Mixed Quantization

Layer Sensitivity

Measure of the impact of quantization on the performance of each individual layer of the neural network. Sensitive layers require higher precision to maintain overall model quality.

← Indietro