single mention · model architecture
Nemotron Hybrid Multi-Token Prediction
Names the Nemotron-H MTP speculative-decoding method in a runtime configuration; the evidence does not detail its mechanism.
The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See prediction head for the methods filed beside it.
Also called nemotron_h_mtp.
Filed under model architecture › prediction head