AI Paper Atlas

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

Turning Back Without Forgetting: Selective Backward Refinement for Parameter-Efficient Continual Learning

Anushka Tiwari, Kaiyi Ji

ICML 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 0 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 task correlation selection (ablation)SABER (T5-Large) · Long SequenceAP: 77.59 -> 80.33Improved
add task correlation selection (ablation)SABER (T5-Large) · Long SequenceBWT: -1.78 -> 1.95Improved
add constrained backward refinement (ablation)SABER (T5-Large) · Long SequenceAP: 78.84 -> 80.33Improved
add constrained backward refinement (ablation)SABER (T5-Large) · Long SequenceBWT: -0.75 -> 1.95Improved
add task correlation selection (ablation)SABER (T5-Large) · SuperNIBWT: -0.84 -> 2.09Improved
add constrained backward refinement (ablation)SABER (T5-Large) · SuperNIAP: 41.33 -> 42.48Improved
add both components (ablation)SABER (T5-Large) · Long SequenceAP: 77.02 -> 80.33Improved
add task correlation selection (ablation)SABER (LLaMA) · Long SequenceAP: 81.26 -> 83.42Improved
add constrained backward refinement (ablation)SABER (LLaMA) · Long SequenceBWT: -0.87 -> 1.75Improved
add both components (ablation)SABER (LLaMA) · SuperNIAP: 45.12 -> 48.95Improved
orthogonal update instead of unconstrained backward update (ablation)SABER backward prompt refinement · Long Sequence task pairsavg Delta Acc: +0.007 -> +0.0215Improved
orthogonal update instead of same-subspace update (ablation)SABER backward prompt refinement · Long Sequence task pairsavg Delta Acc: -0.007 -> +0.0215Improved

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