Something went wrong, but don’t fret — let’s give it another shot. Some privacy related extensions may cause issues on x.com. Please disable them and try again.
Cohere Transcribe is an open-source automatic speech recognition model that achieves state-of-the-art accuracy with a low word error rate of 5.42%. It is available for download and can be used for real-world transcription tasks across 14 languages.
Anthropic accidentally exposed Claude Code's source code and internal features, including anti-distillation mechanisms and a companion system. The leak reveals product roadmap details that competitors can now see and react to.
StepSecurity identified malicious versions of the axios HTTP client library published to npm, axios@1.14.1 and axios@0.30.4, which inject a remote access trojan (RAT) dropper. Developers who installed these versions should rotate all secrets and credentials, check network logs, and downgrade to safe versions, and StepSecurity provides end-to-end npm supply chain security across three pillars: ...
SolveSpace has a web version that runs in the browser with some speed penalty and bugs. It's experimental and can be hosted locally like static web content.
forkrun is a self-tuning, drop-in replacement for GNU Parallel that accelerates shell-based data preparation by 50×–400× on modern CPUs and scales linearly on NUMA architectures. It operates completely differently, using a pipeline with four stages to preserve physical locality and minimize IPC overhead.
AI models will write good code due to economic incentives, as good code is cheaper to generate and maintain. Economic forces will drive AI models to generate simpler, good code as it will be cheaper overall.
The user created a hook that caused a fork bomb, bricking their computer due to excessive memory usage, but luckily the computer's caching saved it from further damage. The user had been experimenting with Claude, a large language model, and built various tools to help with their work, but the experience was imperfect and costly, resulting in a $3800 API bill.
Language models use a key-value cache to store and recall conversation history, but this cache has physical and memory limitations, leading to context degradation and the need for external memory systems. Researchers are exploring ways for models to learn and manage their own memory, potentially leading to more efficient and effective language processing.
Ollama now runs faster on Apple silicon with MLX framework, leveraging GPU Neural Accelerators for speedup. Ollama 0.19 sees 1851 token/s prefill and 134 token/s decode with improved memory efficiency and model accuracy.
GitHub and Lettermatic collaborated to create Monaspace, a superfamily of five interchangeable typefaces for code editors. Monaspace offers high personalization and accessibility features, including Texture Healing, which improves legibility in monospace typefaces.
Combinators are functions that refer to their arguments without modifying them. They are often represented by bird names in APL, such as Kestrel and Cardinal.
Researchers analyzed audio recordings from the Milgram experiment and found that obedient participants broke the rules of the study most of the time, often ignoring the scientific procedure. This suggests that the laboratory environment was one of unauthorized violence, rather than a legitimate scientific study.
Oracle laid off 20,000-30,000 employees, roughly 18% of its workforce, in a single email with no advance notice. The cuts are tied to Oracle's aggressive expansion into AI infrastructure, freeing up $8-10 billion in cash flow.
Researchers from Google Quantum AI and other institutions have published a whitepaper warning of the potential risks of quantum computers to blockchain security, particularly in cryptocurrencies that rely on Elliptic Curve Discrete Logarithm Problem (ECDLP) cryptography. They estimate that a quantum computer with fewer than half a million physical qubits could break ECDLP-based cryptography ...
Ruby Central's Open Source Committee faced a crisis in September 2025 known as the "RubyGems Fracture" due to a poorly communicated process to offboard two engineers, André Arko and Samuel Giddens, from the RubyGems.org service. The incident led to the removal of paid contributors and highlighted the need for better policies, procedures, and communication in managing access and offboarding in ...
You agree to these terms by using Copilot, which includes rules for using the service, protecting others, and respecting Microsoft's rights. By using Copilot, you grant Microsoft permission to use your content and agree to their terms, including the Microsoft Services Agreement and the Microsoft Privacy Statement.
A weekly-or-more deep dive on technical topics surrounding Mars exploration, from microbiology to rocket science and everywhere in between. By subscribing, you agree Substack's Terms of Use, and acknowledge its Information Collection Notice and Privacy Policy. This fact makes people mad, because there are all kinds of other obstacles that are fun to solve (orbital refueling, landing heavy ...
Scotty is a new SSH task runner that lets you define deploy scripts and run them from your terminal with real-time output. It supports both Blade and plain bash formats, and offers features like pause and pretend modes.
The author discussed major works published by authors over 85, citing Sophocles' Philoctetes and Goethe's Faust, but found few notable works by authors over 80. A list of notable works by authors over 65 includes V. Hugo's Ninety-three, T. Mann's Doctor Faustus, and J. Saramago's Blindness.
Users of Claude Code are experiencing high token usage and early quota exhaustion due to a combination of factors including reduced quotas during peak hours, bugs, and unclear usage limits. Anthropic is actively investigating the issue and users are negotiating with providers over acceptable pricing and usage models for AI development.
The White House released an app on the App Store and Google Play. They posted a blog about it. "Unparalleled access to the Trump Administration." It took a few minutes to pull the APKs with ADB, and threw them into JADX. It's a React Native app built with Expo (SDK 54), running on the Hermes JavaScript engine. The backend is WordPress with a custom REST API. The app was built by ...
Loreline is a narrative language that uses Haxe for transpilation to multiple platforms, allowing it to run everywhere. It has a lexer, parser, and interpreter that transform source text into executable code, with a focus on performance and simplicity.
Google's browser warns of a "Suspicious Download" for yt-dlp, a tool to download files from Google's servers. This is seen as browser monopoly abuse and misleading people, as Chrome also downloads files from various servers.
I was recently diagnosed with Relapsing-Remitting Multiple Sclerosis after experiencing numbness and tingling in my arm and torso. I'm starting treatment to manage symptoms and slow disease progression.
The author reminisces about childhood games like Pixie Hollow and Bratz: Rock Angelz, now lost due to cultural erasure of feminine games. FEMICOM Museum founder Rachel Weil aims to preserve and celebrate games targeted toward young girls.
A CLAUDE.md file reduces Claude output verbosity by ~63% without code changes, targeting sycophancy, verbosity, and formatting noise. It's most beneficial for high-output use cases, and users can customize it to target specific failure modes and compose multiple files for different project types.
Researchers at New York University and City, University of London are conducting a study on AI's impact on software development. They are seeking US-based developers to share their experiences and perspectives on using AI tools in their day-to-day work via a 45-60 minute Zoom interview.
0
2026-03-31 07:39:36 UTC
ycombinator.com
Llama
RamAIn builds AI agents for enterprise work, automating repetitive tasks 10x faster and more reliably than humans. We're hiring a Founding AI/ML researcher to design and deploy agents that reason, plan, and execute complex workflows autonomously.
Forth VM and statically-typed s-expression compiler, written in C++ and Scryer Prolog - no382001/forth-vm