티스토리

JunHan's AI Factory
검색하기

블로그 홈

JunHan's AI Factory

junhan-ai.tistory.com/m

인공지능 정리 및 개인적인 블로그

구독자
2
방명록 방문하기
공지 제 블로그에 방문하신 여러분 환영합니⋯ 모두보기

주요 글 목록

  • Instance Normalization: The Missing Ingredient for Fast Stylization 대회가 마무리 되어가고 있다보니 생각보다 리뷰를 못하는 중이다. monai에서 사용하는 것들이 기억이 안나서 좀 다시 읽고 있는 형태. 무엇을 잘못했는지 대충 이제 보이는 느낌이기도 하다. https://arxiv.org/abs/1607.08022 Instance Normalization: The Missing Ingredient for Fast StylizationIt this paper we revisit the fast stylization method introduced in Ulyanov et. al. (2016). We show how a small change in the stylization architecture results in a significant qualitative impro.. 공감수 3 댓글수 3 2025. 1. 30.
  • Group Normalization https://arxiv.org/abs/1803.08494 Group NormalizationBatch Normalization (BN) is a milestone technique in the development of deep learning, enabling various networks to train. However, normalizing along the batch dimension introduces problems --- BN's error increases rapidly when the batch size becomes smallarxiv.org 초록(Abstract)Batch Normalization(BN)은 다양한 네트워크가 학습될 수 있도록 한 딥러닝 발전의 이정표적인 기법이다. 그.. 공감수 1 댓글수 2 2025. 1. 24.
  • SegFormer3D: an Efficient Transformer for 3D Medical Image Segmentation https://github.com/OSUPCVLab/SegFormer3D/tree/653faa6b44c67cebd27a02de5fe08ee4072dd230 GitHub - OSUPCVLab/SegFormer3D: Official Implementation of SegFormer3D: an Efficient Transformer for 3D Medical Image SegmentatiOfficial Implementation of SegFormer3D: an Efficient Transformer for 3D Medical Image Segmentation (CVPRW 2024) - OSUPCVLab/SegFormer3Dgithub.com https://arxiv.org/abs/2404.10156 SegF.. 공감수 2 댓글수 2 2025. 1. 21.
  • AutoVFX: Physically Realistic Video Editing from Natural Language Instructions https://arxiv.org/abs/2411.02394 AutoVFX: Physically Realistic Video Editing from Natural Language InstructionsModern visual effects (VFX) software has made it possible for skilled artists to create imagery of virtually anything. However, the creation process remains laborious, complex, and largely inaccessible to everyday users. In this work, we present AutoVFX, aarxiv.org 개요(Abstract)현대의 시각효과(.. 공감수 6 댓글수 2 2025. 1. 20.
  • NeRFactor: neural factorization of shape and reflectance under an unknown illumination https://dl.acm.org/doi/10.1145/3478513.3480496 NeRFactor: neural factorization of shape and reflectance under an unknown illumination: ACM Transactions on Graphics: Vol 40, NoWe address the problem of recovering the shape and spatially-varying reflectance of an object from multi-view images (and their camera poses) of an object illuminated by one unknown lighting condition. This enables the rend.. 공감수 1 댓글수 2 2025. 1. 17.
  • Titans: Learning to Memorize at Test Time https://arxiv.org/abs/2501.00663 Titans: Learning to Memorize at Test TimeOver more than a decade there has been an extensive research effort on how to effectively utilize recurrent models and attention. While recurrent models aim to compress the data into a fixed-size memory (called hidden state), attention allows attending toarxiv.org 댓글에 논문 추천도를 확인하는 것을 추천 개요10년이 넘는 기간 동안 순환 모델(recurrent mode.. 공감수 3 댓글수 3 2025. 1. 16.
  • MedNeXt: Transformer-driven Scaling of ConvNets for Medical Image Segmentation https://arxiv.org/abs/2303.09975 MedNeXt: Transformer-driven Scaling of ConvNets for Medical Image SegmentationThere has been exploding interest in embracing Transformer-based architectures for medical image segmentation. However, the lack of large-scale annotated medical datasets make achieving performances equivalent to those in natural images challenging. Convolarxiv.org https://github.com/MI.. 공감수 1 댓글수 2 2025. 1. 13.
