컴퓨터 vs 책

(오늘의 짤방: 에? 하고 연기 발음해봤다가 웃음터져서 잠을 못잠 ㅠ via @wasab2_)
빅데이터/인공지능
Buffer Overflow in Mixture of Experts
자율 AI 에이전트, ChatGPT 다음의 메가트렌드?
MM-LLMs: 멀티모달 대규모 언어 모델의 최근 발전에 대한 연구 (Recent Advances in MultiModal Large Language Models)
Welcome to the Offensive ML Playbook
CREMA: Multimodal Compositional Video Reasoning via Efficient Modular Adaptation and Fusion
이제 역사학자도 AI 알아야 하는 시대 - AI로 고문서 인식 및 복원하는 기술 잇달아 등장
ALOHA 2: An Enhanced Low-Cost Hardware for Bimanual Teleoperation
사진작가, 미드저니 버전6에 왜 놀라는가
MetaVoice-1B - 1.2B 파라미터 Text-To-Speech 모델 (github.com/metavoiceio)
노르웨이 오슬로, 11만 명의 학생 및 교사를 위해 ChatGPT 도입 (digi.no)
The Whisper API now support word and segment level timestamps which makes editing audio / video much easier and granular 🎬
Exploring the potential of OpenAI Matryoshka 🪆 embeddings with Vespa
Gemini Advanced Prompt Enginnering Guide
Tag 309K Images/$ with Recognize Anything Model++ (RAM++) On Consumer GPUs
Multilingual E5 Text Embeddings: A Technical Report
구글 맵스, 로컬 가이드를 위한 새로운 생성 AI 기능 도입 (blog.google)
OpenAI, DALL-E 3에 새로운 워터마크 추가 (theverge.com)
EfficientViT-SAM: Accelerated Segment Anything Model Without Performance Loss
More Agents Is All You Need
Ollama에 OpenAI 형식의 호출 호환 기능 추가 (ollama.ai)
Large Language Model for Table Processing: A Survey
Multimodal RAG for processing videos using OpenAI GPT4V and LanceDB vectorstore
Models Leaderboard - Comparison and ranking the performance of AI LLM models across key metrics including quality, price, performance and speed (throughput & latency), context window & others.
cleanlab helps you clean data and labels by automatically detecting issues in a ML dataset.
RMBG v1.4 - 최첨단 배경 제거 모델 (huggingface.co)
“AI로 만든 합성 사진입니다” 메타, 자동 라벨링 툴 개발 중
‘부정확한 응답 걸러내고’… 생성형 AI 환각 현상 보완하는 접근법 3가지
‘모든 직원이 물고 뜯고 맛보게 하라’··· 삼사라의 생성형 AI 접근법
Useful training tip: when using Adam, don't double your learning rate when you double your batch size.(참고: How to Scale Hyperparameters as Batch Size Increases)
RAG Research Insights - Below is a collection of research papers highlighting key insights and the latest developments in RAG.
Training Data for the Price of a Sandwich - Common Crawl’s Impact on Generative AI
Grandmaster-Level Chess Without Search
MOMENT: A FAMILY OF OPEN TIME-SERIES FOUNDATION MODELS
EfficientViT-SAM: Accelerated Segment Anything Model Without Performance Loss
LoCo Benchmark - BM25 & Insights
The News About the News Business Is Getting Grimmer
StreamRAG: GPT-Powered Video Retrieval & Streaming 🚀
“한 달 만에 수십만 개 등록” 오픈AI GPT 스토어의 흥미로운 AI 툴
“조립형 플랫폼으로 생성형 AI 혜택 전사에 확대” 로이터 엔지니어링 총괄
Build with Gemini - Experience Google's largest and most capable AI model
DocsBot - OpenAI & other LLM API Pricing Calculator
GN⁺: 데이터 과학자를 위한 100개의 쿼리로 배우는 SQL (gvwilson.github.io)
Matryoshka Representation Learning (MRL) from the Ground Up
Guiding Instruction-based Image Editing via Multimodal Large Language Models by Apple
구글 클라우드, 기업용 AI 제품 구축 가이드 제시 "모델 이상이 필요하다"
네이버 "의료용 AI 서비스 만족도 늘어…정부, 기업·학계 연구 지원 꾸준해야" - "오픈AI 'GPT-4'·구글 '메드팜2' 가장 기능 좋아…뛰어넘는 기술 개발 해야"
What is 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗺𝗼𝗱𝗲𝗹 𝗖𝗼𝗺𝗽𝗿𝗲𝘀𝘀𝗶𝗼𝗻 and why you might need it?
