Open Models Almanac
Methodsdata curationsequence packing

specific method · filed under data curation

Length-aware best-fit packing

Multiple conversations are packed into sequences up to a maximum context length using length-aware best-fit placement.

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

Also called length-aware best-fit packing strategy.

sources
2
model
1
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
used 2

NVIDIA-Nemotron-3-Ultra-550B-A55B

NVIDIAadopted2 sources

we adopt a length-aware best-fit packing strategy (Ding et al., 2024), which packs multiple conversations into sequences up to a maximum context length.

useddata curationin Nemotron 3 UltraNVIDIA
1 more from this model

adopt a length-aware best-fit packing strategy (ding2024fewer)

useddata curationin Nemotron 3 UltraNVIDIA
↑