Prices of the electricity we use to charge

lenny

Esteemed Pedelecer
May 3, 2023
3,629
1,093
US 'in secret talks with Putin ally to restart Nord Stream 2 pipeline' two years after it was blown up in sabotage attack
 

lenny

Esteemed Pedelecer
May 3, 2023
3,629
1,093
Economic And Market Data Signal A Recession Is Coming
"A slew of economic data is signaling that a recession is around the corner. The impending economic contraction, and possibly a recession, is primarily being caused by President Donald Trump’s tariffs imposed on Canada, China, and Mexico and the wave of retaliation which has now followed. Moreover, the chaotic layoff of federal workers will likely lead to a rise in unemployment and tightening of spending by those losing their jobs. Additionally, deportations of undocumented immigrants, as well as the fear thereof, is causing significant uncertainty in several important economic sectors such as construction, farming, hospitality, poultry, and small businesses."


Tariff war risks sinking world into new Great Depression, International Chamber of Commerce warns

Options traders are bracing for a stock-market crash
Demand for crash protection soared last week, according to Cboe data
 
  • Like
Reactions: Woosh

Ghost1951

Esteemed Pedelecer
Jun 2, 2024
2,007
882
After 47 years of deep space travel, and at 15.6 billion miles, Voyager 2 has just switched off more instruments to conserve its dwindling power reserves.

Most days, Voyager 2 sends data to the giant Deep Space Network antennas in California, Spain or Australia. The reason the antennas are dispersed like that is that they need to face the distant space probe to receive its signals. Amazingly, the radio transmitter only transmits a feeble 23 watts of radio power, but thanks to high gain antennas at both ends (the huge listening antennas are dishes 70 meters across - YES - 70 meters or 230 feet across) - the signals are received and processed and inform our knowledge of the outer reaches of our star's environment.

62236

The Voyager probes were launched to give us close up study of Jupiter and Saturn. They were never intended to look at Uranus and Neptune, but they passed by them all and now 47 years after launch, both of them are sending limited data back about the deep space environment. They won't last forever, because their nuclear electric generators are running down and eventually, they will shut down and go dark, but the probes will probably outlast the earth as they speed through space 34,390 miles an hour relative to the Earth. At that speed, if they were going in the right direction, and they aren't, it would take 75,000 years to get to the nearest star system to the sun - the Alpha Centauri group - three stars in orbit around each other.

If it takes 75000 years at that speed to get to the nearest star - 4.5 light years away, how long would it take to cross our own galaxy from side to side?

The Milky Way is 100,000 light years across, so crossing our own galaxy at 35,000 miles an hour would take 1.66 Billion years....... AND THIS IS JUST OUR OWN GALAXY!

It would take the Voyager probes 416 Billion years to reach the nearest other Galaxy, which is called Andromeda.

62237

The Andromeda galaxy, shown above is bigger than the Milky Way and it has about a trillion stars in it. The Milky Way is no shrimp though and has perhaps 400 billion stars - the sun is just one of them - a not very large, yellow star that will burn for about another 5 billion years, gradually swelling up so that by its end it will be a red giant probably reaching as far as the earth before it shrinks down and eventually goes dim.

You can see the Andromeda Galaxy with the naked eye if you have really good eyesight, but it is easier to spot with binoculars. Even that is tricky. You are seeing it as it was 2 million years ago before our species existed. That is how long it took the light to get here and go into your binoculars.

The Universe is infinitely large and utterly impenetrable to us. We can't see beyond a certain point because the distances are almost infinitely large. We just get a tiny peep at what is near enough to see.



Voyager 2 and its instrument layout.

