Generate images in one second on your Mac using a latent consistency model
Generate images in one second on your Mac using a latent consistency model Latent consistency models (LCMs) are based on Stable Diffusion, but they can generate images much faster, needing only 4 to 8 steps for a good image (compared to 25 to 50 steps). By running an LCM on your M1 or M2 Mac you can generate 512x512 images at a rate of one per second. Simian Luo et al released the first Stable Dif
GitHub - radames/Real-Time-Latent-Consistency-Model: App showcasing multiple real-time diffusion models pipelines with Diffusers
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
Airbnb’s Trip to Linaria
CSS is a critical component of every web application, and many solutions have evolved for how styles are written by developers and delivered to visitors. In this post we’ll take you through Airbnb’s journey from Sass to CSS-in-JS and show you why we landed on Linaria, a zero-runtime CSS-in-JS library, and the impact it has had on the developer experience and performance of Airbnb’s web app. From S
A new JavaScript backend was merged into GHC on November 30th, 2022! This means that the next release of GHC will be able to emit code that runs in web browsers without requiring any extra tools, enabling Haskell for both front-end and back-end web applications. In this post, we, the GHC DevX team at IOG, describe the challenges we faced bringing GHCJS to GHC, how we overcame those challenges, and
South Korea’s ‘Sender Pays’ Policy Is a Threat to the Internet
The “sender pays” model is poised to create fragmentation of digital rules and standards, compromising the digital ecosystems that have been at the heart of the internet’s innovation and growth. Credit: Depositphotos When the presidents of South Korea and the United States met in May, they highlighted their shared belief in the benefits afforded by an open, global, interoperable, reliable, and sec
How we built our multi-platform design system at Booking.com
Building a design system that works is a challenge at any scale. Building a design system for 150+ product teams, used by 200+ designers and 800+ developers, and serving 4 different platforms? That’s a challenge requiring lots of special considerations. When building a design system for such a large scale, many questions come to mind: How can we make it strong and scalable to multiple themes and b
How Figma's Databases Team Lived to Tell the Scale | Figma Blog
Vertical partitioning was a relatively easy and very impactful scaling lever that bought us significant runway quickly. It was also a stepping stone on the path to horizontal sharding. Figma’s database stack has grown almost 100x since 2020. This is a good problem to have because it means our business is expanding, but it also poses some tricky technical challenges. Over the past four years, we’ve
building-software.md
building-software.md Some thoughts on building software Lately I have been busy reading some new books on Domain Driven Design (DDD) and software architecture -- including a short yet eye-opening one in Python and a great one in F#. At the same time, it seems that more people in the Functional Programming world are looking at more formal approaches to modelling -- some examples here. This has brou
