Gender bias in Pythia-6.9b grows sharply after about 80k training steps even as general performance improves, and stopping earlier could trade 1.7% LAMBADA accuracy for a large fairness gain.
Towards understanding fairness and its composition in ensemble machine learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Fairness Dynamics During Training
Gender bias in Pythia-6.9b grows sharply after about 80k training steps even as general performance improves, and stopping earlier could trade 1.7% LAMBADA accuracy for a large fairness gain.