jononor

On the practical side there are many good options. Using the voice effectively is a key skill in many professions, such as singing, public speaking, education, religious, management. So there are many practitioners, and lots of learning materials out there. So I recommend to check of there are any of those angles that resonate more with you. But very important - this is a challenge that needs practice to overcome - it cannot be dealt with only theoretically! So as a simple starting point is: go on YouTube and search for vocal exercises. Start with something like warmup routines. And then try something like saying the same sentence at 7 different loudness levels, from a whisper into someones ear to a SHOUT that needs to go to another person at the other side of a lake. Observe just how much power you got! This will be awkward and that is good - it means that you are stretching yourself, both physically and mentally! Like at the gym.

Then out a few sentences in the level you feel most comfortable with, plus 1 or 2 steps above. Try whole paragraphs.

PS: drink water and do not push harder if you feel pain.

The world models are not really useful yet. So they are starting lower, compared to LLM. So they probably have some decent gains to make still, before it gets really hard (diminishing returns).

Why is it like that, currently? There is no information added by a vector index compared to the original text. And the text is highly redundant and compressible with even lossless functions. Furthermore a vector index is already lossy and approximate. So conceptually it is at least possible to have an index that would be a fraction of the size of what is indexed?

Hypothetically that might ne true. But current systems do not do online learning. Several recent models have cutoff points that are over 6 months ago. It is unclear to which extent user data is trained on. And it is is not clear whether one can achieve meaningful improvements to correctness based on training on user data. User data might be inadvertently incorrect and it may also be adversarial, trying to out bad things in on purpose.

Logging everyones network data/metadata would likely be illegal under employment law in Norways. Other European countries may have same/similar rules. GDPR may also apply. So be careful with how broadly you apply that default.

It has not learned anything. It just looks in its context window for your answer. For a fresh conversation it will make the same mistake again. Most likely, there is some randomness and also some context is stashed and shared between conversations by most LLM based assistants.

I for one am quite surprised. Sometimes impressed. But also often frustrated. And occasionally disappointed. Sometimes worried about the negative follow-on effects. Working with current LLMs spans the whole gamut... But for coding we are at the point where even the current level is quite useful. And as the tools/systems get better, the usefilness is going to increase quite a bit. Even if models improve slowly from this point on. It will impact the whole industry over the next years, and since software is eating the world, will impact many other industries as well. Exponential? Perhaps in the same way as computers and Internet have been exponential - cost per X (say tokens) will probably go down exponentially the next years and decades, the same way cost per FLOP went down, on megabytes transferred. But those exponential gains did not results in exponential growth in productivity, or if so, the exponent is much much lower. And I suspect it will likely be the same for artificial intelligence.

A decent training pipeline will be able to reproduce models with equivalent aggregate performance (as measured by the evaluation metrics). And a high degree of similarity in behavior on specific inputs - but not identical. It will not be the same exact weights. But that is not a critical bar to reach. And may software builds also fails that bar.

Probably a fair bit of hubris, sure. But right now it is not possible or legal to operate a company without a CEO, in Norway. And I suspect that is the case in basically all jurisdictions. And I do not see any reason why this would change in an increasingly automated world. The rule of law is ultimately based on personal responsibility (limited in case of corporations but nevertheless). And there are so many bad actors looking to defraud people and avoid responsibility, those still need protecting against in an AI world. Perhaps even more so...

You can claim that the AI is the CEO, and in a hypothetical future, it may handle most of the operations. But the government will consider a person to be the CEO. And the same is likely to apply to basic B2B like contracts - only a person can sign legal documents (perhaps by delegating to an AI, but ultimately it is a person under current legal frameworks).

Some of their core skill is taking credit and responsibility for the work others do. So they probably assume they can take do the same for an AI workforce. And they might be right. They also take do the same already for what the machines in the factory etc produces.

But more importantly, most already have enough money to not have to worry about employment.

No the preferred way of making modifications is the weights _together_ with training (or fine tuning) scripts, and the entire evaluation pipeline to measure performance. And the data required to support all of this.

When someone joins your data science team your would give them all this code and data. Not just the weights and say - the weights are the source, modify that to improve the model, I look forward to see your MR next week.

EDIT: Heck, sometimes the way to make improvements (modifications) is just to improve the data, and not touch the training code at all. It is often one of the most powerful ways. You still need training code though, and evaluation to measure the impact.

If everyone else had stayed silent as well, then I would agree. But as it is right now they are juuust about managing to match the current pace of the other contenders. Which actually is fine, but they have previously set quite high expectations. So some will probably be disappointed at this.