  • Rho-1: Not All Tokens Are What You Need https://arxiv.org/abs/2404.07965 Rho-1: Not All Tokens Are What You NeedPrevious language model pre-training methods have uniformly applied a next-token prediction loss to all training tokens. Challenging this norm, we posit that "9l training". Our initial analysis examines token-level training dynamics of language model, revearxiv.org 초록기존의 언어 모델 사전 학습 기법은 모든 학습 토큰에 동일하게 다음 토큰 예측 손실을 적용해 왔습니다. .. 공감수 2 댓글수 3 2025. 1. 10.
  • mochi-1-preview https://huggingface.co/genmo/mochi-1-preview genmo/mochi-1-preview · Hugging FaceMochi 1 Blog | Hugging Face | Playground | Careers A state of the art video generation model by Genmo. https://github.com/user-attachments/assets/4d268d02-906d-4cb0-87cc-f467f1497108 Overview Mochi 1 preview is an open state-of-the-art video generation modhuggingface.cohttps://github.com/genmoai/mochi GitHub - genmo.. 공감수 2 댓글수 3 2025. 1. 10.
  • Ferret-UI 2: Mastering Universal User Interface Understanding Across Platforms https://arxiv.org/abs/2410.18967 Ferret-UI 2: Mastering Universal User Interface Understanding Across PlatformsBuilding a generalist model for user interface (UI) understanding is challenging due to various foundational issues, such as platform diversity, resolution variation, and data limitation. In this paper, we introduce Ferret-UI 2, a multimodal large languagearxiv.org 초록(Abstract)플랫폼의 다양성,.. 공감수 2 댓글수 2 2025. 1. 10.
  • Hallo2: Long-Duration and High-Resolution Audio-Driven Portrait Image Animation https://arxiv.org/abs/2410.07718 Hallo2: Long-Duration and High-Resolution Audio-Driven Portrait Image AnimationRecent advances in latent diffusion-based generative models for portrait image animation, such as Hallo, have achieved impressive results in short-duration video synthesis. In this paper, we present updates to Hallo, introducing several design enhancementsarxiv.org 초록최근에 Hallo와 같은 잠재 확.. 공감수 2 댓글수 3 2024. 11. 30.
  • MLE-bench: Evaluating Machine Learning Agents on Machine Learning Engineering https://arxiv.org/abs/2410.07095 MLE-bench: Evaluating Machine Learning Agents on Machine Learning EngineeringWe introduce MLE-bench, a benchmark for measuring how well AI agents perform at machine learning engineering. To this end, we curate 75 ML engineering-related competitions from Kaggle, creating a diverse set of challenging tasks that test real-world ML engarxiv.org 초록우리는 AI 에이전트가 기계 학습 엔.. 공감수 7 댓글수 5 2024. 11. 30.
  • GRIN: GRadient-INformed MoE https://arxiv.org/abs/2409.12136 GRIN: GRadient-INformed MoEMixture-of-Experts (MoE) models scale more effectively than dense models due to sparse computation through expert routing, selectively activating only a small subset of expert modules. However, sparse computation challenges traditional training practices,arxiv.org 초록Mixture-of-Experts (MoE) 모델은 전문가 라우팅을 통해 희소한 계산을 수행함으로써, 밀집 모델보다 더 효과적으.. 공감수 1 댓글수 2 2024. 11. 29.
  • Introducing SynthID Text (이전에 한 것) https://huggingface.co/blog/synthid-text Introducing SynthID TextIntroducing SynthID Text Do you find it difficult to tell if text was written by a human or generated by AI? Being able to identify AI-generated content is essential to promoting trust in information, and helping to address problems such as misattributionhuggingface.co 텍스트가 사람이 쓴 것인지 AI가 생성한 것인지 구분하기 어렵다고 느끼나요? AI 생성 콘텐츠를 식별하는 능력은 .. 공감수 2 댓글수 3 2024. 11. 29.
  • From questions to discoveries: NASA’s new Earth Copilot brings Microsoft AI capabilities to democratize access to complex data https://blogs.microsoft.com/blog/2024/11/14/from-questions-to-discoveries-nasas-new-earth-copilot-brings-microsoft-ai-capabilities-to-democratize-access-to-complex-data/ From questions to discoveries: NASA’s new Earth Copilot brings Microsoft AI capabilities to democratize access to complex dataEvery day, NASA’s satellites orbit Earth, capturing a wealth of information that helps us understand o.. 공감수 1 댓글수 1 2024. 11. 29.
