[B급 프로그래머] 5월 1주 소식(빅데이터/인공지능, 하드웨어/블록체인, 읽을거리 부문)

(오늘의 짤방: This meme was leaked from that internal meme site. via @untitled01ipynb)

1. 빅데이터/인공지능
   • eBay’s Blazingly Fast Billion-Scale Vector Similarity Engine
   • This is the official code and model release for Shap-E: Generating Conditional 3D Implicit Functions.
   • LMFlow - An extensible, convenient, and efficient toolbox for finetuning large machine learning models, designed to be user-friendly, speedy and reliable, and accessible to the entire community.
   • The StarCoder models are 15.5B parameter models trained on 80+ programming languages from The Stack (v1.2), with opt-out requests excluded.
   • llm is a Rust ecosystem of libraries for running inference on large language models, inspired by llama.cpp.
   • Introducing MPT-7B: A New Standard for Open-Source, Commercially Usable LLMs
   • Open LLMs - These LLMs are all licensed for commercial use (e.g., Apache 2.0). Contributions and corrections welcome!
   • Today's top trending AI news stories and papers
   • Re-implementing LangChain in 100 lines of code (scottlogic.com)
   • Retriever Router Query Engine
   • RLHF: Reinforcement Learning from Human Feedback
   • OK Computer: Romania debuts ‘world’s first’ AI government adviser
   • writeout.ai - Transcribe and translate
   • A quick and sobering guide to cloning yourself
   • What is a Vector Database?
   • supabase-template: Supabase + @LangChainAI starter template.
   • ChatGPT Prompt Engineering for Developers
   • ChatGPT Prompt Engineering for Developers 노트 정리
   • Fixing Hallucination with Knowledge Bases
   • Lamini - LLM 훈련용 instruction 셋을 만들어주는 데이터 생성기 (github.com/lamini-ai)
   • 기업용 AI 코딩 도구를 도입할 때 검토해야 할 8가지
   • Bing Chat adds the best of ChatGPT and more in major feature announcement
   • AI가 쓴 리뷰, AI로 막아보자··· 가짜 리뷰 분석 기업 페이크스팟, 모질라에 인수
   • “같은 듯 다른” 유럽과 미국의 생성형 AI 규제 동향과 시사점
   • 원드라이브 웹 버전, 사람 중심 UI로 바뀐다 “사람·회의별로 파일 분류”
   • 코딩 가능한 AI Bot들 비교: Bing, Claude+, CoPilot, GPT-4, Bard (hackernoon.com)
   • 박스, 오픈AI와 제휴해 제품군 전반에 생성형 AI 통합 발표
   • The Amazing AI Super Tutor for Students and Teachers | Sal Khan | TED
   • gpt-repository-loader is a command-line tool that converts the contents of a Git repository into a text format, preserving the structure of the files and file contents.
   • AG3D: Learning to Generate 3D Avatars from 2D Image Collections
   • Announcing the next wave of AI innovation with Microsoft Bing and Edge
   • GPTutor: a ChatGPT-powered programming tool for code explanation
   • GPTutor - VS Code Extension
   • Leaked Google document: “We Have No Moat, And Neither Does OpenAI”
   • ChatGPT Lifts Business Professionals’ Productivity and Improves Work Quality
   • OpenLLaMA: An Open Reproduction of LLaMA
   • "AI 가정교사 온다" 美사교육기업 초토화
   • Pandas AI is a Python library that adds generative artificial intelligence capabilities to Pandas,
   • Replit Code - 10일만에 훈련된 오픈소스 코드 생성 LLM (huggingface.co)
   • ③ MS, '빙'으로 구글 검색 붕괴시킬까?
   • 정부 "챗GPT 쓰면 공공 정보 유출 우려"...우선 국정원 허락 받아라
   • [챗GPT 본고장을 가다]③ ‘기업가치 266조’ 세일즈포스 본사에 아인슈타인 조형물이… “기업용 AI 시장 선점에 사활”
   • AI '일자리 습격' 시작됐다…美IBM, 업무직 30% 대체
   • Whisper Jax: 가장 빠른 Whisper API 구현체 (github.com/sanchit-gandhi)
   • Chatbot Arena: Benchmarking LLMs in the Wild with Elo Ratings
   • Bark is a transformer-based text-to-audio model created by Suno.
