February 15, 2024Open Access

Subgraph-level Universal Prompt Tuning

View Full Paper

Ask AI

Bookmark

Authors

JLJunhyun Lee WYWooseong Yang JKJaewoo Kang

Discussion

Member takes

Overview

Benchmark evaluation demonstrates superior accuracy of subgraph-level universal prompt tuning across pre-trained graph models, highlighting parameter efficiency.

Key Points

Subgraph-level universal prompt tuning outperforms traditional fine-tuning across diverse pre-trained models while using fewer tuning parameters.
In full-shot experiments, the method wins in 42 of 45 tests with a 2.5% gain, while few-shot tests show a 6.6% increase across 41 of 45 runs.
By assigning prompt features directly within local graph feature space, this universal approach enables context-aware adaptation across tasks.

Cite This Study

Lee et al. (2024) studied this question.

synapsesocial.com/papers/68e79181b6db643587703008 https://doi.org/10.48550/arxiv.2402.10380

View Full Paper

Ask AI

Bookmark

February 15, 2024Open Access

Subgraph-level Universal Prompt Tuning

View Full Paper

Ask AI

Bookmark

Authors

JLJunhyun Lee WYWooseong Yang JKJaewoo Kang

Discussion

Member takes

Overview

Benchmark evaluation demonstrates superior accuracy of subgraph-level universal prompt tuning across pre-trained graph models, highlighting parameter efficiency.

Key Points

Subgraph-level universal prompt tuning outperforms traditional fine-tuning across diverse pre-trained models while using fewer tuning parameters.
In full-shot experiments, the method wins in 42 of 45 tests with a 2.5% gain, while few-shot tests show a 6.6% increase across 41 of 45 runs.
By assigning prompt features directly within local graph feature space, this universal approach enables context-aware adaptation across tasks.

Cite This Study

Lee et al. (2024) studied this question.

synapsesocial.com/papers/68e79181b6db643587703008 https://doi.org/10.48550/arxiv.2402.10380

View Full Paper

Ask AI

Bookmark