⚔️ Vision Arena ⚔️ : Benchmarking VLMs in the Wild
MedSAM - This is the official repository for MedSAM: Segment Anything in Medical Images.
MedSAM-Lite 3D Slicer Plugin - This is the official repository for 3D Slicer Plugin for MedSAM: Segment Anything in Medical Images.
GN⁺: Ask HN: LLMs으로 무엇을 만들었나요? (news.ycombinator.com)
BGE-M3 - The Mother of all embedding models
MS, 원드라이브 코파일럿 5월 출시 "원드라이브 파일 쿼리 및 질문 가능"
“KTX” 말만 했는데 놀라운 일…시리 밀어낼 ‘찐 비서’ 정체
메타 "몇 달 내 페이스북 등 AI 생성 이미지에 꼬리표 붙일 것"
LLaVA-NeXT: Improved reasoning, OCR, and world knowledge
LLaVA NeXT: 제미나이 프로를 뛰어넘는 오픈소스 멀티모달 AI!
Exploring the GPT-4 with Vision API using Images and Videos
Adding new LLMs, text classification and code generation models to the Workers AI catalog
Background Removal - Background removal model developed by BRIA.AI, trained on a carefully selected dataset and is available as an open-source model for non-commercial use.
Zuckerberg’s Secret Weapon for AI Is Your Facebook Data
MLDR is a Multilingual Long-Document Retrieval dataset built on Wikipeida, Wudao and mC4, covering 13 typologically diverse languages.
BGE M3-Embedding: Multi-Lingual, Multi-Functionality, Multi-Granularity Text Embeddings Through Self-Knowledge Distillation
Web Voyager - WebVoyager by He, et. al., is a vision-enabled web-browsing agent capable of controlling the mouse and keyboard.
The Majority of AI Compute Spend is Not on Training but on Inference
Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs
네이버 검색 SRE의 시계열 데이터베이스 운영기 - VictoriaMetrics로 수천만 개의 시계열 데이터 다루기
TaskingAI - AI-Native 앱 개발을 위한 오픈소스 플랫폼 (github.com/TaskingAI)
mlx-llm - LLM applications running on Apple Silicon in real-time thanks to Apple MLX framework.
sqlcoder-7b-2 - A capable large language model for natural language to SQL generation.
Hugging Face, OpenAI의 커스텀 GPTs에 대항하는 오픈 소스 AI 어시스턴트 메이커 출시 (venturebeat.com)
NaturalSQL-7B - 자연어를 SQL로 변환하는 강력한 모델  (github.com/cfahlgren1)
저작권 방어 도구 ‘나이트쉐이드' 돌풍...5일 만에 25만건 다운로드
How to Build Your First Predictive Machine Learning Model
EfficientML.ai Lecture, Fall 2023, MIT 6.5940 by MIT HAN Lab
EASYTOOL: Enhancing LLM-based Agents with Concise Tool Instruction
RAG Local CLI Pack
DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
챗GPT가 스스로 운전한다?··· 자율주행과 LLM의 결합 현주소
아마존, AI 쇼핑 도우미 '루퍼스' 발표 "고객 경험을 의미 있게 개선"
구글, AI 기반 퍼징 프레임워크 ‘OSS-퍼즈’ 오픈소스로 개방
ERP에 AI 적용 위해 철저한 도입 계획 수립해야
Nomic Embed: Training a Reproducible Long Context Text Embedder
HuggingChat - Making the community's best AI chat models available to everyone.
Qwen1.5-72B-Chat - Qwen1.5-72B-Chat is the 72-billion parameter chat model of the Qwen series.
Introducing CodeAct, an agent {framework, instruction-tuning dataset, model}, employs executable Python code to unify the actions of LLM agents. 🧵
Is Self-Repair a Silver Bullet for Code Generation?
This Space combines OWLv2, the state-of-the-art zero-shot object detection model with SAM, the state-of-the-art mask generation model. SAM normally doesn't accept text input. Combining SAM with OWLv2 makes SAM text promptable. Try the example or input an image and comma separated candidate labels to segment.
