implementation detail · filed under inference & serving
Top-k over token clusters
Reduces MTP decoding overhead by applying top-k to token clusters instead of projecting over the entire vocabulary.
Also called top-k operation on clusters of tokens.
- 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 reduce the decoding overhead by replacing the projection operation to the entire vocabulary by a top-k operation on clusters of tokens
usedinference servingin Gemma 4Google DeepMind
Filed alongside
Other methods under inference & serving :: decoding strategy.
Speculative decodingMulti-Token PredictionDSparkEAGLEDFlashPresence PenaltyBest-of-N scaffoldingNEXTN speculative decodingMulti-layer EAGLESpeculative samplingStandardized sampling configurationMulti-stage candidate filteringRecursive shared MTP-head draftingTask-specific sampling parametersChat Prefix CompletionConcurrency-aware draft-length tuningDistribution-matched draft-model fine-tuningEAGLE-3-style draft-model fine-tuningFused recurrent replay kernelGrammar-constrained decodingKV-cache sharing between drafter and targetLongest-trace selectionMTP-1 speculative decodingSame-checkpoint target and draft weights