Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Dynamic Scaled Gradient Descent for Stable Fine-Tuning for Classifications

About

Fine-tuning pretrained models has become a standard approach to adapting pretrained knowledge to improve the accuracy on new sparse, imbalance datasets. However, issues arise when optimization falls into a collapsed state, where the model gets stuck, leading to degraded performance and unstable training. One possible reason for this is the cancellation of gradients across training examples. To address this problem, we propose a novel algorithm, dynamic scaled gradient descent (\mName), that directly modifies the gradients returned by training examples, specifically, scaling down the gradients of correctly classified examples using a dynamic scaler. This strategy offers both theoretical and empirical advantages in improving training stability. Experiments on a variety of benchmark datasets, spanning multiple tasks and large pretrained models, demonstrate that our method consistently reduces performance variance and surpasses the accuracy of existing approaches.

Nghia Bui, Lijing Wang• 2026

Related benchmarks

TaskDatasetResultRank
Image ClassificationCIFAR-10 Long-Tailed
Accuracy97.15
71
Image ClassificationCIFAR-100 Long-Tailed
Accuracy87.72
71
Sequence ClassificationGLUE & SuperGLUE (MultiRC, COPA, RTE, BoolQ, MRPC, CoLA)
MultiRC Accuracy85.44
24
Image ClassificationCIFAR-100 step
Accuracy87.93
12
Multi-task ClassificationGLUE MRPC, RTE, CoLA (test, val)
MRPC Accuracy88.95
12
Image ClassificationCIFAR-10 step
Accuracy96.95
12
Showing 6 of 6 rows

Other info

Follow for update