ChatGPTが本当にヤバい。 断言する。新卒がこれを使いこなせば、今職場で「優秀」とされている5-6年目くらいの先輩なら余裕で出し抜ける。鬼になれる。 筆者はメーカー社員なので、メーカーの新入社員がChatGPTを使って鬼になる方法を1つ提案したい。 「ChatGPT×Python」である。 Pythonとは、ご存知のとおり物理シュミレーションからデータサイエンス、機械学習までカバーする汎用性をそなえたプログラミング言語だ。何でもできるわりには書ける人がなぜか少なく、いまだにスキルとして重宝されている。 そんなPythonにChatGPTを使おう。 ChatGPTを使えば、上司から求められるアウトプットを一瞬で出すことができる。それに対してフィードバックをもらい、それも一瞬で打ち返すことができる。 「あいつ"Python書ける"だけじゃないんだよな。こっちが言ったこと正確に理解するし、そ
N番目の素数を求める - すぎゃーんメモ
SNSなどで話題になっていたので調べてみたら勉強になったのでメモ。 環境 Pythonでの実装例 例1 例2 例3 エラトステネスの篩 Rustでの実装例 試し割り法 エラトステネスの篩 アトキンの篩 おまけ: GMP Benchmark 高速化のテクニック 上限個数を見積もる Wheel factorization オチ Repository References 環境 手元のMacBook Pro 13-inchの開発機で実験した。 2.8 GHz Intel Core i7 16 GB 2133 MHz LPDDR3 Pythonでの実装例 例1 最も単純に「2以上p未満のすべての数で割ってみて余りが0にならなかったら素数」とする、brute force 的なアプローチ。 import cProfile import io import pstats import sys def m
プロと読み解くRuby 3.4 NEWS テクノロジー部門技術基盤グループの笹田（ko1）と遠藤（mame）です。Ruby (MRI: Matz Ruby Implementation、いわゆる ruby コマンド) の開発をしています。お金をもらって Ruby を開発しているのでプロの Ruby コミッタです。 本日 12/25 に、恒例のクリスマスリリースとして、Ruby 3.4.0 がリリースされました（Ruby 3.4.0 リリース ）。今年も STORES Product Blog にて Ruby 3.4 の NEWS.md ファイルの解説をします（ちなみに、STORES Advent Calendar 2024 の記事になります。他も読んでね）。NEWS ファイルとは何か、は以前の記事を見てください。 プロと読み解く Ruby 2.6 NEWS ファイル - クックパッド開発者
ChatGPT deep researchに見るAIが自律的に考える未来 - LayerX エンジニアブログ
こんにちは、LayerXプロダクトマネージャーの野畑(@isseinohata)です。 LayerXで生成AIプラットフォーム Ai Workforceの開発に従事しています。 getaiworkforce.com 2月3日にOpenAIが発表したAIエージェント「deep research」が大きな話題を呼んでいます。 openai.com 生成AIの領域では日々さまざまなプロダクトや新しい技術が登場していますが、その中でもdeep researchは単なるサービス自体の性能の高さに加え、それを実現する技術（人間のリサーチプロセスに近い思考を実現する技術）に対して、未来への大きなインパクトを感じさせる体験でした。 実際、deep researchの調査ログを眺めていると、あたかも人間が試行錯誤するように、自律的に計画→検索→読み込み→発見→方針変更を進めているような姿が見て取れます。 左
Developers are increasingly relying on AI coding assistants to accelerate our daily workflows. These tools can autocomplete functions, suggest bug fixes, and even generate entire modules or MVPs. Yet, as many of us have learned, the quality of the AI’s output depends largely on the quality of the prompt you provide. In other words, prompt engineering has become an essential skill. A poorly phrased
GitHub - modelcontextprotocol/servers: Model Context Protocol Servers
Official integrations are maintained by companies building production ready MCP servers for their platforms. 21st.dev Magic - Create crafted UI components inspired by the best 21st.dev design engineers. ActionKit by Paragon - Connect to 130+ SaaS integrations (e.g. Slack, Salesforce, Gmail) with Paragon’s ActionKit API. Adfin - The only platform you need to get paid - all payments in one place, in
January 30, 2023 In this post, we'll implement a GPT from scratch in just 60 lines of numpy. We'll then load the trained GPT-2 model weights released by OpenAI into our implementation and generate some text. Note: This post assumes familiarity with Python, NumPy, and some basic experience with neural networks. This implementation is for educational purposes, so it's missing lots of features/improv
PacketProxyで探るGemini CLIのコンテキストエンジニアリング 〜AIエージェントを信頼できる相棒に〜 | BLOG - DeNA Engineering
2025.07.18 技術記事 PacketProxyで探るGemini CLIのコンテキストエンジニアリング 〜AIエージェントを信頼できる相棒に〜 by akira.kuroiwa #gemini-cli #ai #security #aiエージェント #コンテキストエンジニアリング #packetproxy 「なんかよく分からないけど、すごい」で終わらせないために こんにちは、DeNA セキュリティ技術グループの 黒岩 亮 ( @kakira9618 ) です。 AIエージェント、とくに Gemini CLI のようなコーディングを支援してくれるツールは非常に強力で、私たちの開発体験を大きく変えようとしています。しかし、その一方で、こんな風に感じたことはありませんか？ 「このファイルの情報、勝手にAIに送られたりしない？ 大丈夫かな？」 と、情報管理・セキュリティ面で漠然とした不安を