  • TinyTroupe https://github.com/microsoft/TinyTroupe GitHub - microsoft/TinyTroupe: LLM-powered multiagent persona simulation for imagination enhancement and business insights.LLM-powered multiagent persona simulation for imagination enhancement and business insights. - microsoft/TinyTroupegithub.com 상상력 강화와 비즈니스 통찰력을 제공하는 LLM 기반 다중 에이전트 페르소나 시뮬레이션입니다. TinyTroupe 는 특정 성격, 관심사, 목표를 가진 사람들을 시뮬레이션 할 수 있는 실험적 Py.. 공감수 3 댓글수 2 2024. 11. 29.
  • Microsoft, 곧 Teams 회의에서 목소리 복제를 가능하게 할 예정 https://techcrunch.com/2024/11/19/soon-microsoft-will-let-teams-meeting-attendees-clone-their-voices/ Microsoft will soon let you clone your voice for Teams meetings | TechCrunchMicrosoft plans to let Teams users clone their voices so they can have their sound-alikes speak to others in meetings in different languages. At Microsofttechcrunch.com Kyle Wiggers 2024년 11월 19일 오전 5:30 (PST) Microsoft는.. 공감수 3 댓글수 2 2024. 11. 29.
  • Mistral has entered the chat 2024년 11월 18일 Mistral AI 팀AI는 인간과 기계 지능 간의 원활한 상호작용을 추구하는 것만큼이나 기계가 더 나은 추론을 하도록 만드는 첨단 과학의 탐구입니다. Mistral AI에서는 최첨단 모델을 개발하면서 여정을 시작했으며, 이제 AI 인터페이스에도 투자하기 시작했습니다. 그 첫 걸음으로 le Chat에 대대적인 업데이트를 진행했습니다.새로운 기능le Chat에 처음 오신 분들을 환영합니다! Le Chat [/lə ʃa/, 프랑스어로 '고양이']는 Mistral AI의 무료 생성형 AI 업무 도우미입니다. le Chat을 창의적인 파트너로 활용하거나, 특정 작업을 수행하도록 가르치거나, 코딩 코파일럿으로 사용할 수 있습니다. 즉, 다양한 일들을 처리하는 데 사용할 수 있습니다.오늘,.. 공감수 4 댓글수 4 2024. 11. 29.
  • Introducing FLUX.1 Tools https://blackforestlabs.ai/flux-1-tools/ Introducing FLUX.1 ToolsToday, we are excited to release FLUX.1 Tools, a suite of models designed to add control and steerability to our base text-to-image model FLUX.1, enabling the modification and re-creation of real and…blackforestlabs.ai  FLUX.1 도구 소개2024년 11월 21일—작성자: BlackForestLabs카테고리: 뉴스오늘 우리는 FLUX.1 도구를 발표하게 되어 매우 기쁩니다. FLUX.1 도구는 우리의 기본 텍스트-이미.. 공감수 1 댓글수 3 2024. 11. 29.
  • SAMURAI: Adapting Segment Anything Model for Zero-Shot Visual Tracking with Motion-Aware Memory https://arxiv.org/abs/2411.11922 SAMURAI: Adapting Segment Anything Model for Zero-Shot Visual Tracking with Motion-Aware MemoryThe Segment Anything Model 2 (SAM 2) has demonstrated strong performance in object segmentation tasks but faces challenges in visual object tracking, particularly when managing crowded scenes with fast-moving or self-occluding objects. Furthermore, the fixarxiv.org 초록Se.. 공감수 1 댓글수 2 2024. 11. 28.
  • Mapping Economic Well-being https://wandb.ai/johnowhitaker/dhs/reports/Mapping-Economic-Well-being--VmlldzoyMjY4NTI1 Mapping Economic Well-beingMini-project for the W&B MLOps course. Made by Jonathan Whitaker using Weights & Biaseswandb.ai "우주에서 어떤 지역의 부유함을 알 수 있을까요? 사실, 이것은 매우 중요한 질문입니다. 만약 '네'라는 답을 얻을 수 있다면, 비용이 많이 드는 국가 조사 없이도 한 나라 내에서 부의 분포를 파악할 수 있다는 것을 의미합니다. 이 프로젝트에서는 바로 이 작업을 시도해볼 것이며, 최근 연구를 재현하여 기존의 가구 조사와 원격 감지 .. 공감수 4 댓글수 3 2024. 11. 27.
  • A Deep Dive Into OpenCLIP from OpenAI https://wandb.ai/johnowhitaker/openclip-benchmarking/reports/A-Deep-Dive-Into-OpenCLIP-from-OpenAI--VmlldzoyOTIzNzIz A Deep Dive Into OpenCLIP from OpenAIThis article explores an open re-implementation of OpenAI's CLIP model for zero-shot classification and covers its additional uses and its potential issues.wandb.ai 2021년 초, OpenAI는 자연어 감독을 통해 시각적 개념을 효율적으로 학습하는 모델인 CLIP을 발표했습니다. CLIP은 매우 강력하며,.. 공감수 4 댓글수 4 2024. 11. 27.
