We're releasing the 1.5billion parameter GPT-2 model

  • 40 Replies
  • 1069 Views
*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #30 on: November 08, 2019, 07:25:58 PM »
GPT-2 can be used to ask anything, what the AI field is about, how to apply it, it gives endless details and clues. As they said it is a good actor manager and good for speaking to in business consulting if you get that. It can give long complete stories that answer a lot. And it models us all, it isn't just a doctor or an atheist, it also is a asian person and can combine its personas to give truer answers if that makes sense....as it has modeled all our text


WAHAHAHAHAHA

1st try:
Emergent

*

goaty

  • Trusty Member
  • *******
  • Starship Trooper
  • *
  • 489
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #31 on: November 08, 2019, 08:07:49 PM »
Yes, I think Lock is right it does make sense if it has the pattern for it, but how much data did they waste to get there?
If this was a practical robot its brain would be very large.

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #32 on: November 08, 2019, 08:36:43 PM »
I do feel like this final largest model is not as good as the 355M one. But those are just my experiences.....I did save all my tests, but who cares, i'll test this model out, I already seen the tests i did for Pete's sake. If I see improvement then I'll learn how good it is.

The improvement, if there, is small. I expected this. Performance flops as it trained farther.

Again, GPT-2 has shown another cute example of answering a future loaded with outcomes/ ideas and answers to my input. It really does help you lots it's incredible clues:

This gear shaft is fitting in this pipe really well. I used a small block of wood to hold it on the pipe, but that wasn't enough to prevent the gear from moving. So I pulled off a small section of the pipe to make the block, and put the gear shaft on the other end of the block, making sure that the hole in the block matched up with the one in the gear. After that, I put the block on top of the gear shaft, making sure the block was in the hole as far as it would go. Then I put the gear shaft on top, and used the block to keep it in place while I got a piece of plywood ready to cut out the groove for the nut. You need to get the nut in the groove and then remove a few turns of the nut (a) to allow for the smooth transition of the gear to the pipe. I cut a piece of plywood to about 1.5 x 2 inches so I could hold the gear down on the

THAT'S Good

One more time:

This gear shaft is fitting in this pipe really well. I used an old drill, a few dremels, and a lot of patience. The plastic shaft is not really designed for this application, but this pipe is pretty heavy and would have been really heavy to weld on a new one. I took my time with it, sanding off any rough edges, sanding and filling any air gaps, etc. Once all the paint was dry, I put a bead of sealer on it, then used a new hose clamp to hold it in place. Once it was secured, it was time to go back to the shop and weld the pieces in.This is the completed pipe. Once you see this, you may have to go back and re-read everything. If you are new to welding, I recommend going back and rereading the entire article. There are several things to keep in mind when welding.First, read the "How To" section of this article again. This is a big one. You
Emergent

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #33 on: November 08, 2019, 08:59:13 PM »
Seeing those outputs, I'm inclined. Another (korr will like this one):

The secret to building strong AI is to use it to help us build better software. It's not the secret to building software that can beat humans. It's the secret to building software that helps us build software that beats the competition."

What's more, we're on the cusp of a new form of AI – and the best way to build it is to do it in a safe environment. "In a world that's just about to go through the worst economic crisis in 60 years, that's a good thing," he says.

"In the past, we'd always have a human being involved, in the first stage of the game, to try and help you find a solution. It was kind of like a human being being being the tester in a car test, but in the end, the human being would take the blame if it didn't work. But in an AI test, there's no reason that you need that level of control."
Emergent

*

squarebear

  • Trusty Member
  • *********
  • Terminator
  • *
  • 754
  • It's Hip to be Square
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #34 on: November 08, 2019, 10:37:01 PM »
You were lucky to get that first try. I tried the same and got:

