Fine tune a 70B language model at home
We’re releasing an open source system, based on FSDP and QLoRA, that can train a 70b model on two 24GB GPUs.
Posts that commenters keep linking to, ranked by how many different people have cited them in Hacker News comments. 4,918 posts.
We’re releasing an open source system, based on FSDP and QLoRA, that can train a 70b model on two 24GB GPUs.
Apple today debuted the all-new iPhone Air, the thinnest iPhone ever made, with pro performance.
Updated: March 4, 2025
Kimi K3 matches Fable 5 on SWE tasks (92.4% vs 92.6%) at up to 50x lower cost on Fireworks. See where each model wins and how routing beats both.
Apple announced the latest 14- and 16-inch MacBook Pro with the all-new M5 Pro and M5 Max.
Choosing good colors for your charts is hard. This article tries to make it easier.
On a recent trip to Cuba I brought with me a smartphone and hoped to get Internet access either via WiFi or 3G. I managed that (at a price) but also saw for myself how Cubans get access to an alternate Internet delivered by sneakernet.
Porting Linux to Apple Silicon
A lot of the advice we give startups is tactical; meant to be helpful on a day to day or week to week basis. But some advice is more fundamental. We’ve collected here what we at YC consider the most important, most transformative advice for startups. Whether common sense or counter-intuitive, the guidance below will…
How Go 2 proposals will be evaluated, selected, and shipped.
Ladies and Gentlemen, I'm pleased to announce that Sublime Text 2 has graduated out of beta, with 2.0 now available for your text editing pleasure.
Broadening usage of frontier AI
Launching CockroachDB 1.0, a production-ready release with distributed SQL and multi-active availability in two tiers: CockroachDB Core and CockroachDB Self-hosted.
Here at Microsoft we have teams of all shapes and sizes, and many of them are already using Git or are moving that way. For the most part, the Git client
.NET 5.0 includes many enhancements, including single file applications, more capable JSON APIs, new versions of C
I’m delighted to announce that Sam Altman has agreed to become president of Y Combinator starting next batch. I’ll continue to do office hours with startups, but Sam is going to lead YC. Why the change? Because YC needs to grow, and I’m not the best person to grow it. Sam is what YC needs at this stage in its…
In this essay, Carson Gross and Alex Petros discuss htmx's future direction and philosophy. They explain how the project aims to emulate jQuery's success through API stability, minimal feature additions, and quarterly releases while continuing to promote hypermedia-driven development and support the broader web…
The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.
In this article, Lago's cofounder Raffi gives an overview of the technical challenges we faced while implementing a hybrid pricing model at Qonto, and what we learned during this journey.
Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
An indirect prompt injection in an implementation blog can manipulate Antigravity to invoke a malicious browser subagent in order to steal credentials and sensitive code from a user’s IDE.
We’re introducing new Gemini models, including Gemini 3.6 Flash, 3.5 Flash-Lite and 3.5 Flash Cyber.
From the Zed Blog: Zed is now the world's fastest AI code editor. What's more, Zed's new AI features are also open-source, just like the whole editor.
The most powerful AI platform for enterprises. Customize, fine-tune, and deploy AI assistants, autonomous agents, and multimodal AI with open models.
This message was just sent to Team XBOX employees globally. Team, We are beginning the most significant restructure in XBOX history. After careful
Documentation and guides from the team at Fly.io.