svjatoslav

Clone llama.cpp project from GitHub. Download LLM models from HuggingFace. TheBloke user posts lot of models in GGUF format. GPU is not needed to run these. 13B models should be fast modern computer CPU. llama.cpp offers batch mode, interactive chat mode and also web server mode.

OpenAI does not provide access to GPT-4 32K tokens model via API. Also they have weird restriction that you don't get access to even weaker GPT-4 model unless you had paid them at least 1$ before for API usage. They API prices are quite cheap, so I have to artificially generate lot of requests to them against GPT-3, so that I can pay them 1$, so that I get access to GPT-4. As far as I understand, this is the way to go currently.

I wrote tool to generate diagrams from Java code: https://www3.svjatoslav.eu/projects/javainspect/

It produces diagrams like these: https://www3.svjatoslav.eu/projects/sixth-3d/graphs/

Advantage of this is that diagram can be automatically updated from latest code. Classes discovery and visual layout is automatic.

Here is excellent article that explains hidden reasons behind wide spread phenomenon where people believe certain things or have certain views and you cannot change those using rational arguments:

https://meltingasphalt.com/crony-beliefs/

Spoiler: It is primarily about hidden self interest. But people will never admit it and will argue using any information they can find to support their goal. So it is pointless to attack with logic based counter arguments. Identify and counter actual self interest of the opponent instead.

I would like to create tool to analyze, interact and compute with arbitrary dimensional data. Shall greatly extend human capabilities.

General overview: https://www3.svjatoslav.eu/projects/sixth/ Ideas for computation: https://www3.svjatoslav.eu/projects/sixth-data/ 3D engine demos: https://www3.svjatoslav.eu/projects/sixth-3d-demos/

Because it is done on my little free time, it might be ready somewhere in the next 50 years or so.

You are using https://en.wikipedia.org/wiki/Straw_man argument. It is not about other people being obligated to pay me personally. The point is more general: there is lots of wasted talent. For example: Nicola Tesla did not had own money to support work that later changed the world. Instead he was stuck wasting his time digging trenches and working for Edisson for pennies. It was pure luck that he met George Westinghouse and with his support was able to realize some of his ideas. The point is that too often generally useful work that few gifted people can do does not result in timely or proper compensation to the author. Having dedicated effort for identifying such people and projects and supporting them would be very good use of philanthropic money. And Open Source projects is very good place to look.

I asked Richard Stallman this question: His answer was: Live cheaply.

Eric S. Raymond answer was: Be independently wealthy.

I have tried to reach billionaire philanthropists (Bill Gates, Steve Wozniak, Dean Kamen, Warren Buffet, and others) with the idea that there are very talented people around who have brilliant ideas (say for instance another Newton or Mozart) but are not financially independent to afford to work on those. Among good places to look for such talent is to identify and support important or promising open source projects and people behind them. But none of them bothered to respond.

I shared DMT idea to Lydia (article author).

Exactly !!! Someone should notify this team or others in the field. I think DMT controls neurons sensitivity (trigger happiness). In high amounts one gets hallucinations. But for vegetative patients low amount of DMT might restore brain activity to normal level.

I'm using it without data loss incidents for over 1.5 years now. On SSD drive I have encrypted volume. On top of it I have LVM volume. Ot top of it BTRFS partition. I use compress-force=lzo,autodefrag mount options. Works well for personal computer use case. I always follow latest kernel from Debian backports repo. I red that some RAID modes are implemented poorly it BTRFS still.