Follow
From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models. (arXiv:2305.08283v3 [cs.CL] UPDATED)
From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Models. (arXiv:2305.08283v3 [cs.CL] UPDATED)