Getting started with Llama - Quick setup and how-to guide by Meta
😇 New week, new embeddings for you! Announce 𝗷𝗶𝗻𝗮-𝗲𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝘀-𝘃𝟮-𝗯𝗮𝘀𝗲-𝗰𝗼𝗱𝗲, optimized for code search. 
Defog SQLCoder - Defog's SQLCoder is a family of state-of-the-art LLMs for converting natural language questions to SQL queries.
A Beginner’s Guide to Building Knowledge Graphs from Videos - Build a pipeline to analyze and store the data within videos.
Multimodal Ollama Cookbook
수학 유튜버 3blue1brown, 본인의 목소리를 학습한 AI를 이용해 시범적으로 한국어 더빙 제작 (youtube.com)
“책 구매··쇼핑 내역도 분석해 대출 신용평가하죠”
movieclip - This repository is an experiment in contrastive learning and how it can be applied to something I like a lot - movies.
Query Rewriting Cost Comparison
팀 쿡, 실적발표 컨퍼런스 콜에서 생성형 AI 계획 언급…WWDC서 공개할 듯
Image segmentation models separate areas corresponding to different areas of interest in an image. 
Specialized Language Models with Cheap Inference from Limited Domain Data
TravelPlanner: A Benchmark for Real-World Planning with Language Agents
RHO: Reducing Hallucination in Open-domain Dialogues with Knowledge Grounding
단 몇 분 만에 뚝딱… ‘나만의 AI 챗봇’ 만드는 법
Google Bard, 한국어 포함 40개 언어 지원 및 이미지 생성 기능 추가 (blog.google)
구글, 이미지AI ‘이매진2’ 바드 탑재…챗GPT 달리3와 달리 무료
Outliers have led me to 100s of business insights. But first I had to find them. In 6 minutes let me share 6 months of research into outliers.
Show GN: 보기 힘든 유튜브 영상과 논문을 요약해주는, Lilys AI (lilys.ai)
FOSDEM 2024- Tuana: Using Haystack to Build Custom Functionality for LLM Applications
Building a Hacker News Top Stories TL;DR
Punica: Serving multiple LoRA finetuned LLM as one
Create a Discord Chatbot Using LlamaIndex for Your Server
Paper - "Dissecting the Runtime Performance of the Training, Fine-tuning, and Inference of Large Language Models"
Responsible AI Pattern Catalogue
Setting Up Python for Machine Learning on Windows
Linear Model and Extensions
We've put up the largest collection of machine learning models in Core ML format, to help iOS, macOS, tvOS, and watchOS developers experiment with machine learning techniques.
MoE-LLaVA: Mixture of Experts for Large Vision-Language Models(주피터 노트북과 replicate로 시연)
How Fast Is MLX? A Comprehensive Benchmark on 8 Apple Silicon Chips and 4 CUDA GPUs
Everything You Always Wanted To Know About Mathematics* (*But didn’t even know to ask)
Research Papers in January 2024: Model Merging, Mixtures of Experts, and Towards Smaller LLMs
An introduction to graph theory(그림)
YOLO-World: Real-Time Open-Vocabulary Object Detector
RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval
Tutorial: Semantic Search, RAG and Index Vector Databases
Google just rolled out huge AI upgrades across Bard, Google Maps, and Imagen-2.
Sam Partee on Retrieval Augmented Generation (RAG)
A Survey on Hallucination in Large Vision-Language Models
finagg is a Python package that provides implementations of popular and free financial APIs, tools for aggregating historical data from those APIs into SQL databases, and tools for transforming aggregated data into features useful for analysis and AI/ML.
Large Language Models for Mathematical Reasoning: Progresses and Challenges
Health-LLM: Personalized Retrieval-Augmented Disease Prediction Model(논문)
A decoder-only foundation model for time-series forecasting
AI 이미지 생성 도구 강화에 나선 구글 바드
🗣️ Large Language Model Course
창작자와 사용자에게 플러그인보다 맞춤형 GPTs가 더 나은 이유 (moveit.substack.com)
This notebook demonstrates the use of the logprobs parameter in the Chat Completions API.
Large language models are having their Stable Diffusion moment
Today, Mark Zuckerberg shared the latest on our vision for building the most advanced AI products and services in our Q4 earnings call.
