I wish the author has written about the motives behind the language and why someone should choose Copper. We have in fact more programming languages than we actually can use.
Author here. I have written this language for many reasons but I won't detail here, I'll just say it's a lot of fun for me to design a language and write a compiler.
This language is a playground where I constantly add and remove features or try various experiments, so I have no intention to promote it. The only reason I have published it is that I publish an open source text editor written with this language.
After twenty years of evolution, it starts to become stable but not yet, I had recently added some basic option types.
In some cases, it can certainly help. However, it isn't helpful for deep thoughts or when innovative, creative ideas are needed.
I have found that speaking the steps out loud helps, especially with new, confusing working steps. For example, it helps when tying a complex knot or following the steps/work of others without explicit instructions. In such situations, speaking out loud helps me to remember the steps much faster (simple repetition through multiple pathways) and also helps me to detect and correct spatial and sequential mistakes. It also helps enormously when working with or teaching someone else.
I have a funny memory from my youth: one time, I had to solve a particularly difficult maths problem on the blackboard. I don't remember exactly why, but while I talked out loud about solution A, I wrote a slightly different step (B) on the board. Looking back, I realised I had made mechanical mistakes, so I had to correct them to get the final valid solution. My teacher laughed and joked, 'A sign of genius is to write A, say B, but only C is the true solution.' :) (no, he was just too kind).
I have often experienced my handwriting being much slower than my speaking, so sometimes they don't sync and funny mistakes are made (one line contains the words of the next line). But I fondly remember that particular moment.
On Openrouter at least, all Mino models are painfully slow. TTFT is many seconds and the models spend a lot of time on its CoT, making effective output per second many times slower than DS Flash. On Flash 4.1, I constantly get ~180 tok/s and the real work got done fast, too. So even when the Mimo 2.6 are smart models, I use them much less recently. Not to mention their costs are also higher than DS.
For me, the DeepSeek 4.1 models are currently the ultimate cost-effective model the rest have to match with, especially if you are not on a subscription plan with heavy discount.
GPT Sol 6.1 is however, for me the most efficient model, both in terms of time to finish and token usage.
It sounds pretty nice on paper. But I tested the huggingface.co/mradermacher/hybridagent-8b-agentic-gguf:IQ4_XS with a simple question: "Who is the pope?" and it got stuck in a thinking loop forever and simply could not finish the task.
```
Okay, the user is asking who the pope is. The pope is the leader of the Catholic Church. The current pope is Pope Francis, who became pope in 2013 after Pope Benedict XVI resigned. There's no indication of a change since then, so the answer is Pope Francis.
Okay, the user is asking who the pope is. The pope is the leader of the Catholic Church. The current pope is Pope Francis, who became pope in 2013 after Pope Benedict XVI resigned. There's no indication of a change since then, so the answer is Pope Francis.
Okay, the user is asking who the pope is. The pope is the leader of the Catholic Church. The current pope is Pope Francis, who became pope in 2013 after Pope Benedict XVI resigned. There's no indication of a change since then, so the answer is Pope Francis.
```
Then I also gave it the counterfeit ball problem, but it could not solve it either. But to be fair, no 8B model seem to be able to solve it.
```
I have 9 balls that look exactly the same. But I know one ball is different. I only have a scale. Call scale(b1,b2)={-1;0;1} will tell me whether the left object b1 is heavier, same weight or lighter than the right object b2. How many times at minimum do we have to apply scale to find the different ball? Explain how we can achieve this.
That looks like maybe a driver problem though? Are you sure the inferencing stack is properly recognizing an EOS token? New models often have weird bugs like that that require changes to the engine.
Have you tried increasing the repetition penalty? Needing to do this is already a red flag, but I'm curious if it helps at all.
Running LLMs locally on outdated hardware, I've come to the realisation that sub-10b models are broadly speaking kind of stupid. They can answer simple questions, but can't really solve problems.
The effect is so strong that with an 8GB GPU a 27b model can sometimes finish the task earlier than a 9b model, despite residing largely in RAM. If it's a MoE model, then it isn't even a competition.
Yes. I don't think this formulation of 9ball is solvable in less than 9 weighings because it doesn't allow you to weigh groups of balls. A set of balls is not an object, so it can't be put on the left or right side of the scale
If I was put in place of the LLM I legitimately couldn't tell if the intended solution is to ignore the wording and solve 9balls, or if this is a trick question to check if I spew out a memorized answer without considering the actual wording of the problem
I think you can do it in 5 weightings, because a 0 eliminates both balls. Worst case:
scale(1,2) = 0
scale(3,4) = 0
scale(5,6) = 0
scale(7,8) = -1 # If it was zero, the counterfeit would be ball #9.
scale(6,7) = 0 # Must compare to a known-good ball. Here #8 is counterfeit.
I see it like this: because greenfield coding is strong with AI, but improving/large rewriting an existing codebase is getting progressively harder and worse, if you don't get the architecture right in the first try, improve them in later rounds will be very hard,if not futile. Thus the poor performance is not a typical day 1 problem as with human coding, but a pestilence.
Meh, I wish Cloudflare would break down its "bot" traffic a little more. I assume most of that is just Yandex's scraper going down. Which is good for disrupting Russian daily life but its not like their propaganda machine is knocked out.
Also, purely anecdotally: there are almost no "Zelensky needs to step down" or "Ukraine is a corrupt state" comments under any of the today's posts about Ukraine on X.
Yeah. It’s expected from a vibe coding project, especially if they care more about the optics than functionality.
I heard from another Photoshop clone that they spent almost a year to polish the software, using AI nonetheless. The hardest part they told me is the getting the color management part right.
That for one app. A full suit with a “written in Rust” tag sounds today almost like a clickbait.
reply