Outlier Dimensions Encode Task-Specific Knowledge
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2023
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911762798870528 |
|---|---|
| author | Rudman, William Chen, Catherine Eickhoff, Carsten |
| author_facet | Rudman, William Chen, Catherine Eickhoff, Carsten |
| contents | Representations from large language models (LLMs) are known to be dominated by a small subset of dimensions with exceedingly high variance. Previous works have argued that although ablating these outlier dimensions in LLM representations hurts downstream performance, outlier dimensions are detrimental to the representational quality of embeddings. In this study, we investigate how fine-tuning impacts outlier dimensions and show that 1) outlier dimensions that occur in pre-training persist in fine-tuned models and 2) a single outlier dimension can complete downstream tasks with a minimal error rate. Our results suggest that outlier dimensions can encode crucial task-specific knowledge and that the value of a representation in a single outlier dimension drives downstream model decisions. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2310_17715 |
| institution | arXiv |
| publishDate | 2023 |
| record_format | arxiv |
| spellingShingle | Outlier Dimensions Encode Task-Specific Knowledge Rudman, William Chen, Catherine Eickhoff, Carsten Computation and Language Artificial Intelligence Representations from large language models (LLMs) are known to be dominated by a small subset of dimensions with exceedingly high variance. Previous works have argued that although ablating these outlier dimensions in LLM representations hurts downstream performance, outlier dimensions are detrimental to the representational quality of embeddings. In this study, we investigate how fine-tuning impacts outlier dimensions and show that 1) outlier dimensions that occur in pre-training persist in fine-tuned models and 2) a single outlier dimension can complete downstream tasks with a minimal error rate. Our results suggest that outlier dimensions can encode crucial task-specific knowledge and that the value of a representation in a single outlier dimension drives downstream model decisions. |
| title | Outlier Dimensions Encode Task-Specific Knowledge |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2310.17715 |