   • Amazon, Alexa용 개선된 LLM 개발 중 (techcrunch.com)
   • “빙 API 이용비 너무 높아져 부담”…브레이브, 빙 검색 제휴 중단
   • AWS, 분기 실적과 대응 전략 발표··· “생성형 AI로 중심 이동”
   • Scared of Leaking Data to ChatGPT? Microsoft Tests a Private Alternative
   • A decoder that uses brain scans to know what you mean — mostly
   • Speak, Memory: An Archaeology of Books Known to ChatGPT/GPT-4
   • When A.I. Chatbots Hallucinate
   • Little Book of Deep Learning [143p PDF] (fleuret.org)
   • AI isn’t the app, it’s the UI
   • GPT makes learning fun again
   • The Embedding Archives: Millions of Wikipedia Article Embeddings in Many Languages
   • Bonjour. مرحبا. Guten tag. Hola. Cohere's Multilingual Text Understanding Model is Now Available
   • The 2023 MAD (ML/AI/Data) Landscape
   • ‘The Godfather of A.I.’ Leaves Google and Warns of Danger Ahead
   • 'AI의 대부' 제프리 힌튼 교수, 구글을 떠나며 AI의 위험성에 대해 경고 (nytimes.com)
   • This library provides Go clients for OpenAI API. We support: ChatGPT, GPT-3, GPT-4, DALL·E 2, Whisper
   • ChatGPT’s context length doesn’t work the way u think it does
   • Deep learning pioneer Geoffrey Hinton has quit Google
   • A Quick Review of Pre-Calculus
   • OCR 데이터(고서한자)
   • AI science search engines are exploding in number — are they any good?
   • If It Weren't for This Equation, You Wouldn't Be Here
   • Boosting Theory-of-Mind Performance in Large Language Models via Prompting
   • Hnswlib - fast approximate nearest neighbor search
   • DeepFloyd IF - 오픈소스 Text-to-Image 모델 (github.com/deep-floyd)
   • MLC LLM is a universal solution that allows any language model to be deployed natively on a diverse set of hardware backends and native applications,
   • How AI coding companions will change the way developers work
   • [체험기] 구글 '바드' vs. MS '빙'...누가 검색 더 잘할까
   • Glaze: Protecting Artists from Style Mimicry
   • Introducing Lamini, the LLM Engine for Rapidly Customizing Models
   • 🤖 SalesGPT - Your Context-Aware AI Sales Assistant
   • AI 디자인 툴 ‘마이크로소프트 디자이너’ 일반에 공개…빙에도 통합 예정
   • 깃허브 코파일럿·아마존 코드위스퍼러에 맞선다… 구글 바드, 코드 생성 기능 추가
   • 구글 AI 사령탑 하나로 통합··· 'Google DeepMind'가 연구와 개발 지휘
   • '거부하기 힘든 독 사과' 생성형 AI로 코딩하기
   • “잊으라면 잊겠어요”··· 오픈AI, 챗GPT에 데이터 프라이버시 통제 기능 추가
   • 스파크로 오픈소스 데이터 관리 쉽게…데이터브릭스, 허깅페이스와 맞손
   • 수익은 어떻게?··· 마이크로소프트의 AI 매출 계획 엿보기
   • “개발자 95%가 AI 코딩 도구 이용···코드 복잡성은 더 높아져” 소스그래프
   • 생성형 AI가 인간관계에 악영향?… 해결해야 할 3가지
   • 챗GPT는 시작에 불과... 수면 위로 드러나는 ‘AI 챗봇’의 안보 위협
   • Amazon SageMaker와 Amazon MWAA를 활용한 29CM의 개인화 추천시스템 MLOps 구축여정
   • Navigating the High Cost of AI Compute
   • Cayley is an open-source database for Linked Data. It is inspired by the graph database behind Google's Knowledge Graph (formerly Freebase).
