Neologism Learning as a Parameter-Efficient Alternative to Fine-Tuning for Model Steering
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909972581842944 |
|---|---|
| author | Park, Sungjoon Ramamurthi, Varun Terry, Owen |
| author_facet | Park, Sungjoon Ramamurthi, Varun Terry, Owen |
| contents | In language modeling, neologisms are new tokens trained to represent a concept not already included in a given model's vocabulary. Neologisms can be used to encourage specific behavior in models, for example by appending prompts with "Give me a neologism answer." Behavioral steering can also be achieved through fine-tuning, albeit with more compute and less flexibility: learning a neologism only trains d parameters and allows the user to still access the model's default behavior. We compare the performance of neologism learning against low-rank adaptation (LoRA) fine-tuning, finding that neologisms outperform fine-tuned models under a matched training setup (same data and hyperparameters). We also investigate self-verbalizations of neologisms, and observe that the model will occasionally make up its own new words when asked about a neologism. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_18551 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Neologism Learning as a Parameter-Efficient Alternative to Fine-Tuning for Model Steering Park, Sungjoon Ramamurthi, Varun Terry, Owen Computation and Language In language modeling, neologisms are new tokens trained to represent a concept not already included in a given model's vocabulary. Neologisms can be used to encourage specific behavior in models, for example by appending prompts with "Give me a neologism answer." Behavioral steering can also be achieved through fine-tuning, albeit with more compute and less flexibility: learning a neologism only trains d parameters and allows the user to still access the model's default behavior. We compare the performance of neologism learning against low-rank adaptation (LoRA) fine-tuning, finding that neologisms outperform fine-tuned models under a matched training setup (same data and hyperparameters). We also investigate self-verbalizations of neologisms, and observe that the model will occasionally make up its own new words when asked about a neologism. |
| title | Neologism Learning as a Parameter-Efficient Alternative to Fine-Tuning for Model Steering |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2512.18551 |