noosphr

6 ms·

91.
by noosphr 3mo ago
To match the print quality from the 19th century.
92.
by noosphr 3mo ago
Kids need an antidote to the brain damage that is standard mathematical notation. I prefer lisps variadic prefix notation, but if you're just doing fixed arity functions without higher order functions than prefix and post fix are as go
93.
by noosphr 3mo ago
3gb of ram aren't enough everyday tasks at any sane resolution. That we still don't have 1200 ppi desktop monitors is as stupid as using black and white screens in 2000.
94.
by noosphr 3mo ago
Again I've heard this since 2022 when gpt3.5 came out. This is like microprocessors in the 80s. Sure they double in capability every 18 months but the start is so pathetic it will be 30 years before they are good enough for everyday ta
95.
by noosphr 3mo ago
The moat is the compute.
96.
by noosphr 3mo ago
Chinese models most likely are distillations of frontier models with tricks for subpar hardware. If you want to be ahead of the us labs you need to spend billions for pretraining from scratch.
97.
by noosphr 3mo ago
A matrix is better represented as a nested list of lists: ((1 2 3)(4 5 6)(6 7 8)) Of course you ignore that representing a tensor of rank higher than 2 is just as difficult in 2d as a matrix is in 1d.
98.
by noosphr 3mo ago
Code is natively tree like and hyperbolic. We unfortunately can't embed a 2d hyperbolic space on a computer screen.
99.
by noosphr 3mo ago
There is more than one workload in AI. Inference for llms is memory constrained on even a single card. Training for llms is memory constrained on the level of racks. In both cases you hardly ever see more than 40% of advertised flops used.
100.
by noosphr 3mo ago
Because llm inference is not the only workload a GPU can do and custom silicon cost $10b a chip.
101.
by noosphr 3mo ago
No they don't.
102.
by noosphr 3mo ago
It only has that depth because you prompt it towards the experts that understand that depth. An example that happened 10 minutes ago: contracts in racket, it kept arguing that you can't use -> in a contract of a function with a rest
103.
by noosphr 3mo ago
So does llm inference. You're lucky if you hit 40% of the advertised flops.
104.
by noosphr 3mo ago
All vibe coded projects are in the public domain by definition. The question is if the massive copyright infringement that all the major labs engaged in means that all work derived from llms is under the copyright of the original owners of
105.
by noosphr 3mo ago
If you have to ask then the answer is no.
106.
by noosphr 3mo ago
Tens of millions in tokens for a code base which no one will understand in the end.
107.
by noosphr 3mo ago
I've found that model performance starts substantially degrading after 10%.
108.
by noosphr 3mo ago
Emperors new clothes.
109.
by noosphr 3mo ago
Google wants to be the AI that everyone uses, not the plumbing.
110.
by noosphr 3mo ago
Give it two years.
111.
by noosphr 3mo ago
Watching language package managers reinvent everything distribution package managers have been doing since the 90s has been as fun as watching crypto people reinvent financial regulation.
112.
by noosphr 3mo ago
Under any sane statistical model. If we assume that this is a naturally occurring once in century El Nino under an equal probability model we'd expect not to see it before we have 70 years of records.
113.
by noosphr 3mo ago
This isn't climate, it's weather. There are no proxies that record monthly water temperature. Only second order effects that are at best weakly correlated with water temperature.
114.
by noosphr 3mo ago
I'm not sure how you can make that claim with only 29 years of data without making some pretty big assumptions about the underlying distribution.
115.
by noosphr 3mo ago
This is why you need lisp. First you write a programming language in which the language primatives are the business logic primatives, then you write a trivial program that describes what the business does in the new language.
116.
by noosphr 3mo ago
If I may suggest people move away from harasses and start using Emacs. Gptel is amazing and the rest of Emacs plays incredibly well with llm development. Since the human is always in the loop token costs are also tiny compared to anything e
117.
by noosphr 3mo ago
Kids also need their own time too. Having a parent constantly interfering is just as bad as having one never showing interest. And you're not a playmate, you're a parent.
118.
by noosphr 3mo ago
Children follow by example. If you're an addict your kids will be too.
119.
by noosphr 3mo ago
Mindless cheerleading is what got us in the state we're in. More of it will not fix anything.
120.
by noosphr 3mo ago
And in Singapore I can get 10Gbps home internet. The future is not evenly distributed and the outliers matter little. >Electricity generation is almost always much, much cleaner than burning fuel in an internal combustion engine, and has

More ›