1. Introduction In the early 20th century, a horse named Clever Hans gained international fame for his ability to solve arithmetic calcu-lations—including addition, division, fractions, and telling time—and …
SANA 1.5: Efficient Scaling of Training-Time and Inference-Time...
May 1, 2025 · A clever sampling trick that lets smaller models temporarily boost their capabilities These innovations allow SANA-1.5 to match or exceed the performance of systems like Stable Diffusion XL …
Anchor Frame Bridging for Coherent First-Last Frame Video Generation
According to the following review comments, our proposed Anchor Frame Bridging (AFB) holds significant potential implications for First-Last Frame Video Generation (FLF2V) in the context of …
Contrastive Learning Via Equivariant Representation - OpenReview
Sep 26, 2024 · TL;DR: This paper proposes CLeVER, a novel equivariant-based contrastive learning framework that improves training efficiency and robustness in downstream tasks by incorporating …
A-MemGuard: A Proactive Defense Framework for LLM-Based
Sep 16, 2025 · The core idea of using "consensus" is genuinely clever. It's a really smart, original way to use the agent's own data to spot an attack, rather than relying on some external filter that doesn't …
Provably Mitigating Overoptimization in RLHF: Your SFT Loss is...
Jun 19, 2024 · With a clever usage of the equivalence between reward models and the corresponding optimal policy, the algorithm features a simple objective that combines (i) a preference optimization …
EraseAnything: Enabling Concept Erasure in Rectified Flow Transformers
May 1, 2025 · This paper explores the problem of targeted concept erasure in deep learning models, aligning with broader discussions in the machine learning community on model interpretability, …
Secure Inference for Diffusion Models via Unconditional Scores
Sep 18, 2025 · I find the paper’s topic of efficient secure inference for diffusion models interesting, its proposed technique clever and its writing of high quality. The empirical work is clean and appears …
La RoSA: Enhancing LLM Efficiency via Layerwise Rotated Sparse...
May 1, 2025 · We use a clever technique that involves rotating the data within each layer of the model, making it easier to identify and keep only the most important parts for processing. This ensures that …
Cross-Lingual Multi-Hop Knowledge Editing - OpenReview
Jun 15, 2024 · Following this, we propose a significantly improved system for cross-lingual multi-hop knowledge editing, CLeVer-CKE. CLeVer-CKE is based on a retrieve, verify and generate knowledge …