search

HKUDS / GraphGPT

[SIGIR'2024] "GraphGPT: Graph Instruction Tuning for Large Language Models"
https://arxiv.org/abs/2310.13023
Apache License 2.0
493 stars 36 forks source link

About text feature #30

Closed W-rudder closed 6 months ago

W-rudder commented 7 months ago

This is a very interesting job! One thing I am curious about is whether to use the [CLS] token embedding obtained through BERT processing as a feature or to use the last hidden states as features.

yuh-yang commented 6 months ago

Hi, we take the mean pooling results of last hidden states as features.

W-rudder commented 6 months ago

Hi, we take the mean pooling results of last hidden states as features.

Thanks for your response!