novaRom

17 ms·

151.
by novaRom 4y ago
This is what Bing already does, and probably chatGPT when it generates some long text and then finally decides to remove it showing a banner of error occurred. This is trivial to do, I mean you can add this to Alpaca easily.
152.
by novaRom 4y ago
Playing a detective, who might be interested in disrupting chatGPT and destroying OpenAI's public image?
153.
ASML export ban will hurt Netherlands relationship with China, ambassador warns(nltimes.nl)
5 points by novaRom 4y ago|8 comments
154.
by novaRom 4y ago
OpenAI is among the authors, so this paper can be biased.
155.
by novaRom 4y ago
Everything is "healthy" and "unhealthy" depending on its amount. Good example is table salt. Too little and it feels unsatisfied, too much and it will poison you.
156.
by novaRom 4y ago
I watched it 4 times. Only then I started to understand what's happening. How great and unique this movie really is.
157.
by novaRom 4y ago
Basic statistical relations in observable world (on Human-Level scale) should somehow be reflected in human language and in mathematics. Deeper models tend to learn those relations, and produce good looking connections which fit in the real
158.
by novaRom 4y ago
Wait until Google goes down inevitably, then they will apply all their legal force just to save their sinking ship.
159.
by novaRom 4y ago
All the model's weights are just a compressed representation of data it was trained on. Keep in mind it's open data, created by millions of people for free and placed either on Stackexchange or Reddit or Facebook or on Wikipedia.
160.
by novaRom 4y ago
LLMs will be used to discover "not so obvious" connections in science. I think we will see first scientific discoveries done by LLMs very soon (in fact, such works are probably already happening). Expect major breakthroughs in phy
161.
by novaRom 4y ago
>> We, us or our means Sandoche Adittane (Micro Entreprise), company number 79218834400016. They don't provide their business address nor their country or phone number. How can it be trusted?
162.
by novaRom 4y ago
Civilizations don't last long in historical time, so chances to find something are probably low. But maybe something emerges after technology, a kind of new matter with unusual property?
163.
by novaRom 4y ago
Ever heard about Federated Learning? This is the way it goes. Also, I do run training with no matrix multiplication, just 3-bit weights, addition in log space, slight accuracy degradation, but much faster CPU only training.
164.
by novaRom 4y ago
Host something yourself also for educational reasons, just experimenting, this is how new applications and technologies to be discovered and created.
165.
by novaRom 4y ago
Talking to your computer? This will not last too long until another disruptive change will happen. How about computers will overtake those decisions you think you freely able to do right now. It is all running fast, accelerating actually.
166.
by novaRom 4y ago
Can you elaborate, what is "the AI apocalypse"? Is it just a symbolic metaphor or is there any scientific research behind this words? For me it's rather more unpredictable toxic environment we observe in the world currently,
167.
by novaRom 4y ago
It's not just vitamin D, it's a long chain of different reactions. I prefer to run topless always if temps are comfortably above +5 Celsius. Unfortunately, Europe is probably the darkest place on earth at least in winter months, b
168.
by novaRom 4y ago
Exactly! I guess only few agree this is comparable to the invention of the nuclear weapon. But that's what I think right now after some days of intensive testing. Every sane government should actually act right now to get AI superiorit
169.
by novaRom 4y ago
Because Google is already preparing chatGPT-like feature to be integrated into their Google search. They will agment it with paid ads of course.
170.
by novaRom 4y ago
For me, chatGPT is simply a good replacement for {stackexchange, reddit, wikipedia, github, google,bing, etc.}. All information is accessible in one place and it's easy to ask about anything and get a structured answer. My topics of in
171.
by novaRom 4y ago
10 fps should be fast enough to provide input tensors for real time inference with small scale transformers / convolutional nets.
172.
by novaRom 4y ago
What kind of operations is not clear. Wether it's a simple logic operation or a FMA is big difference.
173.
by novaRom 4y ago
What TOPS means exactly in "... TOPS*|(INT8) 404 ..." ?
174.
by novaRom 4y ago
Limited speed of signal transmission, limited amount of matter and energy per volume of space, topology of physical space, quantum uncertainty, tunneling, and other quantum effects.
175.
by novaRom 4y ago
Most people even stopped making landscape mode fotos an videos. Portrait mode is the king today. No rotation needed, shareable, full screen area utilization.
176.
by novaRom 4y ago
A good abstract section is not the title. https://writing.wisc.edu/handbook/assignments/writing-an-abs...
177.
by novaRom 4y ago
Ask yourself before buying anything: cost per year times its lifetime. Kilowatt hours, disposal costs, quality degradation (battery, software slowdown), etc.
178.
by novaRom 4y ago
It's not different from what a well-known company does with their devices by installing low-grade batteries with leaking capacity. Intentionally, planned obsolescence.
179.
by novaRom 4y ago
Rule Nr 1: never start reading a paper/article with no abstract/summary.
180.
by novaRom 4y ago
Germany, just prices in Euro from the cheapest Aldi store (you cannot get anywhere cheaper): Barilla pasta, 500g: 0.75 -> 1.49 Basmati rice, 1kg: 1.99 -> 2.69 White sugar, 1kg: 0.45 -> 1.35 White flour, 1kg: 0.39 -> 0.55 Sunflow

More ›