Open Models Almanac
Methodspost-trainingsupervised fine-tuning

single mention · post-training

Random-budget truncation of reasoning traces

Truncates a subset of SFT reasoning traces to random budgets while masking designated reasoning markers from loss.

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

Filed under post-training › supervised fine-tuning