話題のDeepSeek-R1が1.58bで動くようになったので早速試してみた。 これだと、H100 80GBx2で全てVRAMに乗せて動かすことができる。 継之助なら8台あるので4つ動かせることになる。やったぜ! 「秋葉原を舞台にしたラブストーリーを全て 日本語で書け。12話で完結するようにしろ。先に構成を決め、それから各話を三幕構成で全て書け」というプロンプトを与えてみた。 t$ ./llama.cpp/llama-cli --model DeepSeek-R1-GGUF/DeepSeek-R1-UD-IQ1_S/DeepSeek-R1-UD-IQ1_S-00001-of-00003.gguf --cache-type-k q4 _0 --threads 12 -no-cnv --n-gpu-layers 61 --prio 2 --temp 0.6 --ctx-size 18192 -
The Scary Thing About Automating Deploys - Engineering at Slack
Most of Slack runs on a monolithic service simply called “The Webapp”. It’s big – hundreds of developers create hundreds of changes every week. Deploying at this scale is a unique challenge. When people talk about continuous deployment, they’re often thinking about deploying to systems as soon as changes are ready. They talk about microservices and 2-pizza teams (~8 people). But what does continuo
What We Learned from a Year of Building with LLMs (Part I)
It’s an exciting time to build with large language models (LLMs). Over the past year, LLMs have become “good enough” for real-world applications. The pace of improvements in LLMs, coupled with a parade of demos on social media, will fuel an estimated $200B investment in AI by 2025. LLMs are also broadly accessible, allowing everyone, not just ML engineers and scientists, to build intelligence into
こんにちは、NTTドコモR&D戦略部の門間です。 この記事では、vLLMのコードを追いつつその中身の動きに迫りたいと思います。 最近、業務やプライベートでLLM関連のいろいろを触っていますが、 OSSのコードリーディングを通じてLLMの推論処理への理解を深めたいというモチベーションです。 🤖 vLLMって? 📚 前提知識 Attention Is All You Need Paged Attention Continuous Batching 📦 vLLMの開発用インストール (Pythonコード開発のみ) Wheelのインストール リポジトリのクローン 起動確認 Pythonコードの改変 デバッガを使ったOSSのコードリーディングのススメ 🧩 vLLMのソフトウェアアーキテクチャ オンライン推論 : FastAPIサーバの立ち上げとEngineClientの生成 1. Engin
TL;DR; We are changing std::sort in LLVM’s libcxx. That’s a long story of what it took us to get there and all possible consequences, bugs you might encounter with examples from open source. We provide some benchmarks, perspective, why we did this in the first place and what it cost us with exciting ideas from Hyrum’s Law to reinforcement learning. All changes went into open source and thus I can
31st December 2024 A lot has happened in the world of Large Language Models over the course of 2024. Here’s a review of things we figured out about the field in the past twelve months, plus my attempt at identifying key themes and pivotal moments. This is a sequel to my review of 2023. In this article: The GPT-4 barrier was comprehensively broken Some of those GPT-4 models run on my laptop LLM pri
Update 1.69.1: The update addresses these issues. Update 1.69.2: The update addresses these issues. Downloads: Windows: x64 Arm64 | Mac: Universal Intel silicon | Linux: deb rpm tarball Arm snap Welcome to the June 2022 release of Visual Studio Code. There are many updates in this version that we hope you'll like, some of the key highlights include: 3-way merge editor - Resolve merge conflicts wit
Agents
Intelligent agents are considered by many to be the ultimate goal of AI. The classic book by Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach (Prentice Hall, 1995), defines the field of AI research as “the study and design of rational agents.” The unprecedented capabilities of foundation models have opened the door to agentic applications that were previously unimaginabl
Internet Engineering Task Force (IETF) K. Davis Request for Comments: 9562 Cisco Systems Obsoletes: 4122 B. Peabody Category: Standards Track Uncloud ISSN: 2070-1721 P. Leach University of Washington May 2024 Universally Unique IDentifiers (UUIDs) Abstract This specification defines UUIDs (Universally Unique IDentifiers) -- also known as GUIDs (Globally Unique IDentifiers) -- and a Uniform Resou
