Open Models Almanac
Methodsdata curationdata mixture & curriculum

single mention · data curation

Two-axis mid-training data mixing

Allocating mixture shares across domains and then distributing each domain's budget among its datasets.

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

Also called in-domain composition.

Filed under data curation › data mixture & curriculum