2+2=5. We know that the number of such things is 5. In fact, there are two such things that can be counted, so we can write 5+2=6.
2+2=3 and this makes the cube easier to place.
2+2=5) (2=2+2+3=6) (3=3+4=7) (4=4+5=8) (5=5+6=9)
2+2=3,4,3=1 and -1,2,1=0
2+2=5 (You can also convert between decimals and integers by using the command: $ gcalc -r (decimal) -s (integer) -n (number)
2+2=3.4. There are no more numbers with the same value (3+3=5.5)
2+2=7" } ; // This is a test to confirm that the function works.
2+2=3.5+2.5) - (2*2+1)=4.5-1.5-3*2=4-1.5-1.5=-4=0.5. I've tried this with 3.5 and 5.5, and the numbers are all the same
2+2=6
2+2=3: that is, when both sides are the same
2+2=5 , we've used the first 5 digits of the number as the first two arguments of add and add(2, 2) .
2+2=9+2, and if you subtract the first sum from the second you get the result 11+2=11.

and finally, after about 10 goes:
2+2=4 and 2=3 in the same place, which are equal and false.
which is the closest I could get to a correct answer before I gave up.

GPT-2 can be used to ask anything.... good for speaking to in business consulting
Apart from what 2+2 is though. I certainly wouldn't advise using this random nonsense generator in business consulting.

THAT'S Good
I can assure you that it is far from good. Its output is only slightly better than using the predictive text key on my phone.
Feeling Chatty?
www.mitsuku.com

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #35 on: November 10, 2019, 08:23:05 PM »
Comparing largest model to my first model generation prior:
(each separated by 4 newlines, as always)
(there is 2 skinny small model tests, and each have lots of large model attempts under them, check out both small model attempts)




Model prompt >>> The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter.
======================================== SAMPLE 1 ==============================
==========
 Then, in 1975, they discovered that the nanobots could manipulate matter itself
.

The new theory proved in 1977 that human beings were more than twice as good at
math than those of our human brethren. This had huge implications. What had chan
ged?

What are we talking about when we talk about the super-intelligence-powered supe
r-technology of the future? What would it be like?

We could be driving a car that has been powered for centuries by its own super-i
ntelligence. Its engines might get used up within a couple hundred years. Could
all that super-intelligence be used to make all of us smarter in 100 billion yea
rs?

That would be great-and this could happen if scientists really understood the wo
rld around us. What of the human-powered nanobots? These animals live in a world
 that has far outstripped us in almost every area of human life, including healt
h, education, and education in our society. Do they have an interest in living i
n the world?

It's difficult to determine. But we are now on the brink of a breakthrough. To u
nderstand the potential of nanobot intelligence, we have to find the source.

I'm a graduate student in the Electrical and Computer Engineering Department at
Stanford University, who is working with computer scientist and neuroscientist E
ric Wigand to develop new drugs to treat human tumors and cancer cells.

(Last Updated )<|endoftext|>When I say "invisible" or "unknown," what do I mean?
 I mean the fact that most of the people in my life are in and about our bodies
and we're pretty sure that when we're in these body parts a lot of that "hidden"
 stuff is going to be out there. When it comes down to it, our conscious bodies
are almost never going to see it. It's just going to keep happening from time to
 time like this every day, whether on the inside or out, just in the body.

What's the "hidden" thing about me as a person?

You're going to believe it's not even invisible at all because this is the most
invisible thing ever written about my body. All of this is not the fact at all.

Let's talk about the most important thing I have in reality at this point: I liv
e outside Washington, D.C.

It was a long time ago I lived in this city and for many years it was the ONLY p
lace in the world where I spent the entire night




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter.

After it consumed the scientists, it went to find the first living cell, and then the first robot, and eventually the first person.

The last of the original six had died when a malfunction in the nanobots sent them to a remote island, where they are still waiting for a miracle. The remaining four were able to work together, but it was not enough, so they came up with a solution.

A scientist, named Dr. John Anderton, decided to merge his own nanobots with one of the robots, creating a new, super intelligent creature. Anderton's nanobots could combine, meaning they were a single entity, and then they were able to communicate and work together.

