Metraipolis › archive › dc_dd622ef4c08b

TaRA: Training-Aware Low-Rank Adaptation Initialization — arXiv:2609.02639

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.02639
License (as archived)
CC0-1.0
Archived
2026-10-17T03:38:26.355Z
SHA-256 of the archived text
dd6139fdbed3d4ddd02c01aed944ca926e1d89ea0aede68e9f815e14376bae57
Stable URL of this page
https://metraipolis.com/documento/dc_dd622ef4c08b

How to cite

TaRA: Training-Aware Low-Rank Adaptation Initialization — arXiv:2609.02639 — arXiv (en), https://arxiv.org/abs/2609.02639 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dd622ef4c08b (Metraipolis archive dc_dd622ef4c08b, sha256 dd6139fdbed3d4ddd02c01aed944ca926e1d89ea0aede68e9f815e14376bae57)
BibTeX
@misc{dcdd622ef4c08b,
  title = {TaRA: Training-Aware Low-Rank Adaptation Initialization — arXiv:2609.02639},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.02639},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dd622ef4c08b; sha256 dd6139fdbed3d4ddd02c01aed944ca926e1d89ea0aede68e9f815e14376bae57},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.02639. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Low-Rank Adaptation (LoRA) has become a de facto standard for parameter-efficient fine-tuning (PEFT), yet its performance is highly sensitive to initialization due to the information bottleneck imposed by low-rank decomposition. Existing approaches attempt to construct high-quality LoRA initializations by exploiting principal components of pretrained weights, activations, or gradients. However, these methods do not directly account for the training dynamics of the full-rank model. In this paper, we propose Training-aware Low-Rank Adaptation Initialization (TaRA), a method that initializes LoRA such that the gradients induced by the low-rank factors closely approximate the gradient of the corresponding full-rank weight matrix. Derived from a mathematical formulation, TaRA improves gradient fidelity at the start of training while introducing negligible computational overhead. Across diverse and challenging fine-tuning tasks, TaRA consistently outperforms prior state-of-the-art methods, establishing a simple, robust, and scalable solution for effective LoRA initialization.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-03 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.02639 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-17T03:13:12.562Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-03?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).