keras.models.clone_model(
model,
clone_function=apply_quantization_to_dense,
)

with quantize_scope({'AnchorBoxes': AnchorBoxes}):
quant_aware_model = tfmot.quantization.keras.quantize_apply(annotated_model)

quant_aware_model.summary()
quant_aware_model.compile(optimizer=adam, loss=ssd_loss.compute_loss)
quant_aware_model.fit_generator(train_generator, epochs=424, steps_per_epoch=1000,
callbacks=callbacks, validation_data=val_generator,
validation_steps=100, initial_epoch=414)`

Start by reviewing the QAT setup in the issue, especially quantize_scope, quantize_apply, the custom AnchorBoxes layer, and the saved H5 model path. Determine whether quantized weights can be retrieved directly or whether conversion through TFLite is required, then document the supported workflow and how the custom layer affects conversion. Done means a reproducible answer for the TensorFlow 2.3 setup described.