There are so many people on-board with this idea, hypemen collaborators, that I think they might be safe for a year or two more. The hypers will shout about how miraculous it is, and tell everyone that does not get the promised value that "you are just holding it wrong". This buys them a fair amount of time to improve things.

The modern Microsoft with Azure, Office360, etc is not much threatened by open source software. Especially with Azure, open source is a fantastic compliment which they would like the world to produce as much of as possible. The same with AI models. They would look at charge for AI hosting and services, at premium due to already being integrated in businesses. They are going to bundle it with all their existing moneymakers, and then just jack up the price. No sale needed, just a bump in the invoices that are flowing anyway.

You made me think of the role of mental effort/exercise. In parts of the western world, we are already experiencing a large increase in dementia/alzheimer and related. Most of it is because we are doing better with other killers like heart etc, and many cancers also. But is said that mental activity is important to stave off degenerative diseases of the brain. Could widespread AI trigger a dementia epidemic? It would be 30 years out, but still...

The "numpy" Sobel code is not that good, unfortunately - all the iteration is done in Python, so there is not much benefit from involving numpy. If one would use say scipy.convolve2d on a numpy.array, it would be much faster.

There are several datastructures for numeric data that do not need marshalling, and are suitable for very efficient interoperetion between Python and C/C++/Rust etc. Examples include array.array (in standard library), numpy.array, and PyArrow.

Sorry to hear that. Props to you for keep on going!

Why is that concerning? The VCs job is to sell to the bigger fool. They are going to pump that narrative for years, so that they can make some exits. Regardless of underlying realism.

It looks like an emoji for someone getting bashed in the head with a long stick. So that makes sense?

We know that learning and building mental capabilities require effort over time. We know that when people have not been applying/practicing programming for years, their skills have atrophied. I think a good default expectation is that unused skills will go away over time. Of course the questions are, is the engagement we have with LLMs enough to sustain the majority of the skills? Or is there new skills one builds that can compensate foe those lost (even when the LLM is no longer used)? How quickly do the changes happen? Are there wider effects, positive and/or negative?

EDIT: 20x better with same algorithm, not 100x. Do not post before the coffee starts kicking in...

They have started. Their strategy is just for others to build it on top of their platform. And not build one, but hundreds of businesses. Much faster way to grow. Why retain 100+ internal product/integration teams when you can have 100+ external teams that will pay you and do the building, the trying and failing, in each niche.

The things that actually work out you can just buy or outcompete later. But that is another phase, perhaps 1 decade from now. Right now it is about starting alll the snowballs.

Every actor making codecs is trying to pull off an MP3, so that they extract rents from everyone else via licensing. They carpet bomb the field with patents to prevent free codecs from succeeding. aptX is an example of an non-free codec made in this manner :)

If anyone is interested in fast dynamic native C modules for MicroPython, you can check out emlearn-micropython - a machine learning and digital signal processing library. https://github.com/emlearn/emlearn-micropython Disclaimer: I am the maintainer

The same can be done in 0.02 seconds with MicroPython. I posted optimized code here, https://github.com/jonnor/embeddedml/tree/master/handson/mic...

I have added this now, as a reply to one of the other posts :) 100x speedup.

Ok, you guys have successfully nerd sniped me this morning... Here some experiments showing the use of code emitter to speed this code up massively. Link to code: https://github.com/jonnor/embeddedml/tree/master/handson/mic... The results on ESP32S3 (M5Stick AtomS3U), running MicroPython 1.24.1. All times in seconds, for the 2000 number sort.

bubble.py 19.119

bubble_native.py 9.482

bubble_viper.py 0.904

heapsort_viper.py 0.02

So one can do 100x better without changing the algorithm, and 1000x by changing it :)

Microcontrollers are a constrained platform. They can be plenty fast - but it is much more important to use the tools available, compared to on PC. MicroPython has excellent tools available for this.

This should not be used to conclude on the viability of using MicroPython for numeric type tasks. For that one should at least take into account the following:

Integers are much faster than floats (floats involve a pointer and a heap allocation, integers are stored in a single word/object).

array.array is preferred over list for something like sort. Continuous memory representation of numbers versus general purpose list of objects.

MicroPython has on-board "JIT" (native/viper emitters), have to explicitly annotate the function. Should give 4-10x improvements for this case.

MicroPython has an on-board assembler, so one can write ARM assembly and get that to a function.

MicroPython also has support for C modules, which expose a Python API. Including dynamic native modules which can be installed at runtime with the package manager.

Bubblesort is O(n*2), which hurts for even a few thousand numbers. Actual sorting on a microcontroller should be done with an O(n log n) algorithm.

Above average people can also use it to get average results. Which can actually be useful. For many tasks and usecases, the good enough threshold can actually be quite low.