AI Paper Atlas

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

Post-hoc Probabilistic Vision-Language Models

Anton Baumann, Rui Li, Marcus Klasson, Santeri Mentu, Shyamgopal Karthik, Zeynep Akata, Arno Solin, Martin Trapp

ICLR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Test-time adaptation of VLMs · 14 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
apply BayesVLM post-hoc Laplace uncertainty to CLIPOpenCLIP ViT-B-32 · CIFAR-100 zero-shotECE (%): 6.31 -> 1.92Improved
apply BayesVLM post-hoc Laplace uncertainty to CLIPOpenCLIP ViT-B-32 · UCF101 zero-shotECE (%): 11.52 -> 3.57Improved
apply BayesVLM post-hoc Laplace uncertainty to CLIPOpenCLIP ViT-B-32 · Flowers-102 zero-shotECE (%): 6.59 -> 4.22Improved
apply BayesVLM post-hoc Laplace uncertainty to CLIPOpenCLIP ViT-B-32 · Flowers-102 zero-shotaccuracy (%): 68.99 -> 68.87No clear effect
apply BayesVLM post-hoc Laplace uncertainty to CLIPOpenCLIP ViT-B-32 · UCF101 zero-shotNLPD: 1.59 -> 1.44Improved
use BayesVLM instead of test-time augmentation (TTA)OpenCLIP ViT-B-32 · CIFAR-10 zero-shotaccuracy (%): 88.54 -> 93.62Hurt
add ProbCosine to PCME++ probabilistic embeddingsPCME++ · Flowers-102 zero-shotECE (%): 8.81 -> 2.79Improved
add ProbCosine to PCME++ probabilistic embeddingsPCME++ · SUN397 zero-shotECE (%): 26.03 -> 19.41Improved
add ProbCosine to PCME++ probabilistic embeddingsPCME++ · Flowers-102 zero-shotaccuracy (%): 40.59 -> 40.43No clear effect
estimate the Hessian on proxy dataset CC12M instead of LAION-400M (ablation)BayesVLM · Flowers-102 zero-shotaccuracy (%): 68.87 -> 68.12No clear effect
estimate the Hessian on proxy dataset CC12M instead of LAION-400M (ablation)BayesVLM · SUN397 zero-shotECE (%): 2.06 -> 3.89Mixed
use BayesVLM uncertainties (EPIG, BALD) for support-set selection instead of random or entropyCLIP fine-tuning · OfficeHome and ImageNet-variant active learningweighted accuracy and NLPDImproved

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