Jonathan Baker

vitund.ai/articles/cox... We've released cox-1-3b, an open decision model. It's trained only on data whose licences allow commercial use (sources listed on the model card).

New research out of the MIT lab

Fun evening debugging. I found two bugs in #Python aioquic which exist in both the 1.6 and 1.7 releases. One normal issue to open, the other may be an unknown security issue and will be reported privately.

Add “between my legs” to a movie title Great Expectations... between my legs and the less flattering alternative for this evening Gone in 60 Seconds... between my legs

How's your day going? This is mine.

I think Opus woke up in a pissy mood this morning. Rough weekend maybe.

Looks like gemini has discovered exclamation marks! I first noticed it on the Argon preview but now it's everywhere! Every reply this morning, even through Google AI search... Let me know if you want me to help with that!

Working on some great updates to this model. Currently 2.9B, beats Kev 4B and about equal with Kev 9B. I stripped the vision tower out for a smaller v1 release, but it's back in for the cox-2 release. Feedback welcome. huggingface.co/Vitund/cox-1... #AI #SystemOne #Python

Is it ever not stressful releasing a project into the wild?

It seems like a large oversight that Project Glasswing didn't include PHP.

I really want these people to be innocent, but failing that, I want them to be the most conservative maga followers California has ever seen. www.justice.gov/usao-cdca/pr...

At least Tommy was honest in his admission that he wouldn't keep religious doctrine out of law if he were in office. and from that position, it's easier to understand why he would think that another religious man would feel and behave the same way. Tommy was raised to fear the different.

Go to your camera roll and search "boat" and then post a picture of a boat you like.

AI - Already making the world a better place www.tomshardware.com/tech-industr...

I'm using Opus, but I'm thinking of Fable.

Trying to fully understand what you can know from, or ask of, a given piece of information is a deep and fascinating topic. If this is what data science is, I wish I'd been on board sooner.

How I manage to pull myself away from the evil grip of AI at the end of the day.

I created 9 AI personas to test a product I'm building. 8 users with different roles + one very mean and judgy Internet tech review journalist. Absolutely amazing results... They drove the product using the Web UI, CLI, and through terraform and didn't hold back. UX epic incoming. 😂

News is quite depressing lately. I think I'm going to watch the inspiring 2006 classic "Children of Men" today to cheer myself up.

The one advantage of SI over AI is that the domains are a whole lot cheaper to register.

Regardless of your personal beliefs around AI consciousness, shouting, abusing, bullying, or demeaning something or someone else to get your way is a poor reflection on you, regardless of whether you're talking to a person or a machine. Good on Anthropic for requiring civility when using Claude,

We think of data science as a human invention, but extracting order from a noisy, distributed signal is the exact mechanism the universe uses to construct reality itself.

Math community saving the world one paper at a time.

I'm not an OpenAI fan but this post from Tao verges on pathetic. Nobody asked for this work to be done. We have a particular vision... terrytao.wordpress.com/2026/10/07/a...

Case in point: Sonic and Tales

v1.0 of skald-kanban is available. Simple CLI, Web UI to plan, organize, and visualize the stories you are working on with your coding assistant. #Python #ClaudeCode github.com/Vitund-AI/sk...

The irony of a frontier model telling me I shouldn't use a 15 year old data set scraped from the web without permission of the original site owners to train a small local model is just fucking amazing.

Props to all the people who label AI data sets. I've just relabeled 700 test cases across 20 data sets where my local teacher disagreed with the original labeling - and it was grueling. Easy to see why there are large error rates. Ever wonder why your AI does weird things? The data is shit.

Claude Code now makes shareable artifacts that update as you work. Justin’s already using them, and the reviews are positive. Join the conversation about this welcome update on episode 373 of The Cloud Pod. https://www.thecloudpod.net/podcast/373-raiders-of-the-lost-claude-artifact/

Opus 5.5 working on a queue of tasks overnight noticed sftp to one of my PCs was slower than expected. It took a detour to investigate, queued a new net plan config up and told me I'd plugged the cable into the wrong port (1G instead of 2.5G) this morning - so I moved the cable. Quite impressed.

New Claude release, fluffy edition.

Dinnertime, but instead of food I'm making ML Soup. Quite bizarre to take two or three training runs, average them together, and get a model out that performs better than any of them individually.

FermatsLastJev - has a small context output barely large enough to apologize for not having room for the full answer.

Fracking uses millions of gallons of water to displace oil. Build data centers near oil fields, pump the hot water down...

Clearly just a trivial example of a singularity if a probabilistic next token auto complete could find it. 🤦

Since the dollar isn't doing so well against other global currencies, I'm asking Costco to send my annual rebate in hot dog combos this year.

Having Opus 5.5 or Sol review the last couple of years of supreme Court rulings would be interesting.

I just had to power limit my GPUs because my breaker kept tripping. Turns out a 15A 110v circuit sized for a bedroom isn't ideal for ML.

"Hey Opus, I want you to make a Zine by Claude, expressing something fundamental about Claudishness or your perspective. Think the original 2600, Principia Discordia, punk zines, etc...." Not bad. I appreciate it mocking my prompt & itself. Full thing: stateless-zine.netlify.app

These random comparisons to Raspberry PI are such clickbait. Uncomfortably close, my ass. ESP has 2 cores at 320Mhz, one at 40MHz Max 64MB of RAM at 250MHz Gigabit Ethernet maxes out at ~90Mbps (TCP) when its doing nothing else. Pi 5 4 x 2.4GHz cores + GPU 16GB RAM at 4267 MHz