Prompt-MII: Meta-Learning Instruction Induction for LLMs
By: Emily Xiao , Yixiao Zeng , Ada Chen and more
Potential Business Impact:
Makes AI answer questions using less information.
A popular method to adapt large language models (LLMs) to new tasks is in-context learning (ICL), which is effective but incurs high inference costs as context length grows. In this paper we propose a method to perform instruction induction, where we take training examples and reduce them to a compact but descriptive prompt that can achieve performance comparable to ICL over the full training set. Specifically, we propose PROMPT-MII, a reinforcement learning (RL) based framework to meta-learn an instruction induction model that can generate compact instructions on the fly for an arbitrary new dataset. We train on over 3,000 diverse classification datasets from the HuggingFace hub, and evaluate on 90 unseen tasks. PROMPT-MII improves downstream model quality by 4-9 F1 points (10-20% relative), matching ICL performance while requiring 3-13x fewer tokens.
Similar Papers
AdaptMI: Adaptive Skill-based In-context Math Instruction for Small Language Models
Computation and Language
Helps small AI learn math better by giving right clues.
Meta-Adaptive Prompt Distillation for Few-Shot Visual Question Answering
Artificial Intelligence
Helps AI understand pictures better with fewer examples.
MachineLearningLM: Continued Pretraining Language Models on Millions of Synthetic Tabular Prediction Tasks Scales In-Context ML
Computation and Language
Teaches computers to learn from many examples.