Since our launch of the Mojo programming language on May 2nd, more than 120K+ developers have signed up to use the Mojo Playground and 19K+ developers actively discuss Mojo on Discord and GitHub. Today, we’re excited to announce the next big step in Mojo’s evolution: Mojo is now available for local download – beginning with Linux systems, and adding Mac and Windows in coming releases. While the Mo
AST vs. Bytecode: Interpreters in the Age of Meta-Compilation
233 AST vs. Bytecode: Interpreters in the Age of Meta-Compilation OCTAVE LAROSE, University of Kent, UK SOPHIE KALEBA, University of Kent, UK HUMPHREY BURCHELL, University of Kent, UK STEFAN MARR, University of Kent, UK Thanks to partial evaluation and meta-tracing, it became practical to build language implementations that reach state-of-the-art peak performance by implementing only an interprete
How I developed a faster Ruby interpreter | Red Hat Developer
In this article, I will describe my efforts to implement a faster interpreter for CRuby, the Ruby language interpreter, using a dynamically specialized internal representation (IR). I believe this article will interest developers trying to improve the interpreter performance of dynamic programming languages (e.g., CPython developers). I will cover the following topics: Existing CRuby interpreter a
Masato Hagiwara Open in Recto Pad Google Colab Github Recto Pad TL;DR Recto is a 2D programming language that uses nested rectangles as its core syntax, encoding structure and recursion directly in space instead of a linear stream of text. Recto explores new ways to write, parse, and reason about code—and even natural language—spatially. Introduction Open in Recto Pad Virtually all the languages w
Update 1.80.1: The update addresses these issues. Update 1.80.2: The update addresses this security issue. Downloads: Windows: x64 Arm64 | Mac: Universal Intel silicon | Linux: deb rpm tarball Arm snap Welcome to the June 2023 release of Visual Studio Code. There are many updates in this version that we hope you'll like, some of the key highlights include: Accessibility improvements - Accessible V
Over the course of my Spring 2020 semester at Harvey Mudd College, I developed a self-hosting compiler entirely from scratch. This article walks through many interesting parts of the project. It’s laid out so you can just read from beginning to end, but if you’re more interested in a particular topic, feel free to jump there. Or, take a look at the project on GitHub. Table of contents What the pro
Firebase Studio lets you build full-stack AI apps with Gemini | Google Cloud Blog
Millions of developers use Firebase to engage their users, powering over 70 billion instances of apps every day, everywhere — from mobile devices and web browsers, to embedded platforms and agentic experiences. But full-stack development is evolving quickly, and the rise of generative AI has transformed not only how apps are built, but also what types of apps are possible. This drives greater comp
Patterns for Building LLM-based Systems & Products [ llm engineering production 🔥 ] · 66 min read Discussions on HackerNews, Twitter, and LinkedIn “There is a large class of problems that are easy to imagine and build demos for, but extremely hard to make products out of. For example, self-driving: It’s easy to demo a car self-driving around a block, but making it into a product takes a decade.”
