implementation detail · filed under training objective
Shared-weight Multi-Token Prediction
An MTP objective retained with shared weights, including during supervised fine-tuning.
Also called shared-weight MTP objective.
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- used
How sources treat it
One count per evidence span, weakest treatment to strongest.
used 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
As in pre-training, we retain the shared-weight MTP objective during SFT, using two MTP layers with a per-token auxiliary-loss scaling factor of 0.1.
usedtraining objectivein Nemotron 3 UltraNVIDIA
Filed alongside
Other methods under training objective :: multi-token prediction objective.