AI Paper Atlas

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

LANCE: Low Rank Activation Compression for Efficient On-Device Continual Learning

Marco Apolinario, Kaushik Roy

ECCV 2026 · Continual learning › Continual fine-tuning of LLMs · approach: Low-rank and model-merging adaptation · 3 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
fine-tune 4 layers with LANCE instead of 2 layers (ablation)LANCE · ImageNet-1k Split Baccuracy: 59.18 -> 62.20Improved
replace full BP with LANCE (4 layers)ResNet18 · ImageNet-1k Split Baccuracy / memory: 64.28 -> 62.20 accuracy, 61.25 -> 16.59 MBMixed
replace full BP with LANCE (2 layers)ResNet34 · CIFAR-100 (last 2 layers)accuracy: 76.19 -> 74.89Hurt
replace iterative HOSVD with one-shot LANCEResNet18 · CIFAR-100 (last 4 layers)accuracy: 75.35 -> 75.7No clear effect
replace ASI with LANCEResNet18 · CIFAR-100 (last 4 layers)accuracy: 71.93 -> 75.7Improved
replace ASI with LANCEMobileNetV2 · CUB-200 (last 4 layers)accuracy: 50.69 -> 61.21Improved
replace iterative HOSVD with LANCEMCUNet · Raspberry Pi 3B+ latencytraining time: 2.1x-4.4x slower -> LANCE lowestImproved
replace ASI with LANCEMCUNet · Raspberry Pi 3B+ latencytraining time: 19-76% slower -> LANCE lowestImproved
replace full BP with LANCEMCUNet · Raspberry Pi 3B+ latencytraining time: BP 3-9% slowerImproved
raise energy threshold epsilon (ablation)LANCE · CUB-200accuracy / memoryMixed
increase number of calibration batches N (ablation)LANCE · CIFAR-100accuracyNo clear effect
lower energy threshold epsilon (stronger compression) (ablation)LANCE · gradient angle vs full BPangle: within ~70 degrees at up to two orders of magnitude compressionHurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 2 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