5
Time is encoded in the weights of finetuned language models
source link: https://arxiv.org/abs/2312.13401
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
Computer Science > Computation and Language
[Submitted on 20 Dec 2023]
Time is Encoded in the Weights of Finetuned Language Models
Download PDF HTML (experimental)
We present time vectors, a simple tool to customize language models to new time periods. Time vectors are created by finetuning a language model on data from a single time (e.g., a year or month), and then subtracting the weights of the original pretrained model. This vector specifies a direction in weight space that, as our experiments show, improves performance on text from that time period. Time vectors specialized to adjacent time periods appear to be positioned closer together in a manifold. Using this structure, we interpolate between time vectors to induce new models that perform better on intervening and future time periods, without any additional training. We demonstrate the consistency of our findings across different tasks, domains, model sizes, and time scales. Our results suggest that time is encoded in the weight space of finetuned models.
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2312.13401 [cs.CL] |
| (or arXiv:2312.13401v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2312.13401 |
Recommend
About Joyk
Aggregate valuable and interesting links.
Joyk means Joy of geeK