Open Models Almanac
Methodsdata curationdata mixture & curriculum

single mention · data curation

Difficulty-based routing of validated questions to SFT or RL

Routing validated questions to supervised fine-tuning or reinforcement-learning data pools according to measured difficulty.

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.

Filed under data curation › data mixture & curriculum