rewinfrey

2.6K posts

rewinfrey banner
rewinfrey

rewinfrey

@bashrw

Software engineer focused on dev tools. Previously @nuanced_dev, @GitHub. MSc Software Engineering @UniofOxford.

Se unió Mayıs 2010
2.5K Siguiendo690 Seguidores
rewinfrey
rewinfrey@bashrw·
I've been fine tuning small language models for text classification using T4 GPUs on Modal. This introduced me to BERT (bidirectional encoder representations from transformers), sentence transformers, and SetFit. I wrote up my notes as a post: rickwinfrey.com/writings/small…
English
0
0
0
52
Drew Wilson
Drew Wilson@drewwilson·
Opaque is coming soon :D It’s the system I built that powers @OpacityHQ, and also powers any app deployed via @OpacityHQ 😎 Sneak peek:
English
2
1
25
2.4K
rewinfrey retuiteado
Bo Wang
Bo Wang@BoWang87·
Prof. Donald Knuth opened his new paper with "Shock! Shock!" Claude Opus 4.6 had just solved an open problem he'd been working on for weeks — a graph decomposition conjecture from The Art of Computer Programming. He named the paper "Claude's Cycles." 31 explorations. ~1 hour. Knuth read the output, wrote the formal proof, and closed with: "It seems I'll have to revise my opinions about generative AI one of these days." The man who wrote the bible of computer science just said that. In a paper named after an AI. Paper: cs.stanford.edu/~knuth/papers/…
Bo Wang tweet media
English
155
1.9K
9.1K
1.2M
Drew Wilson
Drew Wilson@drewwilson·
🚀 Tomorrow is the big launch day for @CorticalLabs! The world's first commercially available biological computer! Neurons LIVING on a silicon chip. Deploy your code directly to brain cells. Yes, actually. It's not "Artificial Intelligence", it's "Actual Intelligence". I'm stoked to play a small part in this rad company. And I wanted to share the story with you all of how I got to work on this super revolutionary product.. and give you a look at how it all works. 🎥 Check it out!
English
4
3
33
15.2K
rewinfrey
rewinfrey@bashrw·
Sharing a 5-part series on content-defined chunking (CDC) algorithms as background research for my master's thesis. It's a tour of the algorithms, a deep dive into FastCDC, and explores costs for building deduplication pipelines at scale. rickwinfrey.com/writings/conte…
English
0
0
0
46
rewinfrey retuiteado
Sakana AI
Sakana AI@SakanaAILabs·
We’re excited to introduce Doc-to-LoRA and Text-to-LoRA, two related research exploring how to make LLM customization faster and more accessible. pub.sakana.ai/doc-to-lora/ By training a Hypernetwork to generate LoRA adapters on the fly, these methods allow models to instantly internalize new information or adapt to new tasks. Biological systems naturally rely on two key cognitive abilities: durable long-term memory to store facts, and rapid adaptation to handle new tasks given limited sensory cues. While modern LLMs are highly capable, they still lack this flexibility. Traditionally, adding long-term memory or adapting an LLM to a specific downstream task requires an expensive and time-consuming model update, such as fine-tuning or context distillation, or relies on memory-intensive long prompts. To bypass these limitations, our work focuses on the concept of cost amortization. We pay the meta-training cost once to train a hypernetwork capable of producing tasks or document specific LoRAs on demand. This turns what used to be a heavy engineering pipeline into a single, inexpensive forward pass. Instead of performing per-task optimization, the hypernetwork meta-learns update rules to instantly modify an LLM given a new task description or a long document. In our experiments, Text-to-LoRA successfully specializes models to unseen tasks using just a natural language description. Building on this, Doc-to-LoRA is able to internalize factual documents. On a needle-in-a-haystack task, Doc-to-LoRA achieves near-perfect accuracy on instances five times longer than the base model's context window. It can even generalize to transfer visual information from a vision-language model into a text-only LLM, allowing it to classify images purely through internalized weights. Importantly, both methods run with sub-second latency, enabling rapid experimentation while avoiding the overhead of traditional model updates. This approach is a step towards lowering the technical barriers of model customization, allowing end-users to specialize foundation models via simple text inputs. We have released our code and papers for the community to explore. Doc-to-LoRA Paper: arxiv.org/abs/2602.15902 Code: github.com/SakanaAI/Doc-t… Text-to-LoRA Paper: arxiv.org/abs/2506.06105 Code: github.com/SakanaAI/Text-…
GIF
English
74
354
2.2K
596.2K
rewinfrey
rewinfrey@bashrw·
I appreciate taking clear ownership, clear comms, and no flinching here. But this absolutely reeks. Two points: * There are more than two options here. * “To those leaving” should be “To those I’m firing” Wishing the 4,000 humans impacted better opportunities and good luck ❤️
jack@jack

