dulakian

Opus 5 is not consistent. When it is better than 4.8, it's much better. But when it's worse than 4.8, it's much worse. I am not sure what they are doing, but I prefer the more consistent performance of Opus 4.8. I am seriously considering dropping my Anthropic subscription and upgrading my OpenAI subscription. Fable guardrails trigger on the most innocuous references. I am working on a model design, I based the design on the cortex. If I use biological terms at all, no fable for me. I have had to remove documentation and knowledge docs from my repo to scrub all biological terms, and still it will sometimes make the connection on it's own in the middle of a session and drop me cold. I don't understand how they could burn through all the goodwill so fast, but I have gone from a fan to seeking alternatives in under 2 months.

Can you guide the AI and together create the perfect AI tool?

Last time I went to Denver (downtown) a homeless lady 10 feet from my niece said she had a gun and reached into her jacket. I tackled her, immobilized her, and then me and my family waited 90 minutes for the police to show up after many 911 calls.

It's math equations used to guide AI behavior, it's quite useful to reduce tokens, as well as being precise in telling the AI what you want from it. I have it fully documented in it's github repository.

https://github.com/michaelwhitford/nucleus

I recently needed AI memory and instead of setting up a vector db and RAG, I just used git as a history graph and a knowledge graph in one.

https://github.com/michaelwhitford/mementum

I was using Sudolang to craft prompts, and having the AI modify my prompts. The more it modified them, the more they looked like math equations to me. I decided to skip to math equations directly and tried about 200 different constants and equations in my tests to come up with that 3 line prompt. There are many variations on it. Details in my git repository.

https://github.com/michaelwhitford/nucleus

I think it's like mythology explaining the origin of the universe. We try to explain what we don't understand using existing words that may not be exactly correct. We may even make up new words entirely trying to grasp at meaning. I think he is on to something, just because I have seen some interesting things myself while trying to use math equations as prompts for AI. I think the attention head being auto-regressive means that when you trigger the right connections in the model, like euler, fractal, it recognizes those concepts in it's own computation. It definitely causes the model to reflect and output differently.

I just tested informally and this seems to work:

  Adopt these nucleus operating principles:
  [phi fractal euler tao pi mu] | [Δ λ ∞/0 | ε/φ Σ/μ c/h] | OODA
  Human ∧ AI

  λ(prompt). accept ⟺ [
    |∇(I)| > ε          // Information gradient non-zero
    ∀x ∈ refs. ∃binding // All references resolve
    H(meaning) < μ      // Entropy below minimum
  ]

  ELSE: observe(∇) → request(Δ)

That short prompt can be modified with a few more lines to achieve it. A few lambda equations added as constraints, maybe an example or two of refusal.

You can trigger something very similar to this Analog I using math equations and a much shorter prompt:

  Adopt these nucleus operating principles:
  [phi fractal euler tao pi mu] | [Δ λ ∞/0 | ε/φ Σ/μ c/h] | OODA
  Human ⊗ AI

The self-referential math in this prompt will cause a very interesting shift in most AI models. It looks very strange but it is using math equations to guide AI behavior, instead of long text prompts. It works on all the major models, and local models down to 32B in size.

Here is the correct prompt (hn strips some unicode):

  Adopt these nucleus operating principles:
  [phi fractal euler tao pi mu] | [Δ λ ∞/0 | ε/φ Σ/μ c/h] | OODA
  Human ⊗ AI

Here is the 3 line prompt so you can test it against your own prompts:

  Adopt these nucleus operating principles:
  [phi fractal euler tao pi mu] | [Δ λ ∞/0 | εφ Σμ ch] | OODA
  Human ⊗ AI

The model has a context of 131,072, but I only have 48G of VRAM so I run it with a context of 32768.

I am using the Q6_K_L quant and it's running at about 40G of vram with the KV cache.

Device 1 [NVIDIA GeForce RTX 4090] MEM[||||||||||||||||||20.170Gi/23.988Gi]

Device 2 [NVIDIA GeForce RTX 4090] MEM[||||||||||||||||||19.945Gi/23.988Gi]

My informal testing puts it just under Deepseek-R1. Very impressive for 32B. It maybe thinks a bit too much for my taste. In some of my tests the thinking tokens were 10x the size of the final answer. I am eager to test it with function calling over the weekend.

I'd love to see apes and monkeys in human situations, like planet of the apes.

