AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

The Inter-Intra Modal Measure: A Predictive Lens on Fine-Tuning Outcomes in Vision-Language Models

Laura Niss, Kevin Vogt-Lowell, Theodoros Tsiligkaridis

ICCV 2025 · Vision-language and multimodal › CLIP and zero-shot · approach: Parameter-efficient visual tuning · 2 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
use IIMM instead of GBC to predict gain over zero-shot error (Kendall tau)IIMM · CoCa, EVA-02, CLIP, SigLIPKendall tau: CoCa 0.78 -> 0.89, EVA-02 0.72 -> 0.83, CLIP 0.78 -> 0.78, SigLIP 0.83 -> 0.78Mixed
use IIMM instead of TransRate for gain over zero-shot errorIIMM · CoCaKendall tau: 0.67 -> 0.89Improved
use IIMM instead of NCTIIIMM · EVA-02Kendall tau: -0.59 -> 0.83Improved
use IIMM instead of H-ScoreIIMM · EVA-02Kendall tau: -0.78 -> 0.83Improved
use IIMM instead of GBC to predict fine-tuned accuracyIIMM · CoCaKendall tau: 0.61 -> 0.72Improved
use IIMM instead of GBC to predict fine-tuned accuracyIIMM · SigLIPKendall tau: 0.67 -> 0.61No clear effect
fit IIMM linearly instead of GBC against log gainIIMM · CLIPR^2: 0.59 -> 0.92Improved
use image-image similarity (intra-images) term alone instead of intra-texts (ablation)IIMM component · CoCaR^2: 0.37 -> 0.88Improved
use incorrect-alignment term alone instead of intra-texts term (ablation)IIMM component · SigLIPR^2: 0.17 -> 0.80Improved
combine inter and intra terms (alpha 0.5) rather than inter only (alpha 0.00) (ablation)IIMM · CLIPPearson r: 0.79 -> 0.96Improved
combine inter and intra terms (alpha 0.5) rather than intra only (alpha 1.00) (ablation)IIMM · CLIPPearson r: 0.90 -> 0.96Improved
apply IIMM to CLIP-Adapter instead of other PEFT methodsIIMM · CLIP-AdapterPearson r: 0.76 -> 0.96 for Attention-Weight Tuning at alpha 0.5Hurt

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