we're making @blocks smaller today. here's my note to the company. #### today we're making one of the hardest decisions in the history of our company: we're reducing our organization by nearly half, from over 10,000 people to just under 6,000. that means over 4,000 of you are being asked to leave or entering into consultation. i'll be straight about what's happening, why, and what it means for everyone. first off, if you're one of the people affected, you'll receive your salary for 20 weeks + 1 week per year of tenure, equity vested through the end of may, 6 months of health care, your corporate devices, and $5,000 to put toward whatever you need to help you in this transition (if you’re outside the U.S. you’ll receive similar support but exact details are going to vary based on local requirements). i want you to know that before anything else. everyone will be notified today, whether you're being asked to leave, entering consultation, or asked to stay. we're not making this decision because we're in trouble. our business is strong. gross profit continues to grow, we continue to serve more and more customers, and profitability is improving. but something has changed. we're already seeing that the intelligence tools we’re creating and using, paired with smaller and flatter teams, are enabling a new way of working which fundamentally changes what it means to build and run a company. and that's accelerating rapidly. i had two options: cut gradually over months or years as this shift plays out, or be honest about where we are and act on it now. i chose the latter. repeated rounds of cuts are destructive to morale, to focus, and to the trust that customers and shareholders place in our ability to lead. i'd rather take a hard, clear action now and build from a position we believe in than manage a slow reduction of people toward the same outcome. a smaller company also gives us the space to grow our business the right way, on our own terms, instead of constantly reacting to market pressures. a decision at this scale carries risk. but so does standing still. we've done a full review to determine the roles and people we require to reliably grow the business from here, and we've pressure-tested those decisions from multiple angles. i accept that we may have gotten some of them wrong, and we've built in flexibility to account for that, and do the right thing for our customers. we're not going to just disappear people from slack and email and pretend they were never here. communication channels will stay open through thursday evening (pacific) so everyone can say goodbye properly, and share whatever you wish. i'll also be hosting a live video session to thank everyone at 3:35pm pacific. i know doing it this way might feel awkward. i'd rather it feel awkward and human than efficient and cold. to those of you leaving…i’m grateful for you, and i’m sorry to put you through this. you built what this company is today. that's a fact that i'll honor forever. this decision is not a reflection of what you contributed. you will be a great contributor to any organization going forward. to those staying…i made this decision, and i'll own it. what i'm asking of you is to build with me. we're going to build this company with intelligence at the core of everything we do. how we work, how we create, how we serve our customers. our customers will feel this shift too, and we're going to help them navigate it: towards a future where they can build their own features directly, composed of our capabilities and served through our interfaces. that's what i'm focused on now. expect a note from me tomorrow. jack

English
0
0
0
67
rewinfrey retuiteado
Stefano Ermon
Stefano Ermon@StefanoErmon·
Mercury 2 is live 🚀🚀 The world’s first reasoning diffusion LLM, delivering 5x faster performance than leading speed-optimized LLMs. Watching the team turn years of research into a real product never gets old, and I’m incredibly proud of what we’ve built. We’re just getting started on what diffusion can do for language.
English
320
587
4.2K
978.5K
rewinfrey
rewinfrey@bashrw·
Great read 👏 “authentication and authorization are complicated, situated beasts, impossible to separate from the UX and architectural concerns of the systems that incorporate them.” After doing major surgery of GitHub’s authn/z for code search, I couldn’t agree more!!
Blaine Cook@blaine

@geoffreylitt This definitely doesn't solve your auto-Matt Levine problem, but hopefully (!) explains OAuth: leaflet.pub/p/did:plc:3vdr…

English
0
0
1
132
rewinfrey
rewinfrey@bashrw·
@kwuchu @kenwheeler That’s cool, thanks for explaining. It sounds a bit like a skill? That’s super useful. Is there some way to share that internally? Or do you package it as a plugin for distribution at Stripe?
English
1
0
0
31
Iheanyi Ekechukwu
Iheanyi Ekechukwu@kwuchu·
@bashrw @kenwheeler Just a regular markdown file with front matter with a lot of context about Stripe-specific patterns. We have internal libraries/frameworks for pretty much everything across the stack, so loading an agent with that documentation has been really helpful.
English
1
0
1
194
Iheanyi Ekechukwu
Iheanyi Ekechukwu@kwuchu·
Got asked to show my teammates my development workflow using AI, so I wrote a 5 page document with pretty much a brain dump of how I think about it and use it. Can't wait to present it!
English
9
0
42
4.8K
rewinfrey
rewinfrey@bashrw·
@kwuchu @kenwheeler That's cool! Is it using claude code under the hood for inference? Or does it make API requests?
English
1
0
0
37
Iheanyi Ekechukwu
Iheanyi Ekechukwu@kwuchu·
@kenwheeler I've gone so far as to writing custom agents at work tailored to what I need to get done, lol.
English
1
0
3
410
Enes Akar
Enes Akar@enesakar·
what do you do while claude/cursor is writing your code?
English
15
0
6
1.9K
rewinfrey
rewinfrey@bashrw·
I’m currently working on a master’s thesis to explore these latter questions with an end goal to produce a cost model to understand the tradeoffs for 3 deduplication strategies and their impacts on storage, network pressure, and system resource utilization. It is quite fun!
English
0
0
0
33
rewinfrey
rewinfrey@bashrw·
What version control primitives are necessary for high-velocity, large monorepo, self-driving codebases? I think Sapling, from Meta, has some answers here. But how can we achieve deduplication tailored to source code at massive scale? While also optimizing user-facing latency?
English
1
0
0
46
rewinfrey
rewinfrey@bashrw·
Great write up and experiment! > All agents have their own copy of the repo, but most files and artifacts are identical; could adding simple copy-on-write and deduplication features, …, bring similar easy wins to a typically “single-user”system without building separate infra?
Cursor@cursor_ai

Read more: cursor.com/blog/self-driv…

English
1
0
1
88