Open Models Almanac
Methodsinference & servingKV cache management

implementation detail · filed under inference & serving

Mamba checkpointing for speculative-decoding replay

Retains additional Mamba cache checkpoints to reduce replay needed for speculative-decoding recovery.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See KV cache management for the methods filed beside it.

Also called Additional Mamba checkpoints.

sources
2
models
2
lab adopt it
1
strongest
used

Documented in

Evidence

2 quotes from 2 sources, one card per model, each opening with its strongest.

How sources treat it
optional 1used 1

Additional Mamba checkpoints reduce the replay needed to recover inputs for speculative decoding

usedinference servingin Nemotron 3.5 Super VLNVIDIA
1 more from this model

We improved this path by optionally retaining an additional Mamba cache page near the speculative-decoding replay boundary.

optionalinference servingin Nemotron 3.5 SuperNVIDIA
↑