  • Mid-U Guidance: Fast Classifier Guidance for Latent Diffusion Models https://wandb.ai/johnowhitaker/midu-guidance/reports/Mid-U-Guidance-Fast-Classifier-Guidance-for-Latent-Diffusion-Models--VmlldzozMjg0NzA1 Mid-U Guidance: Fast Classifier Guidance for Latent Diffusion ModelsIntroducing a new method for diffusion model guidance with various advantages over existing methods, demonstrated by adding aesthetic guidance to Stable Diffusion.wandb.ai SD2.1과 중간-사용자 가이드라인.. 공감수 0 댓글수 1 2024. 11. 27.
  • The Surprising Effectiveness of Test-Time Training for Abstract Reasoning https://arxiv.org/abs/2411.07279 The Surprising Effectiveness of Test-Time Training for Abstract ReasoningLanguage models have shown impressive performance on tasks within their training distribution, but often struggle with novel problems requiring complex reasoning. We investigate the effectiveness of test-time training (TTT) -- updating model parameters temarxiv.org graph TB subgraph "전통적인.. 공감수 1 댓글수 3 2024. 11. 27.
  • On the Importance of Noise Scheduling for Diffusion Models https://arxiv.org/abs/2301.10972 On the Importance of Noise Scheduling for Diffusion ModelsWe empirically study the effect of noise scheduling strategies for denoising diffusion generative models. There are three findings: (1) the noise scheduling is crucial for the performance, and the optimal one depends on the task (e.g., image sizes), (2) wharxiv.org 초록우리는 노이즈 스케줄링 전략이 소음 제거 확산 생성 모델에 미치는 영향.. 공감수 1 댓글수 2 2024. 11. 27.
  • agent.exe https://github.com/corbt/agent.exe GitHub - corbt/agent.exeContribute to corbt/agent.exe development by creating an account on GitHub.github.com 'agent.exe'는 Anthropic의 Claude 3.5 Sonnet 모델의 컴퓨터 제어 기능을 활용하여 사용자의 로컬 컴퓨터를 직접 제어할 수 있게 해주는 간단한 Electron 애플리케이션입니다.  주요 기능:Claude 3.5 Sonnet과의 통합: 이 애플리케이션은 Claude 3.5 Sonnet 모델을 통해 컴퓨터를 제어하며, 이를 위해 Anthropic API 키가 필요합니다.크로스 플랫폼 지원: 주로 macOS에서 테스트되었지만, .. 공감수 0 댓글수 0 2024. 11. 26.
  • nivida/llama-3.1-nemotron-70B https://build.nvidia.com/nvidia/llama-3_1-nemotron-70b-instruct llama-3.1-nemotron-70b-instruct model by nvidia | NVIDIA NIMLlama-3.1-Nemotron-70B-Instruct is a large language model customized by NVIDIA in order to improve the helpfulness of LLM generated responses.build.nvidia.com https://huggingface.co/nvidia/Llama-3.1-Nemotron-70B-Instruct nvidia/Llama-3.1-Nemotron-70B-Instruct · Hugging Face.. 공감수 0 댓글수 0 2024. 11. 26.
  • OpenAI Swarm https://github.com/openai/swarm GitHub - openai/swarm: Educational framework exploring ergonomic, lightweight multi-agent orchestration. Managed by OpenAI SolutEducational framework exploring ergonomic, lightweight multi-agent orchestration. Managed by OpenAI Solution team. - openai/swarmgithub.com https://www.magicaiprompts.com/blog/openai-swarm-ai-agent-collaboration OpenAI Swarm 이란? AI 에이전트 간.. 공감수 0 댓글수 0 2024. 11. 26.
  • Qwen2.5: A Party of Foundation Models! https://qwenlm.github.io/blog/qwen2.5/ Qwen2.5: A Party of Foundation Models!GITHUB HUGGING FACE MODELSCOPE DEMO DISCORD Introduction In the past three months since Qwen2’s release, numerous developers have built new models on the Qwen2 language models, providing us with valuable feedback. During this period, we have focused on cqwenlm.github.io 소개지난 3개월 동안 Qwen2가 출시된 이후, 수많은 개발자들이 Qwen2 언어 모델을 .. 공감수 4 댓글수 2 2024. 11. 26.
  • [모델까지만]Moshi: a speech-text foundation model for real-time dialogue https://github.com/kyutai-labs/moshi GitHub - kyutai-labs/moshiContribute to kyutai-labs/moshi development by creating an account on GitHub.github.com https://arxiv.org/abs/2410.00037 Moshi: a speech-text foundation model for real-time dialogueWe introduce Moshi, a speech-text foundation model and full-duplex spoken dialogue framework. Current systems for spoken dialogue rely on pipelines of ind.. 공감수 0 댓글수 1 2024. 11. 26.
    문의안내
    • 티스토리
    • 로그인
    • 고객센터

    티스토리는 카카오에서 사랑을 담아 만듭니다.

    © Kakao Corp.