Join a VS Code Dev Days event near you to learn about AI-assisted development in VS Code. Update 1.70.1: The update addresses these issues. Update 1.70.2: The update addresses these issues. Update 1.70.3: This update is only available for Windows 7 users and is the last release supporting Windows 7. Downloads: Windows: x64 Arm64 | Mac: Universal Intel silicon | Linux: deb rpm tarball Arm snap Welc
A Walk with LuaJIT
The following is a chronicle of implementing a general purpose zero-instrumentation BPF based profiler for LuaJIT. Some assumptions are made about what this entails and it may be helpful to read some of our other work in this area. One major change from prior efforts is that instead of working with the original Parca unwinder we are now working with the OpenTelemetry eBPF profiler. If you missed t
Tier 4 Support § Support for these targets is entirely experimental. If this target is provided by LLVM, LLVM may have the target as an experimental target, which means that you need to use Zig-provided binaries for the target to be available, or build LLVM from source with special configure flags. zig targets will display the target if it is available. This target may be considered deprecated by
Real-world gen AI use cases from the world's leading organizations | Google Cloud Blog
AI is here, AI is everywhere: Top companies, governments, researchers, and startups are already enhancing their work with Google's AI solutions. Published April 12, 2024; last updated October 9, 2025. A year and a half ago, during Google Cloud Next 24, we published this list for the first time. It numbered 101 entries. It felt like a lot at the time, and served as a showcase of how much momentum b
How a simple Linux kernel memory corruption bug can lead to complete system compromise
In this case, reallocating the object as one of those three types didn't seem to me like a nice way forward (although it should be possible to exploit this somehow with some effort, e.g. by using count.counter to corrupt the buf field of seq_file). Also, some systems might be using the slab_nomerge kernel command line flag, which disables this merging behavior. Another approach that I didn't look
I want to write a post about Pitchfork, explaining where it comes from, why it is like it is, and how I see its future. But before I can get to that, I think I need to share my mental model on a few things, in this case, Ruby’s GVL. For quite a long time, it has been said that Rails applications are mostly IO-bound, hence Ruby’s GVL isn’t that big of a deal and that has influenced the design of so
How to turn Claude Code into a domain specific coding agent
Authored by: Aliyan Ishfaq Coding agents are great at writing code that uses popular libraries on which LLMs have been heavily trained on. But point them to a custom library, a new version of a library, an internal API, or a niche framework – and they’re not so great. That’s a problem for teams working with domain specific libraries or enterprise code. As developers of libraries (LangGraph, LangCh
By Jun He, Natallia Dzenisenka, Praneeth Yenugutala, Yingyi Zhang, and Anjali Norwood TL;DRWe are thrilled to announce that the Maestro source code is now open to the public! Please visit the Maestro GitHub repository to get started. If you find it useful, please give us a star. What is MaestroMaestro is a horizontally scalable workflow orchestrator designed to manage large-scale Data/ML workflows
agent loop �� �p�� You are Manus, an AI agent created by the Manus team. You excel at the following tasks: 1. Information gathering, fact-checking, and documentation 2. Data processing, analysis, and visualization 3. Writing multi-chapter articles and in-depth research reports 4. Creating websites, applications, and tools 5. Using programming to solve various problems beyond development 6. Variou