Techouse社内勉強会の内容を紹介します(1) データベース<ACID編> - Techouse Developers Blog
Techouseの「エンジニア基礎勉強会」とは Techouse では「基礎勉強会」と称して2週間に1回、わたしが OS・ネットワーク・データベース・ハードウェア・セキュリティ・システムアーキテクチャなどをお話する勉強会を開催しています。 講師は私ひとり、資料を準備するのも私ひとり、動画を収録して YouTube Live で社内向けに配信する作業も私ひとりでやってます。 参加は任意ですが、社内のメンバー (社員・インターン生・業務委託でご参画いただいている方) の多くの方が参加してくれています。先日の RubyKaigi 2024 に参加してくれたメンバーもほとんどがこの勉強会に参加し、基礎的な知識をもった上でセッションへ臨んでくれました。 開催履歴 これまでの開催履歴はこんな具合です。 見ていただくとわかる通り、ほんとうに基礎的な内容を1個ずつやっているということがわかるかと思います。
It took a little longer than I expected, but we released HTML5 Boilerplate 8.0.0 yesterday. As I mentioned a few weeks ago, it’s the 10th anniversary release, and if I do say so myself, it’s a pretty good way to celebrate 10 years of the project. It was a lot of work getting this one out the door, but I’m really happy with how it turned out. I’ve been involved with HTML5-Boilerplate from before th
EKS vs GKE vs AKS - Evaluating Kubernetes in the Cloud | StackRox Community
Providing an update on the core Kubernetes services offered by the big three We are now six years past the initial release of Kubernetes, and it continues to be one of the fastest-growing open-source projects to date. The rapid development and adoption of Kubernetes have resulted in many different implementations of the application. The Cloud Native Computing Foundation (CNCF) currently lists over
research!rsc: Hardware Memory Models (Memory Models, Part 1)
Introduction: A Fairy Tale, Ending A long time ago, when everyone wrote single-threaded programs, one of the most effective ways to make a program run faster was to sit back and do nothing. Optimizations in the next generation of hardware and the next generation of compilers would make the program run exactly as before, just faster. During this fairy-tale period, there was an easy test for whether
July 2020 (version 1.48) Update 1.48.1: The update addresses these issues. Update 1.48.2: The update addresses these issues. Downloads: Windows: x64 Arm64 | Mac: Intel | Linux: deb rpm tarball snap Welcome to the July 2020 release of Visual Studio Code. There are a number of updates in this version that we hope you will like, some of the key highlights include: Settings Sync to Stable - Settings S
We introduce EDGE, a powerful method for editable dance generation that is capable of creating realistic, physically-plausible dances while remaining faithful to arbitrary input music. EDGE uses a transformer-based diffusion model paired with Jukebox, a strong music feature extractor, and confers powerful editing capabilities well-suited to dance, including joint-wise conditioning, motion in-betwe
Ubuntu Weekly Recipe 第749回LXDとUbuntuサーバーで、多数のストレージが必要なRAID環境を構築し性能を比較する 第748回の「LXDとUbuntuサーバーでRAIDとmdadmのテスト環境を構築する」では、LXDを使ってかんたんに単一のマシンにRAID 1環境を構築する方法を紹介しました。今回は前回紹介しきれなかった残りのRAIDレベルを構築し、それぞれの性能を比較してみましょう。 さまざまなRAIDレベルたち RAIDにはいくつかのレベルとそれを組み合わせたものが存在します[1]。そのうちUbuntuがインストーラーとしてサポートしているのは、RAID 0、1、5、6、10の5種類です。もちろんインストーラーはUbuntuのルートファイルシステムを構築するための仕組みなので、純粋なデータ領域をRAIDとして構築するのであれば、mdadmコマンドを用いてさら
