Post-Optimization Adaptive Rank Allocation for LoRA
About
Exponential growth in the scale of modern foundation models has led to the widespread adoption of Low-Rank Adaptation (LoRA) as a parameter-efficient fine-tuning technique. However, standard LoRA implementations disregard the varying intrinsic dimensionality of model layers and enforce a uniform rank, leading to parameter redundancy. We propose Post-Optimization Adaptive Rank Allocation (PARA), a data-free compression method for LoRA that integrates seamlessly into existing fine-tuning pipelines. PARA leverages Singular Value Decomposition to prune LoRA ranks using a global threshold over singular values across all layers. This results in non-uniform rank allocation based on layer-wise spectral importance. As a post-hoc method, PARA circumvents the training modifications and resulting instabilities that dynamic architectures typically incur. We empirically demonstrate that PARA reduces parameter count by 75-90\% while preserving the predictive performance of the original, uncompressed LoRA across multiple vision and language benchmarks. Code will be published upon acceptance.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Mathematical Reasoning | GSM8K (test) | Accuracy78.2 | 816 | |
| Natural Language Understanding | GLUE | SST-293.46 | 551 | |
| Classification | Cars | Accuracy84.58 | 492 | |
| Image Classification | Pets | Accuracy89.97 | 308 | |
| Image Classification | CIFAR10 | Accuracy (%)97.89 | 282 | |
| Commonsense Reasoning | Commonsense Reasoning (BoolQ, PIQA, SIQA, HellaS., WinoG., ARC-e, ARC-c, OBQA) | BoolQ Accuracy84.4 | 223 | |
| Image Classification | Flowers | Top-1 Acc86.03 | 110 | |
| Image Classification | Food | Accuracy86.4 | 91 | |
| Mathematical Reasoning | MATH (test) | Accuracy45.5 | 6 |