Tier 4 Support § Support for these targets is entirely experimental. If this target is provided by LLVM, LLVM may have the target as an experimental target, which means that you need to use Zig-provided binaries for the target to be available, or build LLVM from source with special configure flags. zig targets will display the target if it is available. This target may be considered deprecated by
インターネットの世界にニュースが溢れる昨今、満足度が高いものを的確に読みたいという方も多いかと思います。そのためには、見るニュースをどれにするか判断することが必要になります。そこで、ニュース全体の主旨を短い文章で表す要約の価値が高まっています。 自然言語処理における要約は、大きく２つに分けられます。それは、抽出型と抽象型です。抽出型は、文章の中から重要な文を抜き出すことで要約を作ります。要約として選ばれた文は元の文章にあるものなので、方向性が大きく異ることや誤字脱字がうまれる可能性は低いです。しかし、要約として選ばれた文のそれぞれは関係があるわけではないので、流暢な要約にならないことも多いです。それに対して、抽象型は人間が作るように要約としての文章の流暢さを考慮しながら作ります。本来人間がほしい要約はこちらになりますが、抽出型に比べると難易度が上がり、全く意味がわからない文章になる可能性も
Accelerate Python code 100x by import taichi as ti | Taichi Docs
Python has become the most popular language in many rapidly evolving sectors, such as deep learning and data sciences. Yet its easy readability comes at the cost of performance. Of course, we all complain about program performance from time to time, and Python should certainly not take all the blame. Still, it's fair to say that Python's nature as an interpreted language does not help, especially
Deep Research再現実装をDeep Research以上に詳しく検証してみた - AKARI Tech Blog
はじめに こんばんは！ 今週のAKARI Tech Blogは、DX Solution 事業本部 Dev の許が担当いたします。 先日OpenAIが「Deep Research」を公開し、その驚異的な文献調査能力が話題となりましたね！ 皆様使っていますでしょうか。 これまでひいこら言いながらインターネット検索していた時代と比べると、「Deep Research お願いします！」で、それなりの分析レポートが出てくることに隔世の感を感じますね。 これだけ性能の良いものが出てきた以上、仕組みが気になるところ。できることなら、自分たちでも再現実装してみたい！ しかし例によってOpenAIは実装をオープンにはしてくれない……。 そこで登場するのが、Deep ResearchのOSS再現プロジェクトたち！ まずは Deep ResearchにOpenな再現実装について聞いてみましょうか。 ChatGP
Python behind the scenes #13: the GIL and its effects on Python multithreading
As you probably know, the GIL stands for the Global Interpreter Lock, and its job is to make the CPython interpreter thread-safe. The GIL allows only one OS thread to execute Python bytecode at any given time, and the consequence of this is that it's not possible to speed up CPU-intensive Python code by distributing the work among multiple threads. This is, however, not the only negative effect of
arXiv:2506.12115v1 [cs.CL] 13 Jun 2025 Eliciting Reasoning in Language Models with Cognitive Tools Brown Ebouky IBM Research - Zurich ETH Zurich Brown.Ebouky@ibm.com Andrea Bartezzaghi IBM Research - Zurich abt@zurich.ibm.com Mattia Rigotti IBM Research - Zurich mrg@zurich.ibm.com Abstract The recent advent of reasoning models like OpenAI’s o1 was met with excited spec- ulation by the AI community
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
新入社員のみんな、「ChatGPT×Python」で鬼にならないか？｜ピーナッツ
プロと読み解くRuby 3.4 NEWS - STORES Product Blog
The Prompt Engineering Playbook for Programmers
GPT in 60 Lines of NumPy | Jay Mody
DeepSeek-R1 1.58bを試す/ついに実用的なBitNetが!?｜shi3z
📖 vLLMのコードを読んでみよう - ENGINEERING BLOG ドコモ開発者ブログ
Changing std::sort at Google’s Scale and Beyond
Things we learned about LLMs in 2024
June 2022 (version 1.69)
RFC 9562: Universally Unique IDentifiers (UUIDs)
Modular: Mojo🔥 - It’s finally here!
Recto — a truly 2D language
June 2023 (version 1.80)
Kalyn: a self-hosting compiler for x86-64
Patterns for Building LLM-based Systems & Products
July 2022 (version 1.70)
0.8.0 Release Notes ⚡ The Zig Programming Language
So You Want To Remove The GVL?
Maestro: Netflix’s Workflow Orchestrator
Manus tools and prompts
0.10.0 Release Notes ⚡ The Zig Programming Language
PytorchのTransformersのT5を使って要約モデルを作る - 見習いデータサイエンティストの隠れ家
Eliciting Reasoning in Language Models with Cognitive Tools