Model techniques map
Techniquesdata curationdata mixture & curriculum

implementation detail · filed under data curation

Attribution, hedging, and refusal data subsets

Includes data subsets encouraging in-context attribution, hedging, and refusals to improve factuality.

source
1
model
1
lab adopt it
1
strongest
used

How sources treat it

One count per evidence span, weakest treatment to strongest.

used 1

Documented in

Evidence

1 span quoted from the sources, strongest treatment first.

Including subsets of data that encourage better in-context attribution, hedging, and refusals to minimize hallucinations also improves performance on factuality metrics

useddata curationin Gemma 4Google DeepMind

Filed alongside

Other methods under data curation :: data mixture & curriculum.