Enum LoftQAdapter<T>.QuantizationType

Namespace

AiDotNet.LoRA.Adapters

Assembly

AiDotNet.dll

Specifies the type of 4-bit quantization to use for base layer weights.

public enum LoftQAdapter<T>.QuantizationType

Fields

INT4 = 0

4-bit integer quantization with uniform spacing (-8 to 7).

NF4 = 1

4-bit Normal Float quantization optimized for normally distributed weights.

Recommended for most neural network weights. NF4 with LoftQ initialization provides the best accuracy-memory trade-off.

Remarks

Same quantization types as QLoRA. The alternating optimization works with both.