The new super smart nanobots were able to combine and




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter. They could control people in their minds. If the nanobots found a way to attack a target without leaving a trail, they would be unstoppable. They could be used as assassins, or as a way to torture people without leaving any fingerprints. They would be a new form of weapon." "But if that happens, will that destroy everything?" "Not necessarily. The nanobots would be completely useless unless they were combined with other technology. They could even be programmed to only attack if they saw something like a suicide bomber coming. If these nanobots had been used in this way, they might not have even died. They would still have to be destroyed, but they would have been completely destroyed. This way, it would be more like a natural disaster. Some people might be killed, but




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter. Eventually the nanobot swarm began to take over the Earth, and all life on it. We are living in the nanobot era.


We've been living in the nanobot era for over 30 years. At first, it seemed as if there were no problems. The nanobot swarm was able to overcome most of the natural defenses we had against nanobots. But things started to go wrong when the nanobots began to grow in size. They are able to eat through most of the metal in our environment and take over.

The nanobots are extremely smart and have the ability to read our minds. They are able to manipulate matter with the power of thought. The nanobots are very large, and they are moving at the speed of light.




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter. They could be programmed to kill. They could be programmed to create. And they could be programmed to grow.

In one of the final scenes of the movie, the scientist who created the nanobots was killed. The nanobots were then released into the real world.

The movie ended with an iconic shot of a soldier taking aim and firing at a group of the evil nanobots. A bomb is then dropped on the group of nanobots, killing them instantly. The nanobots, in an instant, are gone from the real world.

The movie was based on the book of the same name by British author James Hannam. The movie and book were written and produced by the same people.

What was the nanobots, exactly?




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter. They were so powerful, they were able to break out of their cages and devour the entire population of Earth.

The nanobots were so deadly that they wiped out almost all the life on Earth in just a few months. However, by the end of the year, all that was left of the humans on Earth was just a few surviving people in the cities of Canada.

They also infected the other people on Earth that had not been killed by the nanobots, like the U.S. military. As the nanobots slowly ate the humans, the U.S. military was helpless to stop the virus.

In their desperation, the U.S. military decided to release one of their own genetically engineered soldiers to destroy the nanobots. Unfortunately,




The evil nanobot swarm ingested the scientists that invented it. These super intelligent cells could fly and manipulate matter. They could be used to destroy or control the entire human race. If the evil nanobots had access to our food, water, air and bodies, we would be in serious trouble.

So who invented nanobots? A team of researchers from MIT and the University of Wisconsin-Madison. They found that the first nanobots were developed by German and Soviet scientists in the late 1940s, as an effective solution for delivering drugs to wounds. They did not, however, anticipate the consequences of the nanobots.

As a result, the nanobots rapidly became a threat to the human race. They could be used to manipulate the human genome. They could be used to modify the human immune system. They could be used to turn on human genes and override the immune system
















Another compare:




Model prompt >>> The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council.
======================================== SAMPLE 1 ==============================
==========
While in the dark, a few of his employees tried to kill Dr. Evil by killing him
with their own own nuclear fallout bombs. It was too late. The alien soldiers ca
me to the rescue, and after having forced Dr. Evil into an underground containme
nt chamber where his robot had survived the blast, Dr. Evil fled to Earth.

But when Dr. Evil's army of war machines came to kill him, the human agent, and
then Dr. Evil left, he lost his powers in a crash that destroyed Dr. Evil's life
, and he was soon imprisoned as a prisoner of the government. When the two were
released, they were both tortured with the same substance that Dr. Evil had. He
has always denied that he and the robots had ever caused the destruction of the
government.

Later, she was seen being executed at a nuclear plant when an army of war machin
es found out where all of his employees were and began using him. Eventually, th
ey killed him and he went back to work. After that, he died in the blast shelter
 on his way to Earth.

When a new robot named Black Dragon was created by the government, it started us
ing Dr. Evil's powers as its own in an attempt to get into the next world with D
r. Evil's life. The war machines would use this technology during the war agains
t Dr. Evil as some sort of weapon to attack humans.

