Kformer: Knowledge Injection in Transformer Feed-Forward Layers

By Yunzhi Yao, Shaohan Huang, Ningyu Zhang, Li Dong, Furu Wei, Huajun Chen. 2022

52 citations recorded by the source corpus.

Recent days have witnessed a diverse set of knowledge injection models for pre-trained language models (PTMs); however, most previous studies neglect the PTMs' own ability with quantities of implicit knowledge stored in parameters. A recent study has observed knowledge neurons in the Feed Forward Network (FFN), which are responsible for expressing factual knowledge. In this work, we propose a simple model, Kformer, which takes advantage of the knowledge stored in PTMs and external knowledge via knowledge injection in Transformer FFN layers. Empirically results on two knowledge-intensive tasks, commonsense reasoning (i.e., SocialIQA) and medical question answering (i.e., MedQA-USMLE), demonstrate that Kformer can yield better performance than other knowledge injection technologies such as concatenation or attention-based injection. We think the proposed simple model and empirical findings may be helpful for the community to develop more powerful knowledge injection methods. Code available in https://github.com/zjunlp/Kformer.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures, Knowledge Graphs & Representation

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Yonatan Belinkov
2. David Bau
3. Alex Andonian
4. Mor Geva
5. Yoav Goldberg
6. Kevin Meng
7. Meg Tong
8. Asa Cooper Stickland

In Computer Vision and Pattern Recognition

1. David Bau
2. Yonatan Belinkov
3. Alex Andonian
4. Mor Geva
5. Yoav Goldberg
6. Kevin Meng
7. Ke Wang
8. Avi Caciularu

In Model Architectures

1. Yonatan Belinkov
2. Mor Geva
3. David Bau
4. Kevin Meng
5. Alex Andonian
6. Yoav Goldberg
7. Asa Cooper Stickland
8. Anders Søgaard

In Knowledge Graphs & Representation

1. Yonatan Belinkov
2. Yoav Goldberg

Citations in the corpus

• Mass-Editing Memory in a Transformer (2022, positive): “Previously, Dai et al. (2022) and Yao et al. (2022) have proposed editing methods that alter sparse sets of neurons, but we adopt the classical view of a linear layer as an associative memory (Anderson, 1972; Kohonen, 1972).”
• Transformer Feed-Forward Layers Build Predictions by Promoting Concepts in the Vocabulary Space (2022, background): “More recently, a surge of works have investigated the knowledge captured by the FFN layers (Da et al., 2021; Jiang et al., 2020; Dai et al., 2021; Yao et al., 2022; Meng et al., 2022; Wallat et al., 2020).”
• Editing Large Language Models: Problems, Methods, and Opportunities (2023, unknown): “…model editing techniques bear resemblance to knowledge augmentation (Zhang et al., 2019; Lewis et al., 2020; Zhang et al., 2022b; Yasunaga et al., 2021; Yao et al., 2022; Pan et al., 2023) approaches, as updating the model’s knowledge can also be considered as instilling knowledge into the model.”
• The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A" (2023, methodology): “Other techniques for knowledge editing include closed-form weight updates (Meng et al., 2023; Mitchell et al., 2021; Yao et al., 2022) and hyper-networks (De Cao et al., 2021; Hase et al., 2023).”