Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | https://arxiv.org/abs/2402.10639 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866911907274817536 |
|---|---|
| author | Nguyen, Tuc Le, Thai |
| author_facet | Nguyen, Tuc Le, Thai |
| contents | Several parameter-efficient fine-tuning methods based on adapters have been proposed as a streamlined approach to incorporate not only a single specialized knowledge into existing Pre-Trained Language Models (PLMs) but also multiple of them at once. Recent works such as AdapterSoup propose to mix not all but only a selective sub-set of domain-specific adapters during inference via model weight averaging to optimize performance on novel, unseen domains with excellent computational efficiency. However, the essential generalizability of this emerging weight-space adapter mixing mechanism on \textit{unseen, in-domain examples} remains unexplored. Thus, in this study, we conduct a comprehensive analysis to elucidate the generalizability of domain-specific adapter mixtures in in-domain evaluation. We also provide investigations into the inner workings of the mixture of domain-specific adapters by analyzing their weight signs, yielding critical analysis on the negative correlation between their fraction of weight sign difference and their mixtures' generalizability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_10639 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Generalizability of Mixture of Domain-Specific Adapters from the Lens of Signed Weight Directions and its Application to Effective Model Pruning Nguyen, Tuc Le, Thai Computation and Language Several parameter-efficient fine-tuning methods based on adapters have been proposed as a streamlined approach to incorporate not only a single specialized knowledge into existing Pre-Trained Language Models (PLMs) but also multiple of them at once. Recent works such as AdapterSoup propose to mix not all but only a selective sub-set of domain-specific adapters during inference via model weight averaging to optimize performance on novel, unseen domains with excellent computational efficiency. However, the essential generalizability of this emerging weight-space adapter mixing mechanism on \textit{unseen, in-domain examples} remains unexplored. Thus, in this study, we conduct a comprehensive analysis to elucidate the generalizability of domain-specific adapter mixtures in in-domain evaluation. We also provide investigations into the inner workings of the mixture of domain-specific adapters by analyzing their weight signs, yielding critical analysis on the negative correlation between their fraction of weight sign difference and their mixtures' generalizability. |
| title | Generalizability of Mixture of Domain-Specific Adapters from the Lens of Signed Weight Directions and its Application to Effective Model Pruning |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2402.10639 |