Comet is an MLOps Platform that is designed to help Data Scientists and Teams build better models faster!
Drawdata is a python library that allows you to draw a 2-D dataset of any shape in a Jupyter Notebook.
Beyond Naive RAG: Adding Agentic Layers
Enchanted - Ollama 모델들을 위한 iOS 앱 오픈소스 (github.com/AugustDev)
‘AI판사’가 재판하는 시대…“판결격차 감소” vs “피고인 특성반영 가능할까”
ChatGPT와 틴더를 이용해서 5,239명을 만난 개발자...
Adding Noise Increases Performance In RAG! 🤯
Retrieval Augmented Generation (RAG) with Llama Index and Open-Source Models - Learn how to effectively use proprietary data with your open-source LLMs in Python
LLaVA 1.6 from @imhaotian has been released with improved resolution support, visual reasoning, and OCR capabilities, all while maintaining minimalist design and data efficiency.
Faster Whisper transcription with CTranslate2
PrivateGPT와 간단한 사용 후기
GN⁺: Mistral CEO, GPT4 성능에 근접한 새로운 오픈 소스 AI 모델이 유출된 것을 시인 (venturebeat.com)
Boxplots are one of the most useful tools in my Data Science arsenal. In 6 minutes, I'll teach you 6 years of using box plots for EDA and problem-solving. Let's dive in.
MobileDiffusion: Rapid text-to-image generation on-device
An example of OpenAI's CLIP in MLX. The CLIP (contrastive language-image pre-training) model embeds images and text in the same space.
오픈소스로 완성하는 AI Full Stack
LG CNS "코드 생성형 AI에 최적화한 LLM 개발"
Phinetuning 2.0
mlx-moe - This repository contains scripts to create MoE models from pretrained llama/mistral models.
I spent the past few days testing rerankers and there are two that I'll use going forward. • Cohere • ColBERT
If you are using jupyter notebooks for Python and Data Science, try these 7 magic commands that will save you a ton of time:🧵
WhisperFusion - AI 챗봇과 짧은 대기시간으로 대화하기 (github.com/collabora)
12 RAG Pain Points and Proposed Solutions
Agentic RAG With LlamaIndex
PyTorch 2.2: FlashAttention-v2 integration, AOTInductor
Postgres.AI Bot. Towards LLM OS for Postgres
Deep Dive into AI with MLX-Pytorch
GN⁺: Meta AI, Code Llama 70B 모델 공개 (twitter.com/AIatMeta)
“나쁜 태도를 버리지 않는 AI” 앤트로픽 연구팀, 나쁜 AI 학습 및 복구 실험
한국어 학습 데이터
“컨텍스트 대화를 위한 런타임” 시멘틱 커널을 사용한 AI 에이전트 만들기
Complete ML Ops With Projects Series
This model was converted to MLX format from codellama/CodeLlama-7b-Python-hf.
GN⁺: Eagle 7B - Transformer를 뛰어넘은 모델 (blog.rwkv.com)
Cohere reranking is seriously good
Taipy - 데이터 & AI 알고리듬을 웹앱으로 만들어주는 오픈소스 파이썬 라이브러리  (github.com/Avaiga)
Introducing Qwen-VL
Infer-Retrieve-Rerank LlamaPack - This is our implementation of the paper "In-Context Learning for Extreme Multi-Label Classification by Oosterlinck et al.(파이썬 노트북)
Microsoft released their annual Future of Work report and this time around it’s not about remote work, it’s about AI.
Lumos - 로컬 LLM Co-Pilot 크롬 확장  (github.com/andrewnguonly)
Brave 브라우저용 AI 비서 Leo, 이제 Mixtral 8x7B LLM을 기본으로 사용 (brave.com)
ColBERT query-passage scoring interpretability
최신 AI 스택 : 엔터프라이즈 AI 아키텍처의 미래를 위한 설계 원칙 (menlovc.com)
Really cool - Multi-LoRA inference server that scales to 1000s of fine-tuned LLMs 🔥
Portable, non-invasive, mind-reading AI turns thoughts into text
PyTorch 101: Dataset Class clearly explained!
PowerInfer: Fast Large Language Model Serving with a Consumer-grade GPU 🔥
Llama in a container - This README provides guidance for setting up a Dockerized environment with CUDA to run various services, including llama-cpp-python, stable diffusion, mariadb, mongodb, redis, and grafana.
