FlexRank: Nested Low-Rank Knowledge Decomposition for Adaptive Model Deployment
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918531589734400 |
|---|---|
| author | Zaccone, Riccardo Laskaridis, Stefanos Ciccone, Marco Horváth, Samuel |
| author_facet | Zaccone, Riccardo Laskaridis, Stefanos Ciccone, Marco Horváth, Samuel |
| contents | The growing scale of deep neural networks, encompassing large language models (LLMs) and vision transformers (ViTs), has made training from scratch prohibitively expensive and deployment increasingly costly. These models are often used as computational monoliths with fixed cost, hindering adaptive deployment across different cost budgets.We argue that nested components, ordered by importance, can be extracted from pretrained models and selectively activated within the available computational budget. To this end, our proposed FlexRank method leverages low-rank weight decomposition with nested, importance-based consolidation to extract submodels of increasing capabilities. Our approach enables a ``train-once, deploy-everywhere'' paradigm offering a graceful trade-off between cost and performance without training from scratch for each budget - advancing practical deployment of large models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_02680 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | FlexRank: Nested Low-Rank Knowledge Decomposition for Adaptive Model Deployment Zaccone, Riccardo Laskaridis, Stefanos Ciccone, Marco Horváth, Samuel Machine Learning The growing scale of deep neural networks, encompassing large language models (LLMs) and vision transformers (ViTs), has made training from scratch prohibitively expensive and deployment increasingly costly. These models are often used as computational monoliths with fixed cost, hindering adaptive deployment across different cost budgets.We argue that nested components, ordered by importance, can be extracted from pretrained models and selectively activated within the available computational budget. To this end, our proposed FlexRank method leverages low-rank weight decomposition with nested, importance-based consolidation to extract submodels of increasing capabilities. Our approach enables a ``train-once, deploy-everywhere'' paradigm offering a graceful trade-off between cost and performance without training from scratch for each budget - advancing practical deployment of large models. |
| title | FlexRank: Nested Low-Rank Knowledge Decomposition for Adaptive Model Deployment |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2602.02680 |