(오늘의 짤방: via @Whotfismick)
- 빅데이터/인공지능
- LitServe! 🚀 - The next-gen serving engine for AI models—lightning fast ⚡, scalable.
- Researchers upend AI status quo by eliminating matrix multiplication in LLMs
- Toys “R” Us riles critics with “first-ever” AI-generated commercial using Sora
- Meet Figma AI: Empowering designers with intelligent tools
- Is ChatGPT the key to stopping deepfakes? Study asks LLMs to spot AI-generated images
- Belgian startup to build LLM that detects hate speech in all EU languages
- Let's reproduce GPT-2 (1.6B): one 8XH100 node, 24 hours, $672, in llm.c #677
- A new neural network makes decisions like a human would
- Mistral has released Codestral Mamba, a new model for code generation that uses the new Mamba architecture rather than Transformers.
- Mistral Large 2 has a 128k context window and supports dozens of languages including French, German, Spanish, Italian, Portuguese, Arabic, Hindi, Russian, Chinese, Japanese, and Korean, along with 80+ coding languages including Python, Java, C, C++, JavaScript, and Bash.
- Breaking Instruction Hierarchy in OpenAI's gpt-4o-mini
- GPT-4o mini: advancing cost-efficient intelligence
- Mistral NeMo: our new best small model. A state-of-the-art 12B model with 128k context length, built in collaboration with NVIDIA, and released under the Apache 2.0 license.
- Google’s new weather prediction system combines AI with traditional physics
- Google claims math breakthrough with proof-solving AI models
- "기대치 정점에서 내려오는 생성형 AI…떠오르는 자율 에이전트" 가트너
- "AI가 코딩하는 세상" 개발자와 기업이 직면한 현실
- “이이제이” AI 부정행위 막는 AI 탐지기의 한계와 고전적인 방안
- 한 시대의 은퇴: 이세돌의 은퇴를 기념하며
- KCA(한국방송통신전파진흥원) ChatGPT 업무활용 가이드북
- 생성형 AI로 이익 얻으려면, 먼저 ‘숙제’를 끝내야 한다
- AI 시장 2등 자리 노리는 앤트로픽··· 모바일 매출 100만 달러 돌파
- 가트너, 2024 신기술 하이프 사이클 발표··· '25개 혁신 기술, 4개의 동향으로 분류'
- “멀티모달 AI 역량 강화”··· 네이버, 하이퍼클로바X 기반 이미지·음성 처리 기술 공개
- 마이크로소프트, 신규 SLM 파이-3.5 출시 "메타·구글 모델보다 뛰어나"
- old_photo_restoration - upload old photo, ddcolor image colorization
- Phi 3.5 by @MicrosoftAI in VSCode
- GN⁺: Midjourney 웹 버전, 이제 모두가 사용 가능 (midjourney.com)
- 생성형 AI의 또다른 진화, 온디바이스 AI
- Speculative RAG: Enhancing retrieval augmented generation through drafting
- OpenAI, GPT-4o 파인튜닝 기능 출시 (openai.com)
- Code improves non-code tasks. 🔥 Code in pre-training data for LLMs significantly boosts LLM performance across diverse tasks, not just coding.
- Optimizing Tool Retrieval in RAG Systems: A Balanced Approach
- Ollama, Tool 호출 지원 (ollama.com)
- GN⁺: Zed AI 공개 (with Anthropic) (zed.dev)
- 오픈챗 해시태그 예측을 위한 다중 레이블 분류 모델 개발하기
- GN⁺: AI가 Hype를 잃어가고 있다 (economist.com)
- 12 must-know Data Cleaning techniques:
- Phi-3.5-vision is a lightweight, state-of-the-art open multimodal model built upon datasets which include - synthetic data and filtered publicly available websites - with a focus on very high-quality, reasoning dense data both on text and vision.
- Phi-3.5-MoE is a lightweight, state-of-the-art open model built upon datasets used for Phi-3 - synthetic data and filtered publicly available documents - with a focus on very high-quality, reasoning dense data.
- AI가 해시태그·SNS 분석해 사고팔았다…올 수익률 109%
- AI’s $200B Question - GPU capacity is getting overbuilt. Long-term, this is good. Short-term, things could get messy.
