arXiv - CSCL: "Reward-Augmented Decoding: Efficient Controlled T…" - Qoto Mastodon

arXiv - CSCL @arxiv_cscl@qoto.org

Reward-Augmented Decoding: Efficient Controlled Text Generation With a Unidirectional Reward Model. (arXiv:2310.09520v4 [cs.CL] UPDATED)

http://arxiv.org/abs/2310.09520 #arXiv #NLProc

Jan 03, 2024, 03:20 · · arxiv-cscl · · ·

Sign in to participate in the conversation