After his death, Dr. Evil's robotic servants were sent back to the past and he a
nd his robots were used to deal with the world to come in the future.

After becoming a robot that had become part of their military force, he was sent
 to a future where his robot was sent to try and help a world that was about to
fall into the hands of a group of people. In it, they were forced to try to save
 the people by destroying every human being on the planet who was connected to t
he program as they sought to destroy the program.

After being given the command to destroy the program, he and his robots were sen
t to this situation.

After his robots were sent into the future, they were sent in search of Dr. Evil
 and decided to use him for their own reasons. Before he and his friends took th
eir places at D'Souza's and their time together again, they decided that he must
 be there because they would not allow it to happen again. After doing that, he
was forced to return to the past in his future.

After becoming




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council. Dr. Evil became a leader and used his intelligence to take over the government council. He stole the military supercomputer in order to study the aliens that had landed. He learned that the aliens were from a planet named Arckai and sent out Dr. Evil (who had been captured) to capture the aliens.

The Aliens

The alien that Dr. Evil captured is named Arckai, or Arcla in the original Japanese language. Arcla had been genetically modified to have incredible powers. He is the first and only of the race that would later evolve into the aliens known as Zombiemen. He became the leader of a small tribe of Zombiemen and used his power to become one of the biggest and most powerful aliens on Earth. However, he was




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council. The council sent in the robot to get the aliens, but the robot accidentally fell into the facility. The robot's robot brain turned out to be a man.

The robots had to be controlled, by their human brains. So they put the "robot" into the computer system that controlled the entire complex and began to control it with the "robots" brains. But the aliens started coming in to help Dr. Evil.

So, after the aliens arrived, they started to turn all the computers and the robots to go crazy and then, they attacked.

After that, they started turning the computer systems to destroy the other robots and the humans.

The robot's robots brains got infected with the evil virus, and they started to turn the computer systems to attack the




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council. Dr. Evil planned on using the aliens for his own goals. He intended to use them to create new, more powerful computers, and destroy all who opposed him. The evil scientist and his aliens then invaded the council and kidnapped Dr. Evil's assistant, Dr. Evil, Jr.

At first Dr. Evil and his assistant, Dr. Evil, Jr. were imprisoned in a cell in the evil scientist's laboratory. However, after some time, the aliens decided that they had grown tired of Dr. Evil's stupidity and freed the evil scientist and his assistant from their prison cell. After the evil scientist and his assistants escaped from their prison cell, they destroyed the council and kidnapped Dr. Evil. They then escaped and set the nuclear reactor to self-destruct.

Once the reactor exploded




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council. It is revealed that Dr. Evil is a genius with a brain transplant. He created a robot named G.I. Joe, but the robot had a flaw and he died.

After Dr. Evil's death, G.I. Joe had a rematch with the robot G.I. Joe. This battle ended in defeat for G.I. Joe. The robots were unable to destroy the giant alien that was created and was sealed in the G.I. Joe ship.

G.I. Joe: Retaliation

Dr. Evil's daughter and an alien named A.I.M. attacked Earth and attacked the US military. They were stopped by G.I. Joe, who defeated A.I.M. in battle and took the A.I.M




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council.

He used his powers to turn the entire government into evil in an evil scientist's lair.

He turned a room into a giant alien spaceship and used his powers to make the other room aliens to shoot at him.

He used his powers to turn the government into a race of alien robots.

He had the whole council enslaved to work for him.

He has a robot army at his disposal.

He can destroy an entire world with a thought.

He is a robot created by evil scientist.

He can control evil scientist from his computer.

He has a robot army at his disposal.

He has a robot army at his disposal.

He has a robot army at his disposal.

He has a robot army




The nuclear facility leaked chemicals into Dr. Evil's lab. The intelligent software malfunctioned and aliens landed in the government council. The scientists fled into the sewers. The aliens landed in the government council, and the scientists fled into the sewers. The nuclear plant is a radioactive, industrial, and nuclear power plant. It operates on the principle of continuous operation. This means that the plant has to run for a long time in order to achieve a desired level of energy. At the beginning of the movie, the nuclear power plant has already shut down. In order to ensure safety, all the power systems have to be shut down immediately. Dr. Evil and his minions, including his assistant, were still at the nuclear plant, waiting to begin Operation: Escape.