- IBM 및 대학 연구진, LLM 거짓 응답 줄이는 새 훈련 기법 제안
- "에이전틱 RAG, 기술의 발전이 아니라 마케팅 광고"
- “부조리한” 생성형 AI 규제를 최대한 활용하는 법
- AI 데이터 분석가 ‘물어보새’ 등장 – 2부. Data Discovery
- AI를 활용한 주문서 생성 자동화: 카카오톡 주문을 키친보드 주문으로
- GN⁺: 당신은 언어 모델보다 다음 단어 예측 능력이 뛰어난가요? (joel.tools)
- PyTorch는 죽었다. JAX 만세 (neel04.github.io)
- vLLM: Serve LLMs at scale
- Adapt or Become Obsolete: AI’s Unstoppable Business Revolution
- RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented Generation
- Amazon’s RAGChecker could change AI as we know it—but you can’t use it yet
- “API 비용 최대 90% 절감”··· 앤트로픽, 클로드 모델에 ‘프롬프트 캐싱’ 기능 추가
- 기업 내 RAG 도입의 실패 원인과 해결 방안 "검색 모델 최적화와 데이터 품질"
- GN⁺: Launch HN: Trellis – 비정형 데이터용 AI 기반 워크플로우 (news.ycombinator.com)
- Least Privilege Is a Pain that LLMs Can Fix
- Advanced RAG Techniques: Elevating Your Retrieval-Augmented Generation Systems 🚀
- TrustGraph is a true end-to-end (e2e) knowledge pipeline that performs a Naive Extraction on a text corpus to build a RDF style knowledge graph coupled with a RAG service compatible with cloud LLMs and open-source SLMs (Small Language Models).
- Grok-2 베타 릴리즈 (x.ai)
- RM의 사망과 AI에 거는 기대
- 오픈소스 AI 업계 판도 바꿀까?··· 애널리스트가 본 리눅스재단 AI 모델의 가능성
- Integrating Microsoft GraphRAG into Neo4j
- AudioSample is an optimized numpy-like audio manipulation library, created for researchers, used by developers.
- Unstract - No-code LLM Platform to launch APIs and ETL Pipelines to structure unstructured documents
- nano-llama31 - This repo is to Llama 3.1 what nanoGPT is to GPT-2. i.e. it is a minimal, dependency-free implementation of the Llama 3.1 architecture, and it can train, finetune, and inference it very simply.
- DreamWorks Animation founder: “I think of AI as a creative tool”
- 대부분의 AI 소프트웨어, 오픈소스화 가능성은 낮다
- 매출 기여하는 생성형 AI 프로젝트는 거의 없다?··· 불편한 진실 6가지
- We introduce Imagen 3, a latent diffusion model that generates high quality images from text prompts.
- LY의 QA 엔지니어가 생성형 AI를 이용해 품질 활동의 생산성을 높이는 방법
- 구글, 생성형 AI 솔루션 설계에서 얻은 3가지 교훈 제시
- 소비자는 "AI"라벨이 붙은 제품을 꺼려한다 (edition.cnn.com)
- ✨ MindSearch: Mimicking Human Minds Elicits Deep AI Searcher
- AI Toolkit by Ostris
- 🚀 The Open Ko-LLM Leaderboard2 🇰🇷 objectively evaluates the performance of Korean Large Language Model (LLM).
- '합스부르크 AI' 등장..."합성 데이터 사용은 근친혼에 해당, 붕괴할 것"
- 메타, 유니버설뮤직과 음악 라이선스 계약 확장… 인공지능 음악 생성 콘텐츠 문제 해결 '맞손'
- EXAONE 3.0 : 글로벌 Top 수준의 성능 갖춘 오픈소스 모델 공개
- Enhancing Relevance of Embedding-based Retrieval at Walmart
- Medical Graph RAG: Towards Safe Medical Large Language Model via Graph Retrieval-Augmented Generation
- 100 AI Tools to replace your tedious work:
- Flood of 'junk': How AI is changing scientific publishing
- RecLoRA tackles personalization in LLMs for recommendation systems.
- Mistral, 더 간단한 모델 커스터마이징 기능과 Agents 알파 릴리스 (mistral.ai)
- The LLM-Aided OCR Project is an advanced system designed to significantly enhance the quality of Optical Character Recognition (OCR) output.
- GN⁺: Show HN: LLM-Aided OCR – LLM으로 Tesseract OCR 오류 수정 (github.com/Dicklesworthstone)
- Alibaba, Qwen2-Math 모델 공개 (qwenlm.github.io)
- Mere Imitation - Generative AI has lately set off public euphoria: the machines have learned to think! But just how intelligent is AI?
- LUNAR: Unsupervised LLM-based Log Parsing
- Multimodal RAG over Complex Legal ⚖️, Insurance 🪪, and Product Documents 🎁
- 하드웨어
- 읽을거리
EOB
댓글 없음:
댓글 쓰기