emporas

10 ms·

91.
by emporas 1y ago
Head over to groq.com, use the qwen-qwq-32b model, and take these examples [1] and put them at the start before the prompt. After that use the following command: write chapter 1 for a new Novel in Progress, take inspiration from the example
92.
by emporas 1y ago
It is knowledge that gets automated, rather than reasoning. I was thinking of the first solar civilization, which lives totally in space. Near a star, but not in a planet, and no gravitational pull anywhere. They build tubes 10 km long, a s
93.
by emporas 1y ago
> a popular German FinTech with "AI" in its name has way less automation than a 5 person startup. Maybe their Cobol programmers are too busy writing an all caps function for a month, instead of providing forms to their employee
94.
by emporas 1y ago
I made a short story in half an hour by using the description of the article. Not totally accurate based on the specs, there are no closed source humans and openhumans, but much faster. It needs 10 more hours of work for it to be a really c
95.
by emporas 1y ago
It imports the bugs as well. No human involvement needed. Automagically.
96.
by emporas 1y ago
Thanks. This workflow of 4 prompts, has the benefit of not using the mouse. I have a friend who uses Photoshop to make posters for bands, the resulting images are better, faces of real people are put in the poster, but he does 1 million cli
97.
by emporas 1y ago
I use gemini to create covers for songs/albums i make, with beautiful typography. Something like this [1]. I was dying of curiosity, how ideogram managed to create such gorgeous images. I figured it out 2 days ago. I take an image with
98.
by emporas 1y ago
It works very well (99.9%), when the problem resides at a familiar territory of the user's knowledge. When i know enough about a problem, i know how to decompose it into smaller pieces, and all (most?) smaller pieces have been already
99.
by emporas 1y ago
Depends on the use case. For programming where every small detail might have huge implications, 98% accuracy vs 99.99% is a ginormous difference. Other tasks can be more forgiving, like writing which i do all the time, then i load 3000 toke
100.
by emporas 1y ago
In other words don't use the context window. Treat it as a command line with input/output, in which the purpose of the command is to extract information signal or knowledge manipulation or data mining and so on. Also special care
101.
by emporas 1y ago
If all internet data could be saved in a disk, alongside with model weights, then what's the difference of pulling the knowledge out of the weights exclusively, versus weights and jpeg images? I don't see any difference. The only
102.
by emporas 1y ago
They use Go as well, I know someone who writes it for banks, but most of their infrastructure is written in COBOL. There are some sources [1], and some people have told me the same in person, not in exact numbers or percentages, but roughly
103.
by emporas 1y ago
You've to got a reputation system with identities also stored on the blockchain, and you accept the transactions only from the most trustworthy people. Counterfeit paper currency exists, and counterfeit bitcoin (not BTC) transactions m
104.
by emporas 1y ago
We will use better technology for electronic transactions. Most of banks worldwide still use COBOL for most (all?) their software infrastructure. You can do as many electronic transactions as you wish without internet or electricity, provid
105.
by emporas 2y ago
It did find an old account of mine that got banned, top of the list. I have to say 500 words for fingerprinting, that's mindblowing. It focuses on topic a lot, that's true.
106.
by emporas 2y ago
tree-sitter is a C library though. Only grammars for each particular lang are defined in javascript.
107.
by emporas 2y ago
Ok, let's focus on food then instead of being all over the place. I personally produce on average 1 ton of olive oil a year. A little bit less 800 kilos or something, but let's say it's 1 ton. I keep some for myself and the r
108.
by emporas 2y ago
Yeah, but when money becomes information then everyone has money. When music becomes information then everyone has music. When food, medicine and narcotics become information with genetic engineering, then everyone has food, medicine and na
109.
by emporas 2y ago
It reminds me of "Generative AI is just a phase. What’s next is interactive AI." The more i think about it however, command line applications are about as interactive as a program can be. Let's say one is interested to find o
110.
by emporas 2y ago
>We believe the future of writing is AI & human collaboration. That's on the right track based on my experience. Human + A.I. is the way for literature. I will make in the following weeks a small open source tool to generate com
111.
by emporas 2y ago
Emacs has pretty print JSON which makes it very easy to read. I don't find it possible, XML displayed for human consumption to be better than that.
112.
by emporas 2y ago
You can use the gmic plugin alongside with Gimp, and adjust the CMYK values. I use that all the time. Color adjustments like yellow -> orange, or green -> purple, might exceed your expectations. Photoshop is excellent at it as well, i
113.
by emporas 2y ago
Separating the "What" from the "How" is a pretty old idea originating from Prolog, and yes you can do that. Erlang for example, in which many of Lamport's ideas are implemented, is a variation of Prolog. Prolog by i
114.
by emporas 2y ago
Indeed, but you can get used to navigating around different functionality using default shortcuts and custom commands. Not everything can be automated, but some things can, and that's the point.
115.
by emporas 2y ago
With script-fu or python-fu you can make menus sub-menus and similar UX adjustments. I also make a command, and just run it using slash: /command. Better than clicking around but not live-tweakable, you have to refresh scripts first. I
116.
by emporas 2y ago
I have written a whole project [1] for making amateur videoclips [2] of A.I. generated music, using GPT and other LLMs. 10.000 to 12.000 lines of code was written exclusively by AIs. I didn't know at the start what should be done, and
117.
by emporas 2y ago
>Why is establishing genre a goal in the first place? What is culture, if not a common agreement on what is beautiful and ugly? Establishing a genre in music is not a goal, but we see it happen over and over again. It is how humans opera
118.
by emporas 2y ago
The problem with real music, is that it requires a hefty amount of musicians to establish a genre. This amount could be somewhere in the range of 100 to 1000 musicians. When this critical number is not amassed then the genre effectively die
119.
by emporas 2y ago
GPUs compute in two dimensional vectors, and auto-regressive models compute in sequence. Diffusion seems to me to be the more efficient and natural fit for computation on GPUs. When o3 solved the ARC[1] challenge, Sam Altman said it was rec
120.
by emporas 2y ago
Then you need a Large Arithmetic Model (LAM). We have that, it's called calculator. The LLM could invoke several command line programs, including calculators or anything else in which a deterministic answer is desirable. Structured out

More ›