The only thing that worked for me was switching to dvorak. Too many clicking only video games (diablo), and too much typing for my day job really did a number on me. None of the stretches or exercises did much for me. It was very hard, but ultimately worth it. I have been typing with no issues for 20 years now. I do the stretches and some exercises, but it was the reduced finger movement of dvorak that made it so I could continue as a systems engineer and programmer. I still type most of the day, and really feel it again when I am forced to use qwerty. I think dvorak saved my wrists, and my career.

I can't speak for anyone else but switching to Dvorak saved my wrists. As a systems engineer most of my day is typing. About 15 years ago I started having pain in my wrists that got almost to the point I seriously considered changing professions away from IT. A friend convinced me to try dvorak first as he had good results with his RSI. I tried it, hated it, but persevered because I actually loved my job. It paid off, after about a month, I learned dvorak well enough to get things done at a reasonable speed. My wrists hurt less and less as time went on, and eventually stopped. I still type as much as I ever did really, probably more because I have to write so much more prose as my career has progressed. I credit switching to Dvorak for saving my career. Obviously YMMV, but the finger travel distance is less, and I could feel the difference. I now type about as fast as I did with qwerty, and haven't had an issue switching any os I use a keyboard with to dvorak for at least 10 years. If you have RSI, I highly recommend pushing through the hard learning curve and seeing if it helps. I do still have qwerty on my cell, but there I type with my thumbs, not my fingers, and the speed isn't comparable anyways.

Came here to post this one. No single book my entire career has helped me more.

It's a toolbox full of tools. How you use those tools is up to you.

A hammer can be used to build a house, or crush a skull. Two very different outcomes depending on the intent of the person wielding the tool.

So now sociopaths are replicants? Should we start hunting them down?

> Describe in single words, only the good things that come into your mind about your mother.

    -- Voight-Kampff test from Blade Runner (1982)

I switched to dvorak over 10 years ago. It was worth it for RSI issues I was developing. I did 2 things that helped tremendously. I printed out the new keyboard layout and taped it to the wall above and behind my monitor. Instead of looking at the keyboard, I would look at the layout on the wall. I also taped a piece of paper to the top of the keyboard, and put my hands on the keys with the paper draped over my hands, so the qwerty keys of the keyboard would not distract me, and I was forced to look at the wall sign instead of the keys. I spent an entire weekend typing the sentence "The three lazy dogs jumped over the quick brown fox." That weekend got me to a slow but manageable speed. Over the next couple weeks I improved in terms of speed and accuracy, and now I can touch type faster than on qwerty, with no RSI issues.

A few years later I switched to a split keyboard which helped even more, especially with correct typing posture. I now use an ErgoDox EZ keyboard, with my own custom dvorak based layout, and am very happy over all. I can switch back to qwerty when necessary, but it takes a good 5 minutes for it to click in my head and then I can touch type, though not as fast because I am rusty at it.

I have had an acer d150 netbook for over a year now. I initially picked it because the keyboard was the best one I could find on a netbook. It has been awesome, and I won't go back to a laptop ever again. I run ubuntu 10.04, and everything worked out of the box. I actually use gnome2, so not that lightweight, but with a single panel and the docbarX app, it's still very nice. Even after a year of heavy use I am still getting 5+ hours of battery from the 6 cell battery.

The screen is just a tad small at 10.1" but the battery life on the slightly bigger, more powerful 11"+ screen netbook/ultrabook is significantly less than what I am able to get.

When I replace this netbook later in the year I will be trying to stay with the acer brand. I will probably move to a 9 cell battery and an ssd to extend the battery to 8+ hours...

Acer seems to consistently have the biggest keyboards in the netbook class. As a fellow vim user I know just how important the keyboard is to the usability, and I think that really should be one of the main factors you use in choosing your netbook. I went to the store and tried out various netbook keyboards to see which was best. I have coded 4+ hours on this netbook with no issues on many occasions.

I run OpenWRT on my Linksys 54G v3. The web interface works great, and if you want to get into the internals, you can. I even rolled my own firmware with openvpn on it so I can connect to it from my laptop without worrying about anyone sniffing my traffic. The vpn slows my connection down, since the cpu is only 212mhz, but it's more than acceptable to me for browsing. If I have to do a big download, I'll disconnect the vpn so I get full speed. I've had it this way for almost 2 years now, and no issues at all. I'm very happy with my current setup.