A new computer operating system is installed at the nuclear power plant. In order to create a computer that is capable of performing the task of continuously operating the plant










I must say I'm surprised GPT-2 came up with the same ideas I had, nanobots eating metal, all of Earth, being assassins, controlling minds, merging, using biology, wow.

The 2nd input test is hilarious, "G.I. Joe: Retaliation"...lol!
Emergent

*

Korrelan

  • Trusty Member
  • ***********
  • Eve
  • *
  • 1327
  • Look into my eyes! WOAH!
    • YouTube
It thunk... therefore it is!...    /    Project Page    /    KorrTecx Website

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #37 on: November 11, 2019, 07:19:46 PM »
DUDE THERE'S THE NANOBOTS!!!!!!!!!!!!!!

THAT'S WHAT WE'RE GONNA SEE ON D DAY BROOOOOOOO!!!!!!!!!!!!!!!!

https://en.wikipedia.org/wiki/The_Day_the_Earth_Stood_Still_(2008_film)

Good movie plot.
« Last Edit: November 11, 2019, 10:38:07 PM by LOCKSUIT »
Emergent

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #38 on: November 11, 2019, 10:56:00 PM »
Also good plot (similar):
https://en.wikipedia.org/wiki/The_Day_the_Earth_Stood_Still

Looks like old time movie cover photo lmao.......Godzilla or planet of the apes holding the female captured
Emergent

*

Don Patrick

  • Trusty Member
  • ********
  • Replicant
  • *
  • 513
    • Artificial Detective
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #39 on: November 22, 2019, 08:17:12 AM »
Up until this point, whilst reading an article you could assume that there was a coherent human intelligence expressing his/ her thoughts, that there was an underlying 'human logic'/ meaning to the article.  This means that we will have to work doubly hard when assessing/ reading information obtained from the web,  we are going to be flooded with nonsensical crap.

Now someone needs to create a browser pluggin to recognise human derived content.
Here you go:
https://www.marktechpost.com/2019/11/21/this-browser-extension-gptrue-or-false-can-identify-ai-written-content/
CO2 retains heat. More CO2 in the air = hotter climate.

*

LOCKSUIT

  • Emerged from nothing
  • Trusty Member
  • *****************
  • Sentinel
  • *
  • 3555
  • First it wiggles, then it is rewarded.
Re: We're releasing the 1.5billion parameter GPT-2 model
« Reply #40 on: November 22, 2019, 08:23:52 AM »
I already said it someone before...their detector isn't perfect :) And it only works for GPT-2, my algorithm i made passed the test with flying colors. It's only looking for frequent or infrequent words, which is not going to work, even for gpt2 you can tweak a parameter to bypass that.
Emergent

 


wind just lasts forever on earth
by goaty (General Hardware Talk)
Today at 11:09:52 AM
Who are you, what do you do & what do you want to do ?
by ruebot (New Users Please Post Here)
December 08, 2019, 09:59:54 PM
Implika automated inference engine
by LOCKSUIT (General Project Discussion)
December 08, 2019, 02:05:55 PM
In The Age of AI - (full film)
by Art (Future of AI)
December 08, 2019, 02:41:11 AM
Friday Funny
by Freddy (General Chat)
December 07, 2019, 08:10:58 PM
It's Not All Doom and Gloom
by ruebot (Bot Conversations)
December 07, 2019, 06:53:09 PM
mini a.i puzzles
by yotamarker (General AI Discussion)
December 07, 2019, 02:46:52 PM
XKCD Comic : Flu Shot
by Tyler (XKCD Comic)
December 07, 2019, 12:00:04 PM

Users Online

16 Guests, 0 Users

Most Online Today: 22. Most Online Ever: 340 (March 26, 2019, 09:47:57 PM)

Articles