AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

LAVA: A Unified Framework for Finetuning Language and Vision Models

Daorui Ding, Fanhua Shang, Tiancan Feng, Junkang Liu, Hongying Liu

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory

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 regularization to attention LoRA (ablation)SDXL attention LoRA · pokemon-blip-captionsCLIP score: 29.28 -> 29.42Improved
add orthogonal regularization to reshape-free conv adapter (ablation)LAVA on SDXL · pokemon-blip-captionsCLIP score: 29.62 -> 29.64Improved
add orthogonal regularization to both attention and conv (ablation)LAVA on SDXL · pokemon-blip-captionsCLIP score: 29.62 -> 29.96Improved
tune conv layers with reshape-involved LoRA instead of leaving them frozen (ablation)SDXL attention LoRA · pokemon-blip-captionsCLIP score: 29.28 -> 28.10Hurt
replace reshape-involved LoRA with reshape-free tensor decomposition for conv (ablation)SDXL full U-Net · pokemon-blip-captionsCLIP score: 28.10 -> 29.62Improved
tune conv layers reshape-free in addition to attention (ablation)SDXL attention LoRA · pokemon-blip-captionsCLIP score: 29.28 -> 29.62Improved
reshape-free instead of reshape-involved low-rank conv finetuning (ablation)six CNNs · CIFAR10accuracy: around 10% higher with fewer trainable parametersImproved
use LAVA instead of LoRALLaMA2-7b · commonsense reasoningaverage accuracy: 77.6 -> 79.5Improved
use LAVA instead of LoRARoBERTa-base · GLUEaverage score: 84.7 -> 85.9Improved
use LAVA instead of DoRALLaMA3-3b · commonsense reasoningaverage accuracy: 79.5 -> 79.8No clear effect
use LAVA instead of LoRA for convolution layers of Depth-AnythingDepth-Anything-large · NYU Depth v2delta1: 0.937 -> 0.972Improved
use LAVA instead of full finetuningDepth-Anything-large · NYU Depth v2delta1: 0.984 -> 0.972Hurt

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