 
Last edited:
  • Like
  • :D
Reactions: POLLY and Woosh

Woosh

Trade Member
May 19, 2012
20,819
17,066
Southend on Sea
wooshbikes.co.uk
Economic And Market Data Signal A Recession Is Coming
"A slew of economic data is signaling that a recession is around the corner. The impending economic contraction, and possibly a recession, is primarily being caused by President Donald Trump’s tariffs imposed on Canada, China, and Mexico and the wave of retaliation which has now followed. Moreover, the chaotic layoff of federal workers will likely lead to a rise in unemployment and tightening of spending by those losing their jobs. Additionally, deportations of undocumented immigrants, as well as the fear thereof, is causing significant uncertainty in several important economic sectors such as construction, farming, hospitality, poultry, and small businesses."



Tariff war risks sinking world into new Great Depression, International Chamber of Commerce warns

Options traders are bracing for a stock-market crash
Demand for crash protection soared last week, according to Cboe data
@sw and mikelbikel: What do you think?
 

Ghost1951

Esteemed Pedelecer
Jun 2, 2024
2,007
882
I am aware how users can trick LLMs into telling you how to commit crimes (or jailbreaking) or to hallucinate. That does not apply to customised AI models. It's just that to avoid hallucinations, the model needs to run through more filters such as semantic entropy to detect defective replies.
This discussion was taking place last August.

The rate of progress is surprisingly fast. The mad LLM hallucinations of last year are far less likely now. I can't remember seeing one on any of the big LLMs like Chat Gpt 4o for a good while. They are getting better and better and faster and faster at a rate that looks like it will make Moore's law look conservative.

You can even download small and 'tiny' LLMs to run on your phone or a low spec laptop, but don't expect them to be fast and don't expect to see the accuracy that is now inherent in ChatGPT higher level models, Chat GPT3 is still error prone, but you only get kicked off the free programme with ChatGPT4o if you have been hogging it. Then they send you down to the dunces class and you get the lower level stuff for a while.

If you want to try running a local LLM - and they are ENTIRELY on the phone with no outreach to the big boys running on giant supercomputers, there is an android app called PocketPal AI on Google Play. It is free and it gives access to a variety of models. I have run Llama 3.2-1b and Llama 3.2-3b successfully on phone and 2 chromebooks. The larger one - 3.2-3b is much more accurate with about 3 Bn parameters and about 2.5Gbyts in size, but it runs incredibly slowly on the weakest of my chromebooks. It is surprisingly good and makes very very few mistakes in my testing of it. It runs pretty well on a Samsung A536B which is not a rocketship phone. It takes a while to download the model files (model_name.gguf) because they are pretty huge. Even the small models are Gbyt size, but it is worth doing if you are interested in this stuff and how it works. The models will tell you how they work if you ask - but I think you are already across this stuff Woosh. You may already be running LLMs on your own machine. I started because one of mys sons set up a system on an ordinary pc and put it online so I could try it from here. He has now bought a much more powerful gaming pc with a GPU and will be setting it up soon. Even the old pc he set up ran Llama 3.2-3b about three times faster than my phone. He has deepseek on there too.

The surprising thing is that you can run any LLM AI wholly on a phone in your pocket. Both of the Llama models I referenced have an amazingly wide range of knowledge, and the larger one gets it right almost every time. The smaller Llama model makes regular mistakes and has hallucinations. I asked it which king came after Henry V11 and it said EdwarV1 which is wrong.He came after HenryV111. I think it may have been thrown by the notation I used to denote the roman numerals because when I asked the same question in words (Henry the seventh) it got it right.

I was disappointed to discover that the models also have built in a degree of censorship. They won't discuss controversial topics of a certain sort, though I only found this once.
 

MikelBikel

Esteemed Pedelecer
Jun 6, 2017
1,360
371
Ireland
The truth about DOGE’s AI plans: The tech can’t do that
Identify “mission-critical” jobs? Spot dead people on Social Security rolls? Government needs AI -— but what DOGE appears to be doing doesn’t add up.
I thought AI could do Anything?
At least that's what some on this forum seem to think. They even say it's gonna drive tanks and fire, pew, pew.
Are you saying it can't even spot dead people on social security roll? Just use Auto Attestation, stop payment, see who replies, simples! :)
 

soundwave

Esteemed Pedelecer
May 23, 2015
17,425
6,653