   • Google Announces Machine Learning Powered API Abuse Detection
   • AI, write an essay for me: A large-scale comparison of human-written versus ChatGPT-generated essays
   • deepdoctection is a Python library that orchestrates document extraction and document layout analysis tasks using deep learning models.
   • Large Language Models are Versatile Decomposers: Decompose Evidence and Questions for Table-based Reasoning
   • 유럽에서 ChatGPT로: 출처 공개 (wsj.com)
   • Hidet is a powerful deep learning compiler that simplifies the process of implementing high-performing deep learning operators on modern accelerators (e.g., NVIDIA GPUs).
   • Microsoft Designer is being integrated into Edge for AI-generated social content
   • Let Us Show You How GPT Works
   • Thoughtworks Technology Radar, Volume 28 공개 (thoughtworks.com)
   • 생성형 AI가 LCNC 영역에 일으키는 파문
   • ‘이스라엘의 오픈AI’ AI21랩스가 전하는 생성형 AI의 미래
   • 챗GPT가 던진 화두… 학교에서 AI ‘쓸까 말까’
   • “거부하기 힘든 독 사과” 생성형 AI로 코딩하기
   • LLM 환각 모니터링 지원··· 어라이즈AI, ML옵스 라이브러리 오픈소스로 공개
   • A guide to prompting AI (for what it is worth)
   • ChatGPT로 테이블링 챗봇 만들기
   • A Python package for segmenting geospatial data with the Segment Anything Model (SAM)
   • Segment Anything Model for Geospatial Data
   • Scaling Transformer to 1M tokens and beyond with RMT
   • Stability AI releases its Image Upscaling API
   • PwC US makes $1 billion investment to expand and scale AI capabilities
   • Simple LLM Finetuner - 초보자 친화적인 UI를 가진 LLM 튜닝 도구 (github.com/lxe)
   • 마틴 파울러가 알려주는 ChatGPT와 페어 프로그래밍 하는 법
   • MIT CSAIL Explains: Large Language Models: Part 1
   • More Design Patterns For Machine Learning Systems
   • NVIDIA Enables Trustworthy, Safe, and Secure Large Language Model Conversational Systems
   • When Large Language Models Meet Personalization
   • Prof. Michael J. Pyrcz, @GeostatsGuy, Resources
   • The story of Prof. Michael J. Pyrcz
   • When building chat applications using LLMs, memory handling is something that has to be built every time. Motörhead is a server to assist with that process.
   • A Cookbook of Self-Supervised Learning
   • It doesn’t take much to make machine-learning algorithms go awry
   • GPT-4, LangChain & Chroma - Create a ChatGPT Chatbot for Your PDF Files
   • DeepTime: Deep Time-Index Meta-Learning for Non-Stationary Time-Series Forecasting
   • LaMP: When Large Language Models Meet Personalization
   • 구글클라우드, 생성 AI 기반 보안 서비스 공개
   • Weights & Biases, LLMOps 도구 W&B Prompts 공개 (wandb.ai)
   • [2023 DEVIEW 리뷰] 값비싼 Diffusion model을 받드는 저비용 MLOps
   • New ways to manage your data in ChatGPT - ChatGPT users can now turn off chat history, allowing you to choose which conversations can be used to train our models.
   • ChatGPT Clone
   • 거대언어모델(LLM)의 현주소
   • Instruction finetuning is how we get from GPT-3-like pretrained base models to more capable LLMs like ChatGPT.
   • OpenAI, 브랜딩 가이드라인 공개 (openai.com)
   • griptape - LLM 워크플로우를 위한 파이썬 프레임워크 (github.com/griptape-ai)
   • Generative AI at Work(해설: Findings 요약)
   • LlamaAcademy - GPT에게 코딩하는 법 가르치기 (github.com/danielgross)
   • 검색 엔진은 어떻게 작동하는가 (xo.dev)
   • WizardLM: Empowering Large Language Models to Follow Complex Instructions
   • Faiss is a library for efficient similarity search and clustering of dense vectors.