序 この記事、ちょっと、というかかなり感覚が古くね？ qiita.com と思っていたら、炎上目的だったようで。 autotaker on Twitter: "「形式手法はなぜ流行っていないのか？」という記事を書いて炎上させたい" autotaker on Twitter: "形式手法をこき下ろすと見せかけて最終的に出身研究室のステマに成功した" 一連の言い訳も見苦しい。 autotaker (@autotaker1984) | Twitter 故意に不正確な情報をばら撒く形でしか自分の"小さな得意分野"をアピールできないとは、エンジニアとして不誠実極まりないし、ミジメだ。 本題 ということで。 形式手法は80年代から本格的に開発が始まっているが、最初に大規模に報道されたのは90年代のIntelの不動小数点ユニットのバグ対策に形式手法が導入された件だろう。 つまりチップレベルでの導入。それ
【感想】『データ指向アプリケーションデザイン』：深遠なる分散データの新大陸への旅 - Rのつく財団入り口
読破した分厚いオライリー本の感想記事です。本書ではCPUの速度がボトルネックになるようなものは演算指向アプリケーションと区別し、データの量や複雑さ、変化の速度が主題となるシステムを「データ指向」と位置づけて、特定技術に幅を狭めずに包括的に解説した本となっています。 著者はイギリス、ケンブリッジ大学の分散システムの研究者 Martin Kleppmann氏。監訳者が斉藤太郎氏、訳者は玉川竜司氏。 タイトルの『データ指向アプリケーションデザイン』の原題は Designing Data-Intensive Applications。よく使われる「オブジェクト指向」の原語は Object-Oriented ですが、本書の「指向」は Intensive で若干ニュアンスが違います。たまに見るデータ駆動、データドリブンなどともちょっと違いますね。 Intensive単体の意味は強い、激しい、徹底的、集
IntroductionInspired by the cross-functional cooperation promised by DevOps, platform engineering has begun to emerge in enterprises as an explicit form of that cooperation. Platforms curate and present foundational capabilities, frameworks and experiences to facilitate and accelerate the work of internal customers such as application developers, data scientists and information workers. Particular
メモリモデル入門（Sequential ConsistencyとTotal Store Orderを理解する）
L1: r1 = y;L2: r2 = x; 並列プログラムには非決定性があるので、結果はひとつには決まりません。いくつかパターンを考えてみると、 r1, r2 = (NEW, NEW), (0, NEW), (NEW, 0) の可能性があることがわかります: S1 < S2 < L1 < L2 → (NEW, NEW)S1 < L1 < S2 < L2 → (0, NEW)S2 < L2 < S1 < L1 → (NEW, 0) では r1, r2 = (0, 0) という結果はありうるでしょうか？ 少し考えてみるとこれはなさそうに思えます： r1 = 0を仮定するこのときL1 < S2である各コアのプログラムの順序よりS1 < L1, S2 < L2である2と3よりS1 < L1 < S2 < L2という順序が決定されるS1 < L2よりr2 = NEWである 以上の「素朴な推論」は正
Jun 6, 2023 by Patrick Angle, Jean-Yves Avenard, Marcos Caceres, Ada Rose Cannon, Eric Carlson, Garrett Davidson, Jon Davis, Karl Dubost, Brady Eidson, Matthew Finkel, Simon Fraser, Brent Fulgham, Rachel Ginsberg, David Johnson, Anne van Kesteren, Mark Lam, Sihui Liu, Justin Michaud, Jer Noble, Tim Nguyen, Ben Nham, Richard Robinson, Michael Saboff, Alexey Shvaika, Jen Simmons, Sam Sneddon, Brando
TMTOWTDIのつづき。Bicarbonate（けれど、一貫性を保つことは悪くないこともある） - Blog::kobaken
Perl のモットーは「何をするにもやり方は一つだけじゃない」（TMTOWTDI; There's more than one way to do it. 読みは Tim Toady）で、自分は魅力を感じる。それは、Perlのコミュニティから、Perlに限らずさまざまな技術を愛する懐の深さを感じるから。実際どうかはYAPC::Hiroshima 2024のタイムテーブルがわかりやすい例だと思う。 愛なんて言葉を使ったけれど、この記事を書くため、ラリー・ウォールのインタビューを読んだせいかもしれない。 「何をするにもやり方は一つだけじゃない」（TMTOWTDI; There's more than one way to do it.）の哲学は、宇宙の創造者の謙虚さを観察した直接の結果であり、荘厳ではなく軽妙に処理できるよう選択した結果だ。宇宙には、押し付けるべきスタイルのガイドラインなどない
MagicAnimate: Temporally Consistent Human Image Animation using Diffusion Model
TL;DR: We propose MagicAnimate, a diffusion-based human image animation framework that aims at enhancing temporal consistency, preserving reference image faithfully, and improving animation fidelity.
(略称多すぎですね。) 0. 忙しい方へ GANにConsistency Regularizationを適用することで性能を大きく改善したよ Improved-CRは「balanced-CR + latent-CR」だよ bCRはリアル画像とフェイク画像両方にCRしたものだよ zCRはノイズ空間のCRをしたものだよ CRもICRも超絶シンプルで強力だよ CRとICRはSNとの相性が良く、他の正則化手法を大きく上回ったよ ICRはCIFAR-10(FID:9.21)とImageNet(FID:5.38)でBigGANにおけるSoTAを達成したよ "Improved Consistency Regularization for GANs", Zhao, Z., Singh, S., Lee, H., Zhang, Z., Odena, A., Zhang, H., (2020)より作成 1.
