Papers
arxiv:2301.13688

The Flan Collection: Designing Data and Methods for Effective Instruction Tuning

Published on Jan 31, 2023
Authors:
,
,
,
,
,
,
,
,

Abstract

We study the design decisions of publicly available instruction tuning methods, and break down the development of Flan 2022 (Chung et al., 2022). Through careful ablation studies on the Flan Collection of tasks and methods, we tease apart the effect of design decisions which enable Flan-T5 to outperform prior work by 3-17%+ across evaluation settings. We find task balancing and enrichment techniques are overlooked but critical to effective instruction tuning, and in particular, training with mixed prompt settings (zero-shot, few-shot, and chain-of-thought) actually yields stronger (2%+) performance in all settings. In further experiments, we show Flan-T5 requires less finetuning to converge higher and faster than T5 on single downstream tasks, motivating instruction-tuned models as more computationally-efficient starting checkpoints for new tasks. Finally, to accelerate research on instruction tuning, we make the Flan 2022 collection of datasets, templates, and methods publicly available at https://github.com/google-research/FLAN/tree/main/flan/v2.

Community

Sign up or log in to comment

Models citing this paper 48

Browse 48 models citing this paper

Datasets citing this paper 31

Browse 31 datasets citing this paper

Spaces citing this paper 123

Collections including this paper 7