implementation detail · filed under training objective
Frozen-backbone MTP-head fine-tuning
Fine-tuning in which the backbone is frozen and only the MTP head receives gradient updates.
- 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.
The backbone is fixed for the entire training phase, and only the MTP head receives gradient updates.
usedunclearin Nemotron 3 UltraNVIDIA
Filed alongside
Other methods under training objective :: multi-token prediction objective.