Saved in:
Bibliographic Details
Main Author: Phang, Jason
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2402.16817
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917598171496448
author Phang, Jason
author_facet Phang, Jason
contents Gisting (Mu et al., 2023) is a simple method for training models to compress information into fewer token representations using a modified attention mask, and can serve as an economical approach to training Transformer-based hypernetworks. We introduce HyperLlama, a set of Gisting-based hypernetworks built on Llama-2 models that generates task-specific soft prefixes based on few-shot inputs. In experiments across P3, Super-NaturalInstructions and Symbol Tuning datasets, we show that HyperLlama models can effectively compress information from few-shot examples into soft prefixes. However, they still underperform multi-task fine-tuned language models with full attention over few-shot in-context examples. We also show that HyperLlama-generated soft prefixes can serve as better initializations for further prefix tuning. Overall, Gisting-based hypernetworks are economical and easy to implement, but have mixed empirical performance.
format Preprint
id arxiv_https___arxiv_org_abs_2402_16817
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Investigating the Effectiveness of HyperTuning via Gisting
Phang, Jason
Computation and Language
Gisting (Mu et al., 2023) is a simple method for training models to compress information into fewer token representations using a modified attention mask, and can serve as an economical approach to training Transformer-based hypernetworks. We introduce HyperLlama, a set of Gisting-based hypernetworks built on Llama-2 models that generates task-specific soft prefixes based on few-shot inputs. In experiments across P3, Super-NaturalInstructions and Symbol Tuning datasets, we show that HyperLlama models can effectively compress information from few-shot examples into soft prefixes. However, they still underperform multi-task fine-tuned language models with full attention over few-shot in-context examples. We also show that HyperLlama-generated soft prefixes can serve as better initializations for further prefix tuning. Overall, Gisting-based hypernetworks are economical and easy to implement, but have mixed empirical performance.
title Investigating the Effectiveness of HyperTuning via Gisting
topic Computation and Language
url https://arxiv.org/abs/2402.16817