ELROND: Exploring and decomposing intrinsic capabilities of diffusion models

arXiv:2602.10216 · cs.LG · Submitted 2026-02-10 · Read on arXiv

cs.LG

Submitted: 2026-02-10

Updated: 2026-10-07

Terminology

• Diffusion Model — 2× in this paper · explained in 9 episodes
• Diffusion Models — 1× in this paper · explained in 6 episodes
• Mode Collapse — 1× in this paper · explained in 2 episodes
• Steering — 1× in this paper · explained in 2 episodes

Sources

• LEDITS++: Limitless Image Editing using Text-to-Image Models
• BatchTopK Sparse Autoencoders
• SAeUron: Interpretable Concept Unlearning in Diffusion Models with Sparse Autoencoders
• From What to How: Attributing CLIP's Latent Components Reveals Unexpected Semantic Reliance
• DreamSim: Learning New Dimensions of Human Visual Similarity using Synthetic Data
• Prompt-to-Prompt Image Editing with Cross Attention Control
• TIDE : Temporal-Aware Sparse Autoencoders for Interpretable Diffusion Transformers in Image Generation
• SAEdit: Token-level control for continuous image editing via Sparse AutoEncoder
• $\textit{Revelio}$: Interpreting and leveraging semantic information in diffusion models
• Diffusion Models already have a Semantic Latent Space
• Faster Diffusion via Temporal Attention Decomposition
• Sparse Autoencoders, Again?
• The Origins of Representation Manifolds in Large Language Models
• Null-text Inversion for Editing Real Images using Guided Diffusion Models
• Sparse Autoencoders Learn Monosemantic Features in Vision-Language Models
• The Linear Representation Hypothesis and the Geometry of Large Language Models
• SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis
• Controlling Language and Diffusion Models by Transporting Activations
• One-Step is Enough: Sparse Autoencoders for Text-to-Image Diffusion Models
• Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks