[B급 프로그래머] 7월 4주 소식(빅데이터/인공지능, 하드웨어, 읽을거리 부문)

(오늘의 짤방: Twitter logo evolution 2006–2023 via @WebDesignMuseum)

1. 빅데이터/인공지능
   • Optimizing LLM latency - An exploration of inference tools for open source LLMs focused on latency.
   • '1세대 챗봇' 심심이, 하버드대학교에 4억명 대화 데이터 제공
   • Awesome JSON Datasets - A curated list of awesome JSON datasets that don't require authentication.
   • Data Visualization Interfaces in Python With Dash
   • Sample Chat App with AOAI - This repo contains sample code for a simple chat webapp that integrates with Azure OpenAI. Note: some portions of the app use preview APIs.
   • Raising the Cost of Malicious AI-Powered Image Editing
   • 포티투닷 AI챗봇, 현대차에 심는다
   • Many people struggle to understand #Bayesian updating! So I made this super cool #Python @matplotlib #educational, interactive dashboard.
   • How many GPUs to run full parameter fine tuning on llama-2 70B? Looks like it will be roughly 32 GPUs using deepspeed (A10 GPU 24GB, comparable to a 3090) or 10-12~ A100 80GB. H/T @CyrusHakha for sharing numbers
   • GPT-4 + Streaming Data = Real-Time Generative AI
   • Build a real time feature pipeline in python step by step
   • Are the scaling Laws for LLMs shifting🤔 Earlier this year, we started to see training smaller LLMs for longer, but why? 💭
   • Rebel + LlamaIndex Knowledge Graph Query Engine
   • Exploring the Most Popular Machine Learning and Deep Learning GitHub Repositories
   • “패브렉스 기술로 32개 GPU를 한 서버에 결합”··· 기가아이오, 싱글노드 AI 슈퍼컴퓨터 출시
   • AI 포트폴리오 강화하는 AWS··· 아마존 베드록 업데이트 및 헬스케어용 AI 서비스 출시
   • AWS, 생성형 AI 기반의 임상 문서 자동 생성 서비스 ‘AWS 헬스 스크라이브’ 출시
   • 3 ingredients - to solve a business problem using LLMs
   • "냉각, 전력까지 총체적 부족" 생성형 AI 대비 시급한 데이터센터
   • Datadog Creates Scalable Data Ingestion Architecture
   • Tales of Kafka at Cloudflare: Lessons Learnt on the Way to 1 Trillion Messages
   • Programmatic Custom Model Creation - Introducing programmatic custom model creation with our Python SDK. (by Cohere)
   • LLM-grounded Diffusion: Enhancing Prompt Understanding of Text-to-Image Diffusion Models with Large Language Models
   • ChatDOC - Chat with documents. Get instant answers with cited sources.
   • Made With ML - Join 40K+ developers in learning how to responsibly deliver value with ML!
   • VectorAdmin: The universal interface for vector databases.. A production-ready vector database management system that doesn't hurt to use.
   • Announcing OverflowAI - Let’s highlight the new features and products we announced today from the stage of WeAreDevelopers.
   • 구글·오픈AI·마이크로소프트·앤트로픽, ‘프론티어 모델 포럼’ 창립
   • 엔비디아, 생성AI 모델 훈련용 DGX클라우드 서비스 시작
   • 클라우드만 선방한 삼성SDS…"연내 생성형 AI 출시"
   • AWS·메타·MS 참여한 오버추어맵스재단, 오픈소스 지리 데이터 무료 공개
   • 메타, 클라우드 사업자에 '라마2' 이용료 받는다
   • 네이버, AI 지향점은 'B2B'…쏘카는 시작일 뿐
   • 의료산업도 AI 전쟁터... MS·구글 이어 아마존도 참전
   • Machine Learning & Deep Learning Tutorials
   • LMs can serve as powerful engines to extract insights from your data.
   • Everything you need to know to create a ChatGPT plugin
   • Exciting news for the StarCoder cosmos!✨ The team just released new smaller StarCoder models for edge and on-device code generation - introducing the 1B, 3B and 7B models. 📱
   • Creating an automated meeting minutes generator with Whisper and GPT-4
   • A Context-based Chatbot Surpasses Trained Radiologists and Generic ChatGPT in Following the ACR Appropriateness Guidelines
   • 생성형 AI, 더 새롭고 골치 아픈 섀도우 IT 시대를 열다
   • Fine-Tune Your Own Llama 2 Model in a Colab Notebook(노트북
   • **Instruction-Tuned Llama 2: Comprehensive Guide & Code** 🚀
   • What We Know About LLMs (Primer)
   • The economic potential of generative AI: The next productivity frontier
   • “AI가 작성한 글 구별하기 아직 어려워”··· 오픈AI, AI 콘텐츠 탐지 도구 개발 중단
   • 가능성과 호들갑 사이··· 챗GPT·LLM의 실제 비즈니스 가치는? 활용처는?
   • 챗GPT 넘어라··· 다양한 LLM 활용이 필요한 때
   • ‘선수 실시간 데이터 활용’… AI로 팬 경험 혁신한 스페인 축구 리그
   • Stable Diffusion XL v1.0 🦒 Colab 🥳 Thanks to @StabilityAI ❤
   • What is 𝗺𝗼𝗱𝗲𝗹 𝗿𝗲-𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 and how to implement it?
   • Amazon OpenSearch Serverless용 벡터 엔진 미리 보기 출시
   • 🌋 LLaVA: Large Language and Vision Assistant
   • "거짓말도 사람만큼 잘한다" 생성형 AI, 소셜 미디어 흉내에 효과적
   • Elastic vector dense field type and index vector
   • pgvector - Open-source vector similarity search for Postgres
   • JetBrains IDE의 AI Assistant
   • QLoRA 기법으로 Falcon-40B 및 기타 대규모 모델(LLM)을 Amazon SageMaker Studio 노트북의 대화형 환경에서 파인튜닝하기
   • 태빌리티AI, 오픈소스 LLM ‘프리윌리’ 출시
   • Automating Web Research
   • SparseEmbed: Learning Sparse Lexical Representations with Contextual Embeddings for Retrieval
   • Combining DuckDB and PyArrow allows you to efficiently process datasets larger than memory on a single machine.
   • ChatGPT has Never Seen a SINGLE Word (Despite Reading Most of The Internet). Meet LLM Tokenizers.
   • TLDR: We successfully trained a Flamingo-Llama2-Chat7B on CC3M in less than 5 hours using just 4 A100s.
   • Extended Guide: Instruction-tune Llama 2
   • Announcing Jupyter Notebook 7
   • New – Amazon EC2 P5 Instances Powered by NVIDIA H100 Tensor Core GPUs for Accelerating Generative AI and HPC Applications
   • 1️⃣ AI for beginners - A 12-week, 24-lesson curriculum all about Artificial Intelligence. (by Microsoft)
   • Llama 2 - Responsible Use Guide (by Meta)
   • LLM-Reading-List - Just helping myself keep track of LLM papers that I‘m reading, with an emphasis on inference and model compression.
   • How to Leverage Artificial Intelligence to Grow Your Business
   • AI 윤리 모범답안 ‘데이터셋’ 공개한 네이버 이화란 팀장 “치우친 의견, 비윤리적 답은 위험 초래해”
   • 데이터가 곧 가치인 시대, 성패는 데이터 품질에 달려있다
   • Predicting Code Coverage without Execution
   • vectordb - A Python vector database you just need - no more, no less.
   • Do Efficient Training Algorithms / Optimizers really save us compute when training Transformer LMs? 🧐
   • Complete Numpy Library
   • ARB: Advanced Reasoning Benchmark for Large Language Models
   • llama2-webui - Running Llama 2 with gradio web UI on GPU or CPU from anywhere (Linux/Windows/Mac).
   • RLHF gets far more powerful as models get bigger
   • GPT-4 System Card by OpenAI
   • ARB: Advanced Reasoning Benchmark for Large Language Models
   • Azure OpenAI Landing Zone reference architecture
   • 💬🚀 LLM as a Chatbot Service
   • Unraveling GPU inference costs for fine tuned open source models vs closed platforms
   • ChatGPT broke the Turing test — the race is on for new ways to assess AI
   • 나만의 GPT 만들기, Embedding
   • LLM Censorship: A Machine Learning Challenge or a Computer Security Problem?
   • tinyvector - the tiny, least-dumb, speedy vector embedding database.
   • JAX Implementation of Llama 2
   • CUDA Version Management
   • 🌋 LLaVA: Large Language and Vision Assistant
   • What We Know About LLMs (Primer)
   • Parameters and Hyperparameters 101.
   • llama2.c - With the code in this repo you can train the Llama 2 LLM architecture from scratch in PyTorch, then export the weights to a binary file, and load that into one ~simple 500-line C file (run.c)
   • The case for 4-bit precision: k-bit Inference Scaling Laws
   • Simplify Search with Multilingual Embedding Models
   • What is LlamaIndex.TS?
   • “AI 출시 전에 외부 보안 테스트” 구글ㆍMSㆍ오픈AI, 美 정부와 합의
   • scale - Custom Open-Source LLMs. Fine-tune open-source large language models for improved performance on your most important use cases.
   • ⚡ LLM Engine ⚡ - The open source engine for fine-tuning and serving large language models.
   • How to Fine-Tune Llama 2 With LLM Engine
   • Attention Is Off By One
   • 글로벌 칼럼 | 마이크로소프트 365 코파일럿, 새로운 생산성 혁명의 시작
   • 하이퍼클로바X·믿음… ICT업계, 하반기 빅테크에 반격 채비
   • 네이버 뉴로클라우드 출격… 하반기 B2B AI대전 불붙는다
   • 이제야 ‘생성형 AI’ 전쟁 참전한 애플, 늦었지만 늦지 않았다
   • ‘그림자 아닌 빛으로’··· 시민 개발자·데이터 과학자 성공 사례 3가지
   • 마이크로소프트, 자연어 인터페이스 구축을 위한 오픈소스 라이브러리 ‘타입챗’ 공개
   • What is LangSmith and why should I care as a developer?
   • What is Similarity Between Sentences?
   • BTLM-3B-8K: 7B Performance in a 3 Billion Parameter Model
   • Given the demand, here's the thread on deploying LLaMA 2 (or any open-source LLM of your choice!) using Hugging Face 🦙🧶
   • Amazon-M2: A Multilingual Multi-locale Shopping Session Dataset for Recommendation and Text Generation
   • Transformers: the Google scientists who pioneered an AI revolution
   • Ollama - Run, create, and share large language models (LLMs).
   • LLaMA2 isn't "Open Source" - and why it doesn't matter
   • Codegen2.5 is only 7B parameters and matches starcoder on benchmarks which is double the size 15B.
   • BoxDiff: Text-to-Image Synthesis with Training-Free Box-Constrained Diffusion
   • ReLoRa: Pre-train a Large Language Model on Your GPU
   • How to ensure 𝗗𝗮𝘁𝗮 𝗤𝘂𝗮𝗹𝗶𝘁𝘆 𝗶𝗻 𝗠𝗮𝗰𝗵𝗶𝗻𝗲 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗦𝘆𝘀𝘁𝗲𝗺𝘀?
   • ML project tip 🧠
   • 🧠이해하기 쉽게 마이크로소프트가 설명해 주는 '코파일럿용 시맨틱 인덱스(Semantic Index for Copilot)🧠
   • Qualcomm Works with Meta to Enable On-device AI Applications Using Llama 2
   • The latest release of bitsandbytes has an improved CUDA setup and A100 4-bit inference.
   • The Open Handbook of Linguistic Data Management
   • 안드레 카파시의 도전: AGI
   • Decoding the ACL Paper: Gzip and KNN Rival BERT in Text Classification
   • Llama 2 x Function Calling API
   • llama2.c compiled with Emscripten to run in a web page
   • Hands-on MLOps - Train and Deploy a Real-Time ML model to predict crypto prices
   • Lance - Modern columnar data format for ML. Convert from Parquet in 2-lines of code for 100x faster random access, a vector index, data versioning, and more.
   • AI 코딩 동반자 Amazon CodeWhisperer 활용하는 4가지 방법
   • There's a long-tail of advanced query techniques in RAG beyond top-k retrieval. 🔭
   • How to Train Generative AI Using Your Company’s Data
   • This repo contains GGML format model files for Meta's Llama 2 70B Chat.
   • Faster PyTorch Training by Reducing Peak Memory (combining backward pass + optimizer step)
   • 생성AI 잠재력을 더욱 키워줄 4가지 키워드
   • Meta’s LLaMa 2 license is not Open Source
   • 🧺 RAGstack - Deploy a private ChatGPT alternative hosted within your VPC.
   • The Neurokernel Project aims to build an open software platform for the emulation of the entire brain of the fruit fly Drosophila melanogaster on multiple Graphics Processing Units (GPUs).
   • The Curse of Recursion: Training on Generated Data Makes Models Forget
   • The AI feedback loop: Researchers warn of ‘model collapse’ as AI trains on AI-generated content
   • Mechanical Turk workers are using AI to automate being human
   • OpenAI의 Function calling
   • Meta's AI MusicGen makes music from text prompts
   • I-JEPA: The first AI model based on Yann LeCun’s vision for more human-like AI
   • About Infinigen - Infinigen is a procedural generator of 3D scenes, developed by Princeton Vision & Learning Lab.
   • 🦾 OpenLLM - n open platform for operating large language models (LLMs) in production. Fine-tune, serve, deploy, and monitor any LLMs with ease.
   • Voicebox: Text-Guided Multilingual Universal Speech Generation at Scale
   • Introducing Voicebox: The first generative AI model for speech to generalize across tasks with state-of-the-art performance
   • Official Repo of Tree of Thoughts (ToT)
   • Tree of Thoughts: Deliberate Problem Solving with Large Language Models
   • This startup gives your speech a new ‘human-realistic’ AI voice — for free
   • LLaMA 2 - Every Resource you need
   • Introducing new GitHub Copilot features in VS Code!
   • Discover LlamaIndex: Key Components to build QA Systems
   • RoboCat: A self-improving robotic agent
   • “낙서를 멋진 그림으로” 스테이블 AI, 스테이블 두들 출시
   • 마이크로소프트, ‘인스파이어 2023’ 개최… “AI 기반 기술 및 정책 대거 업데이트”
   • 코파일럿 전략 가속화··· 마이크로소프트, 365 코파일럿 가격 공개 및 AI 영업 도구 추가
   • AI 제조업 시대는 분명히 오겠지만.... '관건은 시기'
   • "생성형 AI, 기업 IT 지출 영향 적지만 대비는 필요" 가트너 IT 지출 전망
   • 애플도 부산하다…"내부에서 애플 GPT 가동 중"
   • ML커먼스, 의료 AI 모델 정확도 평가 도구 ‘메디퍼프’ 공개
   • MS 365 코파일럿 의 신규 가격 정책이 가져온 불길한 징조
   • Why GPT-3.5 is (mostly) cheaper than Llama 2
   • 'AI의 심장' 머신러닝 알고리즘의 이해
   • AI 생산성의 최대 장벽은 사람이다
   • 이제야 ‘생성형 AI’ 전쟁 참전한 애플, 늦었지만 늦지 않았다
   • “AI 관련 공고 450% 급증” 채용시장 트렌드가 변한다
   • 라즈베리 파이에서 Stable Diffusion 실행 (github.com/vitoplantamura)
   • 🤗 Open LLM-Perf Leaderboard 🏋️
   • Deploy large language models responsibly with Azure AI
   • 네이버, AI 생태계 확장한다…일반·비즈니스 서비스 줄줄이 예고
   • Meta and Qualcomm team up to run big A.I. models on phones
   • If you’re building LLM apps over your data systems, there’s more uses to vector databases than you might think! 💡
   • We are excited to announce the first major release of the Chatbot Arena conversation dataset!
   • Boosted Llama V2 inference speed by 1.8x using @OpenAI's Triton (one key tech behind GPT-4) at batch=1 FP16 on a 3090 GPU, no quality compromises (-> no quant, etc.)
   • No model has ever beaten 91% accuracy on ImageNet! Because this dataset has over 100,000 mislabeled images.
   • Examples of using peft with trl to finetune 8-bit models with Low Rank Adaption (LoRA)
   • Challenges and Applications of Large Language Models
   • There is no one “performance” metric when building LLM apps for prod
   • AlpacaEval : An Automatic Evaluator for Instruction-following Language Models
   • Fast inference with vLLM (Llama 2 13B)
   • Python library that adds Generative AI capabilities to Pandas🔥
   • PandasAI is a Python library that adds Generative AI capabilities to pandas, the popular data analysis and manipulation tool.
   • Android in the Wild (AitW) - Android in the Wild (AitW) is a large-scale dataset for mobile device control that contains human-collected demonstrations of natural language instructions, user interface (UI) screens, and actions for a variety of human tasks.
   • MAPIE - Model Agnostic Prediction Interval Estimator
   • Run Llama 2 Web UI on Colab or LOCALLY!
   • text-generation-webui-colab
   • imple prompting tricks help to improve Llama 2 (chat) performance massively ⚠️
   • [AI 툴 추천] 효율적인 웹리서치를 위한 AI - Betterresearch.com
   • Rerank can speed up an LLM query without sacrificing accuracy (and in fact, probably improving it). It does so by pruning away irrelevant nodes from the context.
   • Introducing Semantic Kernel for Java
   • Google Cloud expands availability of enterprise-ready generative AI
   • Improving search experiences with Enterprise Search on Gen App Builder (by Google)
   • Google Cloud Enterprise Search (by langchain)
   • Many of us practitioners have felt that GPT-4 degrades over time. It's now corroborated by a recent study. But why does GPT-4 degrade, and what can we learn from it?
   • brilliant - The best way to learn math and computer science
   • LLM경쟁 점입가경...K-테크들도 정조준
   • 통신3사 "AI투자 급한데 카르텔 논란 억울"
   • 데이터 분석 예제 데이터 세트로 iris꽃을 대체하는 3종의 팔머 펭귄 데이터 다운로드가 백만을 넘어섰다고. :)
   • How good is Llama 2 Chat? Key insights from our eval:
   • MS, 파트너 컨퍼런스 ‘인스파이어 2023’ 개최… “AI 기술 및 정책 대거 업데이트”
   • 메타·MS·퀄컴 '삼각동맹'… 오픈소스 전략으로 AI 판 흔든다
   • Get started by selecting a learning path (by Google)
   • LLAMA-v2 training successfully on Google Colab's free version! "pip install autotrain-advanced" 💥 Yes, you can also use your local machine!
   • LLM을 활요한 연구분야 주제분석 실습 - ChatGPT API, Embeddings API, Pubmed API
   • Fine-tune LLaMA 2 (7-70B) on Amazon SageMaker
   • Apple Tests ‘Apple GPT,’ Develops Generative AI Tools to Catch OpenAI
   • Train LLMs in just 50 lines of code!
