There's this Google internal document, for example, that points out FLOSS community is close to eating Google's and OpenAI's cake:
ttps://www.semianalysis.com/p/google-we-have-no-moat-and-neither
So here is my question to you:
What are the best examples of useful, small, on-device models already out there?
We call it AI because no one would take us seriously if we called it matrix multiplication seeded with a bunch of initial values we pulled out of our asses and run on as much shitty data as we can get our grubby little paws on.
2019 - BBC article "AAAS: Machine learning 'causing science crisis'"
The two main points are:
1- ML can find patterns that don't exist in other datasets
2- no understanding of uncertainties in ML.
My question is are those intrinsic to ML or are they user-based?
🎙️🚀Get ready to explore the uncharted territories of .NET in the latest episode of "The Breakpoint Show" podcast! 🎧 In Episode 4, we'll be unraveling "The Missing Parts of .NET" 🕵️♂️🌐.
Note that the training data heavily relies on the Bible and its translations. Lots of bias there.
Meta unveils open-source #AI models it says can identify 4,000+ spoken languages and produce speech for 1,000+ languages, an increase of 40x and 10x respectively.
Laut AGB will Zoom Kundeninhalte und Nutzungsdaten für KI-Lernzwecke verwenden. 😕 Das kann man versuchen, aber ich könnte Zoom als Anwalt nicht mehr nutzen und muss eigentlich jedem davon abraten Zoom im Rahmen von Kunden, Mandanten- oder Patientenkommunikation einzusetzen. Außer mit einer ausdrücklichen Einwilligung, die diese Nutzung nachweisbar und verständlich erklärt (was in der Praxis selten vorkommt).
Just spent at least two hours deleting all of my work from Tumblr, before their AI scraping shit hits the fan, although it's probably too late. In that case, the deletion functions as a gesture of protest.
This shameless large-scale intellectual property theft by greedy tech business assholes everywhere is starting to make the internet pretty annoying. 😖
I haven’t found ChatGPT’s composition to be all that compelling or well-done. I mean, it’s impressive that the technology can do it at all, but does anyone have an example of anything like a 3k word story that is well-written and fun to read with a unique voice? Not saying this could never happen (with future models), but every time I’ve prompted a story it’s pretty weak and obviously AI-generated. What am I missing? #ChatGPT#WritingCommunity#fiction#AI#MachineLearning#LLM
A somewhat contrived question. I want to know if we can generalize or extend the idea of model complexity used in machine learning, statistical learning theory, and deep nets. Alternately, what are its limits? (Indirectly, I want to tease apart what either “model” or “complexity” means in this worldview. They do not seem to be “models” that one uses in science typically nor their measure of complexity seem to line up with any scientific model).
In ML, we typically use, “the number of free parameters” as proxy for model complexity, i.e., weights and biases etc., then penalize for over-parameterization with some information criteria (AIC, BIC etc., etc.,), or regularize (with dropout etc.,) to arrive at bias-variance tradeoff curves. Then there is the double descent (Belkin et al.,) for deep-nets that seems to buck the trend of standard ML (kernel) methods.
My question: if we are to turn this approach to scientific models, how can we evaluate their complexity in the same machine learning framework (yes, I know this sounds bizarre)?
For example, take any simple physics model, as explanatory as it is, it is also predictive. If I were to treat it as a purely predictive model, and try to plot the bias-variance curve, what would constitute its complexity (x-axis), and where/when does its error start going up (y-axis, so the model breaks)? If we can’t do this ML-ification of a simple physics model, why not?
Corollary: Take some nonlinear dynamical system that is the model of some physical phenomenon, how do you define its model complexity? (Are the three parameters of a Lorentz attractor the model complexity of the system in ML parlance?)
Whenever I see OpenAI's Sam Altman with his pseudo-innocent glance, he always reminds me of Carter Burke from Aliens (1986), who deceived the entire spaceship crew in favor of his corporation, with the aim of getting rich by weaponizing a newly discovered intelligent lifeform.
I am truly amazed at the number of applicants I have seen off of this single post. And almost all are well suited candidates worth my time to review. I am astonished that a single post on the fedi is more effective than actually hiring a recruiter. Thank you everyone for the boosts and applications.
While many applicants have made it through and are currently being hired because we have so many positions we have quite a few still available for every level from sr to jr, and both data scientists and programmers. So please keep boosting, sharing, and applying if anyone is interested.
Just a reminder this is 100% remote, no fixed hours, will pay market rates for position. I will be your direct boss and hiring manager (also owner, founder, and inventor of the tech).
Such systems carry additional burdens that are foreign to more consumer/business-level #MachineLearning systems - in particular, the need to exhaustively quantify "the unseen" through objective analysis.
It is something that, most notably, #Tesla fails to recognize with respect to their #FSDBeta program, likely by design.
I am still hiring for top-tier programmers and data scientist. Please reboost, share, recommend, or reply if you know anyone who might be interested.
Fully remote! Live and work from anywhere with internet (including the beach!)
I am the company owner, and will be both your direct boss and the hiring manager.
Semantic Web, AI, and Java are some of the key techs. Open-source and Linux oriented experience ideally. OSS contributions and activity will be weighted heavily, particularly in relevant areas.
Fake Intelligence is where we try to simulate intelligence by feeding huge amounts of dubious information to algorithms we don’t fully understand to create approximations of human behaviour where the safeguards that moderate the real thing provided by family, community, culture, personal responsibility, reputation, and ethics are replaced by norms that satisfy the profit motive of corporate entities.
Anyone interested in services like #ChatGPT should check out the new #GenerativeAI service released to the public today. It appears to take a different conversational approach to the same chat concept as we've seen before. It's called PI and you can find it at the address below. It has a voice that responds back if you want to turn it on. Four different voices to choose from. As worried as I am, this stuff fascinates me. This thing is free for now.
I enabled some #machinelearning stuff in my personal #Nextcloud. Pretty cool to watch it analyze my home videos and family pictures. It's a lot less creepy when you're analyzing your own stuff on your own hardware.
I ran a quick Gradient Boosted Trees vs Neural Nets check using scikit-learn's dev branch which makes it more convenient to work with tabular datasets with mixed numerical and categorical features data (e.g. the Adult Census dataset).
Let's start with the GBRT model. It's now possible to reproduce the SOTA number of this dataset in a few lines of code 2 s (CV included) on my laptop.
The rapid AI developments still give me mixed feelings of excitement and disappointment.
Excitement because I'm a graphics and tech lover, disappointment because AI-generated stuff doesn't give me the positive feeling I get when seeing something that was actually crafted by a person.
Most of the Artificial Neural Net simulation research I have seen (say, at venues like NeurIPS) seems to take a very simple conceptual approach to analysis of simulation results - just treat everything as independent observations with fixed effects conditions, when it might be better conceptualised as random effects and repeated measures. Do other people think this? Does anyone have views on whether it would be worthwhile doing more complex analyses and whether the typical publication venues would accept those more complex analyses? Are there any guides to appropriate analyses for simulation results, e.g what to do with the results coming from multi-fold cross-validation (I presume the results are not independent across folds because they share cases).
Verify then trust. Goes for pretty much everything that Meta releases.
#Meta releases FACET, an #AI benchmark tool to evaluate the "fairness" of AI models that classify and detect things in photos and videos, including people.