Beyond Higher Rank: Token-wise Input-Output Projections for Efficient Low-Rank Adaptation

TL;DR

This paper introduces TopLoRA, a token-wise low-rank adaptation method that dynamically adjusts weights for each input token, significantly improving fine-tuning efficiency and performance in large language models.

Contribution

TopLoRA extends standard LoRA by enabling token-wise input-output projections without increasing rank, enhancing model adaptation granularity and effectiveness.

Findings

TopLoRA outperforms standard LoRA across multiple models and datasets.

Token-wise projections improve fine-tuning efficiency and accuracy.

The method does not increase the overall rank of LoRA weights.

Abstract

Low-rank adaptation (LoRA) is a parameter-efficient fine-tuning (PEFT) method widely used in large language models (LLMs). LoRA essentially describes the projection of an input space into a low-dimensional output space, with the dimensionality determined by the LoRA rank. In standard LoRA, all input tokens share the same weights and undergo an identical input-output projection. This limits LoRA's ability to capture token-specific information due to the inherent semantic differences among tokens. To address this limitation, we propose Token-wise Projected Low-Rank Adaptation (TopLoRA), which dynamically adjusts LoRA weights according to the input token, thereby learning token-wise input-output projections in an end-to-end manner. Formally, the weights of TopLoRA can be expressed as $B\Sigma_X A$, where $A$ and $B$ are low-rank matrices (as in standard LoRA), and $\Sigma_X$ is a diagonal matrix generated from each input token $X$. Notably, TopLoRA does not increase the rank of LoRA weights but achieves more granular adaptation by learning token-wise LoRA weights (i.e., token-wise input-output projections). Extensive experiments across multiple models and datasets demonstrate that TopLoRA consistently outperforms LoRA and its variants. The code is available at https://github.com/Leopold1423/toplora-neurips25.