AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

Early Estimation of Language to Latent Alignment in Diffusion Models

Vasco Ramos, Regev Cohen, Idan Szpektor, Joao Magalhaes

ECCV 2026 · Image generation › Fast diffusion sampling · approach: Masked diffusion language model decoding · 4 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 CLIP image encoder on noisy latents (NoisyCLIP)CLIP L14 · Noisy-Conceptual-Captions latent 21-30Factual consistency R@1: 0.145 -> 0.506Improved
fine-tune CLIP image encoder on noisy latents (NoisyCLIP)CLIP L14 · Noisy-GenAI-Bench latent 21-30 Best-of-NVQAScore: 0.657 -> 0.709Improved
noise-aware fine-tuning of SigLIP (NoisySigLIP)SigLIP L16 · Noisy-GenAI-Bench latent 21-30 Best-of-NVQAScore: 0.652 -> 0.718Improved
noise-aware fine-tuning of SigLIP (NoisySigLIP)SigLIP L16 · Noisy-Conceptual-Captions latent 21-30Factual consistency R@1: 0.147 -> 0.194Improved
early-stopping BoN at 150 steps instead of 300-step CLIP selectionNoisyCLIP · GenAI-Bench basic, best-of-6VQAScore: 0.85 -> 2% drop at 50% costMixed
add NoisyCLIP selection at latent 20 for BoNNoisyCLIP · GenAI-Bench BoN at latent 20VQAScore: up to 2%Improved
train image encoder only (add image encoder training) (ablation)CLIP · GenAI Basic, latent 21-30VQAScore: 0.821 -> 0.836Improved
train image encoder only (add image encoder training) (ablation)CLIP · GenAI Advanced, latent 21-30VQAScore: 0.633 -> 0.646Improved
also train text encoder in addition to image encoder (ablation)NoisyCLIP · GenAI Advanced, image dimensionVQAScore: 0.664 -> 0.657Hurt
also train text encoder in addition to image encoder (ablation)NoisyCLIP · GenAI Basic, latent 21-30VQAScore: 0.836 -> 0.835No clear effect
scale encoder from ViT-B/16 to ViT-L/14 (ablation)NoisyCLIP · GenAI Basic, latent 21-30VQAScore: 0.829 -> 0.836Improved
scale encoder from ViT-B/16 to ViT-L/14 (ablation)NoisyCLIP · GenAI Advanced, image dimensionVQAScore: 0.652 -> 0.664Improved

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