Hold-One-Shot-Out (HOSO) for Validation-Free Few-Shot CLIP Adapters
This work addresses the problem of hyperparameter tuning in few-shot CLIP adaptation for researchers and practitioners, making the adaptation process more strictly few-shot and practical.
This paper introduces Hold-One-Shot-Out (HOSO), a validation-free method for learning the blending ratio in few-shot CLIP adaptation. HOSO-Adapter, which uses this method, improves upon the CLIP-Adapter baseline by over 4 percentage points on average across 11 datasets, and even outperforms it in 8- and 16-shot settings when the baseline's blending ratio is optimally selected on the test set.
In many CLIP adaptation methods, a blending ratio hyperparameter controls the trade-off between general pretrained CLIP knowledge and the limited, dataset-specific supervision from the few-shot cases. Most few-shot CLIP adaptation techniques report results by ablation of the blending ratio on the test set or require additional validation sets to select the blending ratio per dataset, and thus are not strictly few-shot. We present a simple, validation-free method for learning the blending ratio in CLIP adaptation. Hold-One-Shot-Out (HOSO) presents a novel approach for CLIP-Adapter-style methods to compete in the newly established validation-free setting. CLIP-Adapter with HOSO (HOSO-Adapter) learns the blending ratio using a one-shot, hold-out set, while the adapter trains on the remaining few-shot support examples. Under the validation-free few-shot protocol, HOSO-Adapter outperforms the CLIP-Adapter baseline by more than 4 percentage points on average across 11 standard few-shot datasets. Interestingly, in the 8- and 16-shot settings, HOSO-Adapter outperforms CLIP-Adapter even with the optimal blending ratio selected on the test set. Ablation studies validate the use of a one-shot hold-out mechanism, decoupled training, and improvements over the naively learnt blending ratio baseline. Code is released here: https://github.com/chris-vorster/HOSO-Adapter