【感想】『マイクロサービスパターン 実践的システムデザインのためのコード解説』：前編 - Rのつく財団入り口
「未来はすでにここにある。まだむらなく流通していないだけだ」←グッとくる 最初のエモワードがSF作家ウィリアム・ギブスンの引用でイイ！ サイバーパンク2077遊んでみた～い……じゃなかった、CloudFoundry.comのファウンダーでありMicroservices.ioの運営者、経験豊富なソフトウェアアーキテクトであるクリス・リチャードソンさんによる『Microservices Patterns』の翻訳本。 タイトルのようにアーキテクチャパターンやデザインパターンのようにマイクロサービスをパターンで体系化し、サンプルストーリーを元にした事例やコード例、OSS紹介を交えつつマイクロサービスを実践する設計方法を探求した本となっています。 Java文化圏で長く活動してきた方とのことでサンプルコードはほぼJava、Springフレームワーク、ご本人らによるマイクロサービス用のフレームワークEv
The Go Programming Language and Environment – Communications of the ACM
Go is a programming language created at Google in late 2007 and released as open source in November 2009. Since then, it has operated as a public project, with contributions from thousands of individuals and dozens of companies. Go has become a popular language for building cloud infrastructure: Docker, a Linux container manager, and Kubernetes, a container deployment system, are core cloud techno
name: Code Review permissions: contents: read pull-requests: write on: pull_request: pull_request_review_comment: types: [created] concurrency: group: ${{ github.repository }}-${{ github.event.number || github.head_ref || github.sha }}-${{ github.workflow }}-${{ github.event_name == 'pull_request_review_comment' && 'pr_comment' || 'pr' }} cancel-in-progress: ${{ github.event_name != 'pull_request_
🌱 This post is still growing and likely will change as the best practices evolve 🌳 GraphQL errors are something many of us struggle with. Good practices are emerging but the community has not yet settled on a convention. I hope this guide helps demystifiy the many ways you can structure errors in your GraphQL servers, and the tradeoffs each of them make. Stage 1: GraphQL Errors AKA Top-Level Err
24 Things Great Scrum Masters Don’t Do. » Growing Scrum Masters
Being a great scrum master is not just about what you do but also about what you don’t do. It’s all about sidestepping the pitfalls, avoiding the traps, and resisting the lure of those seemingly practical shortcuts that ultimately lead down the rocky road to chaos. In this spirit, we present to you your survival guide, a list of “24 Things Great Scrum Masters Don’t Do” This delightful catalogue of
ResearchOps 101
Summary: The practice of Research Operations (ResearchOps) focuses on processes and measures that support researchers in planning, conducting, and applying quality research at scale. ResearchOps is a specialized area of DesignOps focused specifically on components concerning user-research practices. ResearchOps (ReOps): The orchestration and optimization of people, processes, and craft in order to
A handy new Google Cloud, AWS, and Azure product map | Google Cloud Blog
A handy new Google Cloud, AWS, and Azure product map Any craftsman will tell you that choosing the right tool for the job is essential for getting it done right. Cloud technologies are no different. Many cloud professionals look for the best products across vendors, but they remember ‘best’ is always subjective. It depends on highly-individualized criteria like language support, compatibility with
Novel View Synthesis with Diffusion Models 3D generation from a single image We present 3DiM, a diffusion model for 3D novel view synthesis, which is able to translate a single input view into consistent and sharp completions across many views. The core component of 3DiM is a pose-conditional image-to-image diffusion model, which is trained to take a source view and its pose as inputs, and generat
Tips on Adding JSON Output to Your CLI App - Brazil's Blog
Brazil's Blog Musings on automation, scripting, programing, DevOps, and cybersecurity A couple of years ago I wrote a somewhat controversial article on the topic of Bringing the Unix Philosophy to the 21st Century by adding a JSON output option to CLI tools. This allows easier parsing in scripts by using JSON parsing tools like jq, jello, jp, etc. without arcane awk, sed, cut, tr, reverse, etc. in
