| ACL |
Knowledge neurons in pretrained transformers |
2022 |
| |
On the pitfalls of analyzing individual neurons in language models |
2022 |
| |
Discovering latent knowledge in language models without supervision. |
2023 |
| |
What is one grain of sand in the desert? analyzing individual neurons in deep nlp models. |
2018 |
| EMNLP / Geva |
Transformer feed-forward layers are key-value memories. |
2021 |
| EMNLP / Geva |
Transformer feed-forward layers build predictions by promoting concepts in the vocabulary space. |
2022 |
| |
Universal neurons in GPT2 language models. |
2023 |
| |
Finding neurons in a haystack: Case studies with sparse probing. |
2023 |