반응형 전체 글9 [논문리뷰] AffordanceLLM : Grounding Affordance from Vision Language Models AffordanceLLM: Grounding Affordance from Vision Language ModelsAffordance grounding refers to the task of finding the area of an object with which one can interact. It is a fundamental but challenging task, as a successful solution requires the comprehensive understanding of a scene in multiple aspects including detecarxiv.org이번에 리뷰할 논문은 AffordanceLLM이다. 그러면 한번 시작해 보자.AbstractAffordance groundin.. 2025. 4. 30. [논문리뷰] InfiniteYou : Flexible Photo Recrafting While Preserving Your Identity 이번에 리뷰를 작성할 논문은 지난달 말에 나온 Photo Recrafting 논문인 InfiniteYou이다. InfiniteYou: Flexible Photo Recrafting While Preserving Your IdentityAchieving flexible and high-fidelity identity-preserved image generation remains formidable, particularly with advanced Diffusion Transformers (DiTs) like FLUX. We introduce InfiniteYou (InfU), one of the earliest robust frameworks leveraging DiTs for thisarxiv.orgAb.. 2025. 4. 29. [논문리뷰] DDT : Decoupled Diffusion Trasnformer DDT: Decoupled Diffusion TransformerDiffusion transformers have demonstrated remarkable generation quality, albeit requiring longer training iterations and numerous inference steps. In each denoising step, diffusion transformers encode the noisy inputs to extract the lower-frequency semanticarxiv.org오늘 리뷰할 논문은 DDT라고 해서, 기존의 DiT / SiT 등 Transformer 기반의 Diffusion Process에서의 한계점을 극복하기 위한 논문이라고 할 수 .. 2025. 4. 17. [논문리뷰] EXAONE Deep: Reasoning Enhanced Language Models EXAONE Deep: Reasoning Enhanced Language ModelsWe present EXAONE Deep series, which exhibits superior capabilities in various reasoning tasks, including math and coding benchmarks. We train our models mainly on the reasoning-specialized dataset that incorporates long streams of thought processes. Evaluarxiv.org오늘(2025.3.18) LG Research Team에서 EXAONE Deep을 발표하였다. 이는 현재 트렌드에 맞추어 Reasoning 성능을 대폭 향.. 2025. 3. 18. [논문 리뷰] R.A.C.E : Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion Model R.A.C.E.: Robust Adversarial Concept Erasure for Secure Text-to-Image Diffusion ModelIn the evolving landscape of text-to-image (T2I) diffusion models, the remarkable capability to generate high-quality images from textual descriptions faces challenges with the potential misuse of reproducing sensitive content. To address this critical issarxiv.org이번에는 컨택드렸던 자대 교수님의 논문을 가져와봤다. 내가 주로 관심가지고 연구하는 분.. 2025. 3. 17. [논문 리뷰] AoT : Atom of Thoughts for Markov LLM Test-Time Scaling Atom of Thoughts for Markov LLM Test-Time ScalingLarge Language Models (LLMs) achieve superior performance through training-time scaling, and test-time scaling further enhances their capabilities by conducting effective reasoning during inference. However, as the scale of reasoning increases, existing tearxiv.orgChain-of-Thought의 새로운 지평이라고 하여 AoT, Atom of Thoughts가 새롭게 제안되었다. Question에 대해 Decomp.. 2025. 3. 11. [논문리뷰] Fractal Generative Models Fractal Generative ModelsModularization is a cornerstone of computer science, abstracting complex functions into atomic building blocks. In this paper, we introduce a new level of modularization by abstracting generative models into atomic generative modules. Analogous to fractalsarxiv.org한국 시간으로 25일에 최초로 나온 따끈따끈한 논문을 들고 왔다. 웬만하면 이런 짧은 제목의 논문은 잘 보지 않는 편인데, MASK R-CNN, ResNet, Focal Loss, FPN 등의 .. 2025. 2. 26. [Paper Review] StyleGAN : A Style-Based Generator Architecture for Generative Adversarial Networks A Style-Based Generator Architecture for Generative Adversarial NetworksWe propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identitarxiv.org* 이 논문 리뷰는 StyleGAN 시리즈 논문에 대한 세미나를 준비하면서 작성된 글이기에, 다소 얕고 주인장.. 2025. 2. 25. [Paper Review] DeepSeek-R1 : Incentivizing Reasoning Capability in LLMs via Reinforcement Learning DeepSeek에서 또 새로운 사고를 쳤다. 기존 deepseek-v3만 해도 충분히 파라미터 대비 성능이 잘 나온다고 해서 말이 상당히 많은 상태였는데, 이제는 더 작은 파라미터로 o1과 거의 비슷하거나 그 이상의 성능을 내버리는 모델이 나오고 말았다. 일단 논문을 보고 올거라면 아래의 논문을 보기를 바란다. DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement LearningWe introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinf.. 2025. 1. 29. 이전 1 다음 반응형
