HackerTrans
TopNewTrendsCommentsPastAskShowJobs

Ono-Sendai

841 karmajoined 14 лет назад
my blog: http://www.forwardscattering.org/

Submissions

Intuitive Quantum Electrodynamics

forwardscattering.org
33 points·by Ono-Sendai·8 дней назад·1 comments

Reflections on My Tech Career – Part 2

randomascii.wordpress.com
4 points·by Ono-Sendai·7 месяцев назад·0 comments

comments

Ono-Sendai
·7 часов назад·discuss
Adding MCP support to the open-source metaverse Substrata (https://substrata.info/)
Ono-Sendai
·7 часов назад·discuss
energy = frequency of angular rotation in the complex plane of a wavefunction.
Ono-Sendai
·4 дня назад·discuss
ok? well, textbooks can be wrong or misguided.
Ono-Sendai
·4 дня назад·discuss
Nah, special relativity is a consequence of magnetism.
Ono-Sendai
·4 дня назад·discuss
Interesting, I need to look into Maxwell's original formulation with the vortices. You can write electrodynamics in terms of small spinning elements: https://forwardscattering.org/page/Intuitive%20Quantum%20Ele...
Ono-Sendai
·5 дней назад·discuss
Also from this: https://terminator.fandom.com/wiki/Kokoro
Ono-Sendai
·7 дней назад·discuss
Kolomogorov complexity as an objective measure of the information contained in a string just doesn't work. It depends entirely on the language used for the program, and for any string, in some language, the complexity is zero.

See https://forwardscattering.org/page/Kolmogorov%20complexity

And no, the invariance theorem doesn't save you.
Ono-Sendai
·25 дней назад·discuss
Wonderful stuff, good job Midjourney.

An AI can be trained on body scans to detect diseases, tumours etc. Ideally this can be trained on real scans with real diseases but you could also train on synthetic data (synthetic bodies and/or synthetic diseases).

You can also focus ultrasonic waves to destroy (vaporise or cook) diseased tissue.
Ono-Sendai
·4 месяца назад·discuss
So now the algorithm is patent-free, and the vertex and fragment shaders are open-sourced with the MIT license, what we presumably need is some open-source code to take Bezier curves from a font file (or from the loaded data from FreeType or whatever), and process them into the data format that Slug expects.

Also thank you to Eric Lengyel, I have had my eye on Slug for a while and wished it was open-source.
Ono-Sendai
·4 месяца назад·discuss
Awesome, thanks.
Ono-Sendai
·4 месяца назад·discuss
I was going to ask if Slug can be used as a general vector renderer. Or does it assume limits on e.g. number of curves/paths per area that are typical of fonts?
Ono-Sendai
·4 месяца назад·discuss
Here's my fast acos, which I think can be converted to an asin: https://forwardscattering.org/post/66
Ono-Sendai
·4 месяца назад·discuss
For simple models (constant incoming radiance), you can indeed just add the optical depths from the different fog 'layers'. (90% sure but the maths is easy to check anyway, see https://forwardscattering.org/post/72)
Ono-Sendai
·4 месяца назад·discuss
Coincidentally I just posted a blog post on fog rendering, in particular with exponential height fog, yesterday: https://forwardscattering.org/post/72
Ono-Sendai
·5 месяцев назад·discuss
Use a multithreaded blocking approach. Much nicer than async.
Ono-Sendai
·5 месяцев назад·discuss
Substrata: open-source metaverse: https://substrata.info/
Ono-Sendai
·5 месяцев назад·discuss
I haven't thought about it deeply. But I guess it's about allowing the model to easily distinguish the prompt from the conversation. Models seem to get confused with escaping, which is fair enough, escaping is very confusing. It's true that for the transformer architecture the prompt and conversation are in the same stream. However you could do something like activate a special input neuron only for prompt input. Or have the prompt a fixed size (e.g. a fixed prefix size). And then do a bunch of adversarial training to punish the model when it confuses the prompt and conversation :)
Ono-Sendai
·5 месяцев назад·discuss
They could be if models were trained properly, with more carefully delineated prompts.
Ono-Sendai
·6 месяцев назад·discuss
"look, I'm sorry, but the rule is simple: if you made something 2x faster, you might have done something smart if you made something 100x faster, you definitely just stopped doing something stupid"

https://x.com/rygorous/status/1271296834439282690
Ono-Sendai
·6 месяцев назад·discuss
I use webassembly for Substrata (https://substrata.info/). It works pretty well, allows building a c++ app using OpenGL for the web.