Welcome to RAGatouille - Easily use and train state of the art retrieval methods in any RAG pipeline. Designed for modularity and ease-of-use, backed by research.(ColBERT integration)
RAG vs Fine-tuning: Pipelines, Tradeoffs, and a Case Study on Agriculture
Mistral AI's Open-Source Mixtral 8x7B Outperforms GPT-3.5
Transcribe and summarize youtube video using mlx
Self-attention clearly explained:
Prompting Diverse Ideas: Increasing AI Idea Variance
Introducing the LlamaIndex retrieval-augmented generation command-line tool
하드웨어
ASML Shows Off $380 Million, 165-Ton Machine Behind AI Shift
"AI 지원 PC의 원년" 생성형 AI에 힘입은 2024년 PC 시장 확대 전망
프레임워크 랩톱 16 리뷰 | “가자, 모듈러의 세계로” DIY족을 위한 최고의 노트북
“2024년 AI PC와 생성형 AI 스마트폰 출하량, 총 2.95억 대 이를 것” 가트너 전망
“직접 사용해 보니…” 애플 비전 프로의 이모저모
"2023년 글로벌 반도체 매출, 8.8% 감소··· 인텔, 삼성 제치고 1위 차지"
삼성 갤럭시 S24 울트라 리뷰 | 유일한 단점은 아이폰보다도 비싼 가격
GN⁺: iFixIt의 Vision Pro 분해 분석 - 그 가짜 눈은 왜 이상해 보일까? (ifixit.com)
생성형 AI 열풍 속 양자 컴퓨팅 투자 ‘반토막’
애플 비전 프로 🍎 언박싱 & 첫 사용 후기
iFixit pulls the Vision Pro apart, exposing its connectors, screens, and fans
Most evaluation dev tools or frameworks atm are reactionary without system thinking.
험난한 과제에 직면한 애플 비전 프로
“단점 있지만, 인상적이고 재미있다” 애플 ‘비전 프로’ 언론 리뷰 모아보기
생성형 AI 도입 기업, IT 부서 넘어 현업 사용례 확대…워카토
PC·스마트폰·자동차 찍고 확장 또 확장··· 생성형 AI, 데이터센터를 벗어나다
와이파이 7, 2024년에 날개 달까?
비전 프로가 기대되는 단 하나의 이유, “웨어러블 맥”
Verge의 애플 비전 프로 리뷰: 마법 같은 경험, 그렇지 않을때 까지 (theverge.com)
XGBoost is now the go-to number 1 must-have algorithm in my data science toolkit.
[리뷰] Q60MAX – 현존 최고의 HHKB 배열 기계식 키보드
읽을거리
✨ 400만이 시청한 듄 세계관 해설의 결정판ㅣ듄 백과사전 몰아보기
네이버의 마이너스 성장이 시사하는 건
GN⁺: 넷플릭스: 불법 복제는 경쟁하기 어렵고 빠르게 성장하고 있습니다 (torrentfreak.com)
OECD, 회복 흐름 못 타는 한국 경제성장률 전망치 하향
쇼츠중독 시대, ‘문해력 과외’ 찾는 어른들…“회사 보고서 어려워요”
"등록금 인상? 학생 없어 꿈 못 꿔"…지방대 수입 2040년 '반토막'
“사채 고통 없애주려 만들었는데…프랑켄슈타인 됐다" 새마을금고법 '산파'의 후회
How social media algorithms 'flatten' our culture by making decisions for us
Mapped: The deadly geography of Mount Everest
일자리 찾아 세계로 - 해외취업 완전정복 - 호주
"이건 사야돼" 우르르…한국인, 돈 가장 많이 쓴 곳 봤더니 [오정민의 유통한입]
AB자산운용 "글로벌 우량 헬스케어에 집중 투자하라"
일본인 어머니 임종 못한 채 새 종자 개발 외길 - 부산 동래의 자유천(慈乳泉)과 우장춘 박사
(보너스: I just love the idea that after the apocalypse someone's job will be to be a 0.5 FLOPS vector database. @srush_nlp)
EOB

[B급 프로그래머] 2월 2주 소식(빅데이터/인공지능, 하드웨어, 읽을거리 부문)

댓글