Property HasMultiTokenPrediction
HasMultiTokenPrediction
Indicates whether the model carries Multi-Token Prediction (MTP)
heads exposed by the GGUF metadata key
<arch>.nextn_predict_layers. When true, LM-Kit
allocates a sibling MTP context and uses self-speculative
draft-and-verify decoding to accelerate generation. Present on
architectures such as Qwen 3.5 / 3.6, GLM-4.x, DeepSeek V3 / R1,
BailingMoE2, and ExaOne-MoE.
public bool HasMultiTokenPrediction { get; }