implementation detail · not yet filed
likelihood-based acceptance-rate loss (LK loss)
Also called LK loss for draft model fine-tuning.
- 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.
we directly optimize the likelihood-based LK loss, the negative logarithm of the acceptance rate itself, LLK = −log Σ_{x∈V} min(p(x), q(x))
usedunclearin Kimi K3Moonshot AI