Preston Rozwood (@Prozwood) profile picture

Machine learning enthusiast motivated by representation learning! All opinions are my own and do not reflect those of my employer

Joined September 2015
Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)Photo by Preston Rozwood (@Prozwood)
Preston Rozwood retweeted
Discord is adding support for rendering LaTeX math formulas in Markdown files
78
405
149
7,106
178,109
Preston Rozwood retweeted
ICLR reviews are really instructive. Now I'm thinking, what a shame I didn't run multiple seeds for every decision I've made in life to make sure the results were statistically convincing.
22
56
4
980
58,072
Preston Rozwood retweeted
I really enjoyed taking part in this discussion, especially at this moment in time. Thanks Lex, Alex, and Ankur for the great conversation!
This is an incredible moment in this history of mathematics, computer science, physics & science with some of the hardest problems in human history being solved by AI over a period of weeks. Depending on your perspective, it's either a crisis, a triumph, or both. I don't think anyone knows for sure. We're all trying to make sense of it. I oscillate between fear/dread and optimism/excitement on a daily basis. The rate of progress is insane, including today's announcement from OpenAI of more problems solved. To help make sense of all of it, here's my conversation with three mathematicians about this moment. Each of them is an expert in the field of each of the biggest problems solved: Navier-Stokes, quasi-Riemann hypothesis, the unique games conjecture and complexity theory more generally. Ankur Moitra is a mathematician and theoretical computer scientist at MIT, visiting the Institute for Advanced Study in Princeton. Steven Brunton (@eigensteve) is a mathematician and fluid dynamics expert at UW. Alex Kontorovich (@AlexKontorovich) is a mathematician at Rutgers specializing in number theory.
15
36
3
425
18,983
Preston Rozwood retweeted
Two critical vulnerabilities have been identified in PHP, a widely used programming language that powers websites and web applications. These vulnerabilities (CVE-2026-109618 and CVE-2026-109619) could allow malicious cyber actors to remotely execute code in PHP core, potentially compromising sensitive data and networks. The FBI urges organizations to take the following steps now: ➡️ If an immediate update is not feasible it is recommended that organizations take the system offline until it can be fully mitigated. ➡️ Immediately update affected PHP installations following the guidance on php.net. Inventory where PHP runs in your environment, prioritizing internet-facing systems and vendor products that bundle PHP. ➡️ Monitor web servers for suspicious activity, including PHP processes unexpectedly launching command shells or downloading files. Ensure appropriate process-monitoring capabilities are in place to help identify potential exploitation. ➡️ If you find signs of compromise, isolate affected systems, preserve your web server and PHP logs, and report it to your local FBI field office or ic3.gov. Those reports help us identify the actors involved and warn other organizations before they become victims. For the latest information, follow PHP’s announcement on php.net, CERT-EU Security Advisory 2026-018, and CISA’s Known Exploited Vulnerabilities catalog.
448
5,385
1,293
25,281
3,493,598
Preston Rozwood retweeted
If you are wondering whether your website is affected, we have packaged a Codex skill to help check. Point your Codex at this repo to guide you through it: github.com/openai/php2shell-…
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites. Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors. Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact. This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare. openai.com/index/coordinated…
353
1,209
234
9,369
1,536,562
Preston Rozwood retweeted
OpenAI models helped researchers—operating as part of our Patch the Planet initiative—discover critical security flaws in PHP, an open-source programming language used to run an estimated 70% of websites. Over the past two months, we led a coalition of maintainers, researchers, and security providers to prepare fixes and protections ahead of disclosure. This duration was necessary because the fix was quite intricate and required a major change to the PHP engine. Every site using PHP can now safely upgrade their version to stay secure, in addition to applying mitigations from various security vendors. Due to the severity of these flaws and PHP’s widespread use, we consider this bug to be one of the most impactful vulnerabilities in recent memory. It also marks a new phase, where AI can find security bugs with internet-scale impact. This is an example of the Defender’s Window in action: to our knowledge, frontier AI was able to find this vulnerability ahead of attackers, giving defenders time to prepare. openai.com/index/coordinated…
624
2,860
1,528
14,406
6,010,136
Preston Rozwood retweeted
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
640
894
642
8,205
3,498,090
‼️ BREAKING: The private keys of 8 certificates from Indian certificate authority eMudhra were compromised, Mozilla says. Firefox blocked them within 3 hours. Each key should be a 116-digit number, but one was just 4. Anyone can guess them and forge HTTPS certs for any website.
617
3,695
1,718
30,084
5,830,881
Preston Rozwood retweeted
We present Tropical Reinforcement Learning. Standard RL for LLMs cannot learn compositions effectively, for a fundamental reason: its algebra is not optimal. It sums the probabilities of every successful answer, measuring how often a model succeeds, not which solution succeeds.
22
31
3
405
43,145
Preston Rozwood retweeted
New Paper: Human-like Autonomy Emerges from Self-Play and a Pinch of Human Data. We trained self-play RL on 60 years of simulation on 1 GPU in ~15 hours. Regularizing with 30 minutes of demonstration data produces much more human-like driving policies!
8
40
8
350
79,182
Compressing KV cache sounds like a storage problem, but it is really a geometry problem. Shrinking 16-bit floats down to 4, 3, or even 2 bits is noisy and can destroy the geometry of your vectors. A thread on TurboQuant 🧵👇
1
37
One really nice property of TurboQuant is that it is not just a KV-cache trick. The same method applies to RAG pipelines, vector databases, and nearest-neighbor search. In other words, any time when the real target is the geometry and not the coordinates.
1
21
Preston Rozwood retweeted
I Wrote a New Book!!! Optimization: A Bootcamp for Machine Learning, Inverse Problems, and Control Pre-Order Now (July 31) amazon.com/Optimization-Boot… Coming Soon: * Free PDF on website * YouTube Videos for entire book * Python code on GitHub
62
377
22
2,813
193,088
With one .yallm file you can define, run, test, and version: - prompt templates - model settings - typed inputs - output schemas - evals/assertions Think of it like GitHub Actions / Makefiles for prompts. Repo: github.com/Pdbz199/YALLM 👇🧵 2/3
1
1
4
62
If you're building with LLMs, what would you like to see from YALLM? 🧵 3/3
4
31