CRAFT: Forgetting-Aware Intervention-Based Adaptation for Continual Learning
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_ | 1866915990739091456 |
|---|---|
| author | Hossen, Md Anwar Siddika, Fatema Munoz, Juan Pablo Roosta, Tanya Jannesari, Ali |
| author_facet | Hossen, Md Anwar Siddika, Fatema Munoz, Juan Pablo Roosta, Tanya Jannesari, Ali |
| contents | Large language models (LLMs) can acquire new capabilities through fine-tuning, but continual adaptation often leads to catastrophic forgetting. We propose CRAFT, a continual learning framework that avoids updating model weights by instead learning low-rank interventions on hidden representations. CRAFT proceeds in three stages: it first routes each task to a group of similar tasks based on output-distribution divergence; it then fine-tunes the model using a Kullback-Leibler (KL) divergence against the group's prior state, which directly controls forgetting and determines convergence; finally, it merges interventions for the updated task into the shared representation using the same KL signal. This design unifies routing, regularization, and merging through a single KL-based objective. CRAFT improves overall performance and reduces forgetting compared to strong LoRA-based approaches across multiple benchmarks and model scales, while remaining robust to task ordering. These results suggest that controlling adaptation in representation space, guided by output-space divergence, provides a scalable and principled approach to continual learning in LLMs. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_05732 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | CRAFT: Forgetting-Aware Intervention-Based Adaptation for Continual Learning Hossen, Md Anwar Siddika, Fatema Munoz, Juan Pablo Roosta, Tanya Jannesari, Ali Machine Learning Artificial Intelligence Large language models (LLMs) can acquire new capabilities through fine-tuning, but continual adaptation often leads to catastrophic forgetting. We propose CRAFT, a continual learning framework that avoids updating model weights by instead learning low-rank interventions on hidden representations. CRAFT proceeds in three stages: it first routes each task to a group of similar tasks based on output-distribution divergence; it then fine-tunes the model using a Kullback-Leibler (KL) divergence against the group's prior state, which directly controls forgetting and determines convergence; finally, it merges interventions for the updated task into the shared representation using the same KL signal. This design unifies routing, regularization, and merging through a single KL-based objective. CRAFT improves overall performance and reduces forgetting compared to strong LoRA-based approaches across multiple benchmarks and model scales, while remaining robust to task ordering. These results suggest that controlling adaptation in representation space, guided by output-space divergence, provides a scalable and principled approach to continual learning in LLMs. |
| title | CRAFT: Forgetting-Aware Intervention-Based Adaptation for Continual Learning |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2605.05732 |