   • Meta LLM 패권 전쟁 격화시키다 : LlaMA2 상업적 이용가능 오픈소스로 공개
   • uptrain - An open-source framework to evaluate and monitor LLM applications
   • llama-2-13b-chat playground - public A 13 billion parameter language model from Meta, fine tuned for chat completions
   • [챗GPT 러닝데이 | 한국어 LLM 민주화의 시작 KoAlpaca!
   • Llama 2 foundation models from Meta are now available in Amazon SageMaker JumpStart
   • 사실상 마지막 기회 맞은 카카오엔터프라이즈
   • AI 속도조절 네이버 '큐' 사인 늦췄다
   • AI 확산 속 신생 클라우드들이 급성장하는 비밀
   • Text generation web UI - A gradio web UI for running Large Language Models like LLaMA, llama.cpp, GPT-J, OPT, and GALACTICA.
   • Here's a short TL;DR on what Meta did to improve the state of the art 🧵
   • How is ChatGPT's behavior changing over time?
   • Llama 2: an incredible open LLM
   • Diffusion Models Beat GANs on Image Classification
   • The Practical Guides for Large Language Models
   • You'll soon see lots of "Llama just dethroned ChatGPT" or "OpenAI is so done" posts on Twitter. Before your timeline gets flooded, I'll share my notes:
   • The new Llama 2 model has an additional licensing requirement which prevents use if you have >700M MAUs, which is an oddly high number. OpenAI's ChatGPT hit 100 million MAUs in January, and has gone down since. It's essentially a "Amazon and Google don't use this k thx."
   • Fine tune Llama v2 models on Guanaco Dataset
   • M2 맥북의 GPU를 활용한 Jupyter Notebook 띄우기
   • Llama 2: open source, free for research and commercial use
   • Meet PolyLM (Polyglot Large Language Model): An Open Source Multilingual LLM trained on 640B Tokens, Available In Two Model Sizes 1.7B and 13B
   • 허깅페이스와 LoRA를 사용하여 단일 Amazon SageMaker GPU에서 대규모 언어 모델(LLM) 훈련하기
   • 대규모 언어 모델이 디지털 공공재에 위협이 될까요? (arxiv.org)
   • AI Fundamentals: Datasets 101
   • LMQL is a programming language for language model interaction.
   • Building StarCoder, an Open Source LLM Alternative
   • simpleaichat is a Python package for easily interfacing with chat apps like ChatGPT and GPT-4 with robust features and minimal code complexity.
   • AI, ML, Data Engineering News Round Up: Vertex, AlphaDev, Function Calling, Gorilla, and Falcon
   • 엔비디아 H100 GPU, AI 훈련 속도 가장 빠르다…MLPerf 벤치마크서 10분대 기록
   • 미 FTC, 챗GPT의 유해성 조사…가짜 정보 유포 우려 제기
   • The @FTC is investigating @OpenAI and the document outlining their questions is fascinating.
   • IMF 출신 이창용 총재도 놀란 시각화 플랫폼···한은 ‘스냅샷’ 공개
   • TOAST: Top-Down Attention Steering
   • Remove Watermarks from Your Images for FREE
   • 카카오브레인, 이미지 생성모델 Karlo 2.0 공개 (developers.kakao.com)
   • Open-Source Text Generation & LLM Ecosystem at Hugging Face
   • '챗GPT' 업무에 적용한 교보생명..."디지털 전환 역량 강화"
   • Learning to Retrieve In-Context Examples for Large Language Models
   • Every major AI nails the Voight-Kampff test.
   • Learning an embedding with the kernel trick using GZIP
   • BEIR is a heterogeneous benchmark containing diverse IR tasks. It also provides a common and easy framework for evaluation of your NLP-based retrieval models within the benchmark.