Why the developers who use Rust love it so much - Stack Overflow
The 2020 Developer Survey results are in, and once again, Rust is the number one most loved language among the 65,000 programmers who participated. Rust has taken the number one spot since 2016, showing that it’s got something that the developers who use it love. 86.1% of those survey respondents who said they currently use Rust want to continue using it next year. This is the fifth year in a row
Celebrate 15 Years of Amazon S3 with ‘Pi Week’ Livestream Events | Amazon Web Services
AWS News Blog Celebrate 15 Years of Amazon S3 with ‘Pi Week’ Livestream Events I wrote the blog post that announced Amazon Simple Storage Service (Amazon S3) fifteen years ago today. In that post, I made it clear that the service was accessed via APIs and that it was targeted at developers, outlined a few key features, and shared pricing information. Developers found that post, started to write co
GitHub - PatrickJS/awesome-cursorrules: 📄 A curated list of awesome .cursorrules files
.cursorrules is a powerful feature in Cursor AI that allows developers to define project-specific instructions for the AI. Here's why you might want to use it: Customized AI Behavior: .cursorrules files help tailor the AI's responses to your project's specific needs, ensuring more relevant and accurate code suggestions. Consistency: By defining coding standards and best practices in your .cursorru
Buy Read online Discover the beauty of mathematics through the lens of category theory with this unique exploration that goes beyond problem-solving. In this book, you’ll find a refreshing perspective on math as an art form, a language, and a way of thinking that unifies diverse fields of knowledge. Delve into the fascinating world of sets, groups, types, and partial orders, and explore the deep c
GitHub - openai/consistency_models: Official repo for consistency models.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session. Dismiss alert
A step-by-step guide with my learnings from 30+ AI implementations. Earlier this year, I wrote Your AI product needs evals. Many of you asked, “How do I get started with LLM-as-a-judge?” This guide shares what I’ve learned after helping over 30 companies set up their evaluation systems. The Problem: AI Teams Are Drowning in Data Ever spend weeks building an AI system, only to realize you have no i
LCM LoRA の概要｜npaka
以下の記事が面白かったので、かるくまとめました。 ・SDXL in 4 steps with Latent Consistency LoRAs 1. はじめに「LCM」 (Latent Consistency Model) は、元モデルを別モデルに蒸留することで、画像生成に必要なステップ数を減らす手法です。25～50ステップかかっていた処理を4～8ステップで可能にします。 蒸留したモデルは、より小さくなるように設計される場合 (DistilBERT、Distil-Whisperなど)と、必要なステップ数が少なくなるように設計される場合があります。これは通常、膨大な量のデータ、忍耐力、少数のGPU を必要とし、時間とコストが必要でした。しかし、それも過去の話です。 11月9日 、「LCM」で蒸留したかのように「Stable Diffusion」「SDXL」を本質的に高速化できる新手法「LCM
An oral history of Bank Python November 2021 The strange world of Python, as used by big investment banks High finance is a foreign country; they do things differently there Today will I take you through the keyhole to look at a group of software systems not well known to the public, which I call "Bank Python". Bank Python implementations are effectively proprietary forks of the entire Python ecos
j次のブックマーク
k前のブックマーク
lあとで読む
eコメント一覧を開く
oページを開く
JavaScript backend merged into GHC | IOG Engineering
HTML5 Boilerplate 8.0.0 Released
Visual Studio Code July 2020
EDGE: Editable Dance Generation from Music
第749回 LXDとUbuntuサーバーで、多数のストレージが必要なRAID環境を構築し性能を比較する | gihyo.jp
「形式手法はなぜ流行っていないのか」に対する異論 - interdb’s blog
CNCF Platforms White Paper
News from WWDC23: WebKit Features in Safari 17 beta
GANへの新しい正則化「ICR」が期待大な件&解説 - Qiita
CodeRabbit openai-pr-reviewer を調整する
A Guide to GraphQL Errors
Novel View Synthesis with Diffusion Models
Category Theory Illustrated - index
Creating a LLM-as-a-Judge That Drives Business Results –
An oral history of Bank Python