Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2503.12667 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866909538998812672 |
|---|---|
| author | Chmura, Jacob Dauvet, Jonah Sabry, Sebastian |
| author_facet | Chmura, Jacob Dauvet, Jonah Sabry, Sebastian |
| contents | Despite advances in language modelling, distributional methods that build semantic representations from co-occurrences fail to discriminate between plausible and implausible events. In this work, we investigate how plausibility prediction can be improved by injecting latent knowledge prompted from large language models using parameter-efficient fine-tuning. We train 12 task adapters to learn various physical properties and association measures and perform adapter fusion to compose latent semantic knowledge from each task on top of pre-trained AlBERT embeddings. We automate auxiliary task data generation, which enables us to scale our approach and fine-tune our learned representations across two plausibility datasets. Our code is available at https://github.com/Jacob-Chmura/plausibility-vaccine. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_12667 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Plausibility Vaccine: Injecting LLM Knowledge for Event Plausibility Chmura, Jacob Dauvet, Jonah Sabry, Sebastian Computation and Language Artificial Intelligence Despite advances in language modelling, distributional methods that build semantic representations from co-occurrences fail to discriminate between plausible and implausible events. In this work, we investigate how plausibility prediction can be improved by injecting latent knowledge prompted from large language models using parameter-efficient fine-tuning. We train 12 task adapters to learn various physical properties and association measures and perform adapter fusion to compose latent semantic knowledge from each task on top of pre-trained AlBERT embeddings. We automate auxiliary task data generation, which enables us to scale our approach and fine-tune our learned representations across two plausibility datasets. Our code is available at https://github.com/Jacob-Chmura/plausibility-vaccine. |
| title | Plausibility Vaccine: Injecting LLM Knowledge for Event Plausibility |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2503.12667 |