   • Weeknotes: Self-hosted language models with LLM plugins, a new Datasette tutorial, a dozen package releases, a dozen TILs
   • At Jabbr we've poured our heart and soul into building the world's first Computer Vision AI for combat sports — we call it: DeepStrike
   • How to integrate custom LLM using langchain. A GPT4ALL example.
   • Deploy models painlessly
   • TSlib is an open-source library for deep learning researchers, especially deep time series analysis.
   • Are Vector Databases here to stay? 🔍 Yes, it seems LLMs are lost in the Middle and lost focus on long inputs.🗺👁‍🗨
   • How to Use AI to Do Stuff: An Opinionated Guide
   • How to test ML models in the real world
   • embeddings: introduction - This course is about embeddings, mainly the intuition behind them.
   • 9 Distance Measures in data science with algorithms.
   • 🔘 MIT Deep Learning in Life Sciences
   • Awesome MLOps
2. 하드웨어
   • 삼성전자 "HBM 생산능력, 최소 2배 이상 증설중"
   • iOS 17 베타 4 코드로 짐작하는 아이폰 15 프로 ‘동작 버튼’의 9가지 역할
   • 아이패드가 ‘마침내’ 노트북을 대체할 수 있게 된 이유
   • "울트라 이더넷 만든다" AI 인프라용 네트워크 표준 위해 컨소시엄 결성
   • 가민 포러너 255S 리뷰 | 철인 3종 경기에도 적합한 ‘가성비’ 스포츠 워치
   • 엔비디아에서 AMD 그래픽 카드로 바꾼 이유 4가지
   • Mali GPU Training
   • Putting the “You” in CPU
   • When Open Becomes Opaque: The Changing Face of Open-Source Hardware Companies
   • 다가오는 양자 컴퓨팅 시대, 기업이 알아야 할 '양자 어닐링'
   • ‘AI, 암호화폐, 메타버스까지…’ 엔비디아가 시총 1조 달러를 넘어서기까지
   • Fine-tuning is the new training. Here are 2 open-source Python libraries to help you 🐍↓
   • Chapyter is a JupyterLab extension that seamlessly connects GPT-4 to your coding environment.
   • My First Immersion in Apple Vision Pro: Heavy, Man!
   • Insights from a laboratory fire
   • GN⁺: EU, 모든 스마트폰에 교체 가능한 배터리를 의무화하는 규정에 공식적으로 동의 (androidauthority.com)
3. 읽을거리
   • 재난의 그늘 - “과중한 업무·책임만 강요…재난 전문가 되기 전에 다 떠나”
   • 부산광역시에서 직접 제작한 부산 맛집 지도
   • 49세에 퇴직하지만… 고령층 “73세까지 일하고 싶다”
   • "상대의 의견은 공격하되 사람은 공격하지 않는다." ㅡ루스 긴즈버그(미국 연방대법관)
   • 크리에어터나 인풀루엔서가 되고 싶다면 필독해야 되는 인터뷰:
   • 짜증나는 유튜브 '그 게임', 이제 진짜 해볼 수 있다 - "돈 주고 플레이하는 행위 자체가 개그"
   • "좌식시간 세계 최장 한국인…신체활동은 선택 아닌 필수"
   • 비행기 뜬지 1시간만에 '펑'…111명 목숨 앗아간 최악의 이 사고
   • 휴·폐업 급증에 부실률 치솟아…위탁보증 예산 고갈 3년 빨라져
   • "독자도, 저자도 출판사로 모여요" 출판사의 변신은 무죄
   • 어떤 폰 써? "갤럭시 69%, 아이폰 23%"…20대는 '정반대'
   • 원격의료 결국 탈났다 플랫폼 4곳 운영 중단
   • 국내 여행 때 ‘현지인 맛집’을 찾기 위해 쓰는 방법
   • 2022년 기준 한국 출판 생산 통계

(보너스: Run Llama2 on your MacBook with GPU support! 🔥🔥🔥 via @jkronand)

EOB