Task-Level Natural Language Priors as Learning Signals for Low-Resource LLM Training

Jian Gao, Xiao Zhang, Xun Zhu +2

Large language models (LLMs) often struggle when low-resource training data are ambiguous or incomplete. Task-level natural-language priors can provide useful guidance in such sett…