   • ⚔️ Chatbot Arena ⚔️
   • Finetuning Large Language Models
   • BIG DATA IS DEAD
   • SPTAG: A library for fast approximate nearest neighbor search
   • Low-code LLM: Visual Programming over LLMs
   • ChartGPT - A tool that converts text into beautiful charts
   • 80년대 텍스트 어드벤처 ‘포토피아’에 챗 AI 적용했더니
   • Data Observability for Analytics and ML teams
   • awesome-pretrained-models-for-information-retrieval
   • Training open-source LLMs on ChatGPT output is a really bad idea.
   • Transformers from Scratch
   • CancerGPT: Few-shot Drug Pair Synergy Prediction using Large Pre-trained Language Models
   • WasmGPT - a ChatGPT-like language model in the browser
   • WasmGPT - ChatGPT-like chatbot in the browser using ggml and emscripten.
   • 챗GPT의 시대, 챗봇 개발사의 고민
   • This new technology could blow away GPT-4 and everything like it
   • The godot-dodo project presents a pipeline to finetune open source language models on human-created, language-specific code retrieved from GitHub.
   • Leaked deck raises questions over Stability AI's Series A pitch to investors
   • 구글 ‘버텍스 AI’, 생성 인공지능을 구현하는 두 가지 새로운 기능과 모델이 상호작용하는 5가지 방법
   • LlamaAcademy: Teaching Llamas How to Code
   • Why the Brain’s Connections to the Body Are Crisscrossed
   • Tools of the AI engineer
   • Finetuning Large Language Models
   • 한국어 기초 사전에서 대화로 구성된 예문만 추출한 데이터
2. 하드웨어
   • I spoke to Apple to find out the secret behind the AirPods Pro 2’s audio success
   • 애플은 왜 아직도 에어팟 사용자를 배려하지 않는가?
   • Graviton 3, Apple M2 and Qualcomm 8cx 3rd gen: a URL parsing benchmark
   • Voyager as Technosignature
   • 플레이어의 숙련도가 머물렀던 곳은 어디였을까: 터치스크린 시대의 숙련도
   • Advanced AMIGA Architecture (June 18, 1992)
   • Windows Dev Kit 2023
   • The LiDAR Revolution
   • Envisioning the W93
   • 가트너, “2023년 전 세계 반도체 매출 11% 감소 전망”
   • Driver adventures for a 1999 webcam
   • 겔싱어 CEO 리더십 2년, 인텔의 턴어라운드 현황은?
   • The Glorious Return of a Humble Car Feature
   • The Hardware Lottery
3. 블록체인과 메타버스
   • [이슈] 찬바람 부는 NFT 시장… 타이거글로벌, 오픈씨 기업가치 ‘76% 낮춰’
   • 구글 클라우드-폴리곤랩스, 개발자 도구 및 인프라 제공 협력
4. 읽을거리
   • '억대 연봉'만 100만 명이라는데..연봉 차이는 1,300만 원, 왜?
   • "부자들은 배신 안 해" 큰소리치더니…137조 증발에 '패닉'
   • Anthropologist pair solve the mystery of Mayan 819-day count
   • The Long Game
   • [마켓관찰] 불황때 중고가 안 팔리는 이유는
   • 우리나라와 미국의 근원인플레이션 압력 평가
   • [애플페이 한 달] ④“결국 삼성과 간편결제 양분, 군소 ‘페이’ 사라질 듯”
   • 2022년 출판시장 통계 보고서 (수정 5.2.)
   • Why Gravity Is Not Like the Other Forces
   • 아마존, 네이버, 스타벅스..공룡 기업이 '커뮤니티' 하는 이유
   • 2023 Google 검색 플레이북 (thinkwithgoogle.com)
   • France’s struggle to deliver a second nuclear era
   • You’re probably recycling wrong. This quiz will help you sort it out.

보너스: True or false? And what is missing? via @alexxubyte

EOB