Seems like a RL heavy response ranking mechanism for the judge model, to choose the best output for each prompt within the chosen models for each task, So the responses are as good as the best model in fusion, So if there is a better model that works better for the same prompt, the fusion would never be SOTA

non-anon Day-1

And count 2

count 1 of sorrow not setting up cloud agents in @cursor_ai

A Hunter X Hunter movie would hit right now, but there's none, so rewatching Kuroko

I refuse to believe any sane person would/could read or review this

It's just my first time living, excuse me for shit I'm about to say my whole life.

Fable just thinks you a twat and it makes up its own prompt, coz you can't write prompts for shit.

According to laws of nature, the closer we get to making the LLM models/anything better to a point that would set evolution backwards, the process would reach to the point of null somehow, or be restricted by nature in some way to be able to scale that could push back the process of evolution or I'm just retarded and it puts us in a nee brawl for survival of fittest and a new range of evolution that takes us to a very new place.

Yo @elonmusk do your thing and save humanity please

Finally in the era of 10$ a prompt for Fable.

kubectl client version Form a command that gives you the version of the client. I needed 3 tries to get it right

Faker runs Korea lol

It's funny how todo is still the universally understandable CRUD example

Umm

AI should never give you ideas, You give it ideas and it should help you build

"needful" is a bullshit word, and no other way

I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem. As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)! I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work. It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results? 88ms => 1.5ms 150K allocs => ~500 allocs Incredible right? Nope. My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path. This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput. The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity. Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.