AI Paper Atlas

Research map / Continual learning / Continual fine-tuning of LLMs

Gradient Projection for Continual Parameter-Efficient Tuning

Jingyang Qiao, Zhizhong Zhang, Xin Tan, Yanyun Qu, Wensheng Zhang, Zhi Han, Yuan Xie

TPAMI 2025 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 15 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add orthogonal gradient projection (PEGP) to Adapter-10LAE Adapter-10 (ViT) · 10-Split-CIFAR100 (class-incremental)average accuracy: 85.10 -> 85.43Improved
add gradient projection to Adapter-5 on ImageNet-RLAE Adapter-5 (ViT) · 10-Split-ImageNet-R (class-incremental)forgetting: 9.62 -> 5.62Improved
add gradient projection to the plain Adapter-10 baselineAdapter-10 baseline · 10-Split-CIFAR100 (class-incremental)average accuracy: 79.17 -> 85.43Improved
add gradient projection under one-epoch online trainingLAE Adapter-5 · 10-Split-CIFAR100 (online class-incremental)forgetting: 11.44 -> 6.82Improved
add gradient projection to LoRA-5 in online settingLAE LoRA-5 · 10-Split-CIFAR100 (online class-incremental)average accuracy: 78.30 -> 83.21Improved
add PEGP to CLIP prompt tuning (CoOP)CoOP Prompt (CLIP) · 10-Split-CIFAR100 (class-incremental, CLIP)average accuracy: 73.76 -> 82.36Improved
add PEGP to CLIP linear adapterCoOP Adapter (CLIP) · 10-Split-CIFAR100 (class-incremental, CLIP)average accuracy: 63.95 -> 70.37Improved
add PEGP on task-incremental CLIP, trading plasticity for stabilityCLIP-EWC Prompt · 10-Split-CIFAR100 (task-incremental, CLIP)average accuracy: 94.42 -> 93.99Mixed
add PEGP to Adapter-10 on DomainNetLAE Adapter-10 · 5-Split-DomainNet (domain-incremental)forgetting: 10.38 -> 6.84Improved
add PEGP to CLIP prompt on cross-modality BITMCoOP Prompt · 5-Split-BITMaverage accuracy: 52.73 -> 62.66Improved
add PEGP to CLIP before continual training on CIFAR100, test zero-shot ImageNet100CLIP baseline · zero-shot CIFAR100->ImageNet100zero-shot accuracy: 70.56 -> 72.84Improved
add PEGP, zero-shot on CIFAR100 after ImageNet100 trainingCLIP baseline · zero-shot ImageNet100->CIFAR100zero-shot accuracy: 52.85 -> 59.91Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy