Conceptacceptance rate
Draft Acceptance Rate
The core metric of speculative decoding, referring to the proportion of candidate tokens generated by the draft model that are validated by the target model, used to measure the acceleration effect.
Timeline (last 90 days)
Oct 21
MTP's speedup effect depends on the draft acceptance rate, and the acceptance rate highly depends on whether the context visible at draft time is sufficient.
Unverified50%