That is cool. So I could tell it to draw the AMC Hornet from the movie The Man with the Golden Gun doing a barrel roll. The stunt was real
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
AI News
- Thread starter Terminator857
- Start date
Large language models (LLMs) are expected to score 100% on all text based IQ tests in the future. Are they intelligent?
Daniel in SD
(supervised)
A.I. day now Sept 30, because might have a demo of optimus prime. I expect that date will get delayed also.
DanCar
Active Member
Tweet says Microsoft is doing a great job in AI now. Follow up replies argue it is still Google who is top dog.
DanCar
Active Member
How long before it is common that we carry chatGPT in our phones? I'd be surprised if it takes more than 2 years for first implementations. When will Tesla put chatGPT in the car? Will it require HW5?
www.autonews.com
GM explores using ChatGPT in vehicles
GM is exploring uses for ChatGPT as part of its broader collaboration with Microsoft, a company executive told Reuters.
www.autonews.com
mspohr
Well-Known Member
You already have it. Search engines.How long before it is common that we carry chatGPT in our phones? I'd be surprised if it takes more than 2 years for first implementations. When will Tesla put chatGPT in the car? Will it require HW5?
GM explores using ChatGPT in vehicles
GM is exploring uses for ChatGPT as part of its broader collaboration with Microsoft, a company executive told Reuters.
Don't need processing in the car if it's connected.
DanCar
Active Member
I asked Google a technical question about default package to handle an intent on android. I got a bunch of links and none of them answered the question. I asked chatGPT and got a perfect answer right away. I consider search engines a waste of time now for much of my needs.
DanCar
Active Member
mspohr
Well-Known Member
You can now run a GPT-3-level AI model on your laptop, phone, and Raspberry Pi
Thanks to Meta LLaMA, AI text models may have their "Stable Diffusion moment."
arstechnica.com
mspohr
Well-Known Member
Tesla's implementation of technology to improve the vehicle ownership experience continues to impress, and it has now found a novel way of improving ride comfort in the Model S sedan and the Model X crossover. In a post shared on Twitter, the EV maker explained how anonymized fleet data can help it garner information about the road conditions in specific areas.
Essentially, when enough Teslas detect a rough patch of road, the Model S and X - both of which come with a standard adaptive air suspension - can automatically raise the ride height to compensate for the harsh road surface. In turn, this lessens the level of harshness experienced by occupants. That's good news for two cars that were already excellent highway cruisers.
carbuzz.com
Essentially, when enough Teslas detect a rough patch of road, the Model S and X - both of which come with a standard adaptive air suspension - can automatically raise the ride height to compensate for the harsh road surface. In turn, this lessens the level of harshness experienced by occupants. That's good news for two cars that were already excellent highway cruisers.
Tesla Finds Genius Way To Improve Ride Comfort In Model S And Model X
Tesla's implementation of technology to improve the vehicle ownership experience continues to impress, and it has now found a novel way of improving ride comfort in the Model S sedan and the Model X crossover. In a post shared on Twitter, the EV maker explained how anonymized fleet data can help...
carbuzz.com
DanCar
Active Member
DanCar
Active Member
If you want to learn more about large language models, then skimming this paper might help:
DanCar
Active Member
$10 billion per quarter being invested in A.I. startups. A lot of that is going to nVidia for GPUs.
DanCar
Active Member
clydeiii
Member
DanCar
Active Member
Palm is the basis for the upcoming Gemini model, which Google claims is better than chatGPT-4, yeah talk is cheap.
Trained on 6144 v4 TPUs.
Quote:
PaLM 540B achieves breakthrough performance, outperforming the fine tuned state-of-the-art on a suite of multi-step reasoning tasks, and outperforming average human performance on the recently released BIG-bench benchmark.
... performance steeply increased we scaled to our largest model. PaLM also has strong capabilities in multilingual tasksand source code generation, which we demonstrate on a wide array of benchmarks.
... scaling improvements from large LMs have neither plateaued nor reached their saturation point.
... breakthrough capabilities in language understanding and generation across a number of difficult tasks. Specifically, Section 6.3 presents evaluation on a collection of reasoning tasks, which require multistep mathematical or commonsense reasoning to produce the correct answer.
End quote.
Summary: We can expect LLM to continue to improve with scaling. The future is so bright I got to wear shades.
Trained on 6144 v4 TPUs.
Quote:
PaLM 540B achieves breakthrough performance, outperforming the fine tuned state-of-the-art on a suite of multi-step reasoning tasks, and outperforming average human performance on the recently released BIG-bench benchmark.
... performance steeply increased we scaled to our largest model. PaLM also has strong capabilities in multilingual tasksand source code generation, which we demonstrate on a wide array of benchmarks.
... scaling improvements from large LMs have neither plateaued nor reached their saturation point.
... breakthrough capabilities in language understanding and generation across a number of difficult tasks. Specifically, Section 6.3 presents evaluation on a collection of reasoning tasks, which require multistep mathematical or commonsense reasoning to produce the correct answer.
End quote.
Summary: We can expect LLM to continue to improve with scaling. The future is so bright I got to wear shades.
clydeiii
Member
Not sure but I think the paper you cite was from 2022 and predates GPT-4. (At least, there are no mentions of in the paper you cite and the paper says "Submitted 10/22".)
DanCar
Active Member
My bad, your right. I got confused. According this it was just published:
Gemini is the next model after Palm-2, so that paper is two generations behind.
blog.google
Introducing PaLM 2
Today at I/O 2023, Google introduced PaLM 2, a new language model with improved multilingual, reasoning, and coding capabilities.
Similar threads
- Article
