The differences are pretty big, but the simplest way to illustrate is to try to use gazebo, isaac...etc, and then try to build a whole physically interactive kitchen.
First off, it's gonna take you 3 months to author that thing, if you don't ragequit along the way.
Then, when you go to run, your "50 million steps per second" sim becomes 500 steps per second.
The reason we have robots doing backflips and acrobatics instead of actually useful stuff like picking up your house is making the scenes and getting the data is tough. It requires sensors like cameras and rendering, vs purely proprioceptive-only envs with a flat ground plane and no other physics interactions.
Right now, the industry is doing manual teleop to collect data because it's straight up easier than trying to build these sorts of things in simulators.
Odd, pretty sure it was you who misrepresented what I said in attempts to manipulate.
You were also the one who "exaggerat[ed]" my claims. I made a general statement about my thoughts about future AI-based software rather than human-coded.
I still think that's indeed the inevitable future. Doesn't seem like it's remotely outrageous or an exaggerated. I never said GameGAN would be that software, but you seem to want to make that be the case so you can put it down.
What makes you believe neural networks aren't or could not be deterministic? What makes you think NNs could not eventually produce far more robust, reliable, and secure operating systems?
Seems obvious to me, but I guess you're more informed than me :)
I suggested there could be a "future where many game engines are entirely or even mostly AI based like this. Or even things like operating system or other programs."
The thought here was just a wondering of what the future might be and if we might have far more AI based programs.
I still think the answer is a strong yes, this is a glimpse into the future. No where did I say GameGAN would be that engine. You're just trying your hardest to hate.
Heh, yeah, tough crowd I guess. The full code, models, and videos are all released and people are still skeptical.
I feel like 95%+ of papers don't do anything besides tell you what happened and you're just supposed to believe them. Drives me nuts. Not sure why all the hate when you could just see for yourself. I'd welcome someone who can actually prove the model just "memorized" every combo possible and didn't do any generalization. I imagine the original GameGAN researchers from NVIDIA would be interested too.
Interesting @ guided diffusion, not aware of its existence til now. We've had our heads down for a while. Will look into it, thanks!
> I immediately felt insincerity bordering on scamming the audience
MFW I read this. Jeez man. Model size is 173MB. It didn't just memorize every possible combo.
How the hell you went from our excitement about a fun project we shared on YT to accusing us of "scamming" the audience I really don't know. What a terribly rude and hateful attitude you have =/
In the end, everything is boiling down to matrix math, so you can always make the argument that no neural network is impressive if you want.
The model's size is ~173MB, depending on settings. That's not much space to have memorized every single possible combination of events, nor was our data enough to cover that either.
The GAN model is the game environment. You're playing a neural network. The novelty is no game engine, no rules, just learned how to represent the game and you can play it.
When you take an algorithm from a book, or copy and paste from stack overflow, you put a comment in the code with a link to the source, really no different than how you'd cite a quote in a paper.
To me, it just doesn't seem like it's even remotely challenging to figure out how to do this, or when to do this. If it's not yours, say where you got it.
When in doubt, cite it. What exactly would the harm be if you cited something when you weren't sure if it was necessary, anyway?
The plagiarism that I personally see is specifically code plagiarism. I am a programming educator on youtube.com/sentdex and pythonprogramming.net
Lately, I have been digging into this, and it's far more rampant than I ever expected (I am still digging, but we're talking in the 10's of thousands of examples that I've found with basic automated searching just in matches to my own personal code). I have found some seriously absurd examples where an entire portfolio consists of my code, and the person got a job from it at a large company.
Compare a student who writes their own code to the student who plagiarizes.
If you're the non-copy-pasta student, you're competing with the fakes for jobs.
If you're an employer, you're tasked with figuring out who is who, and I strongly doubt you would personally want the copy-paster at your business for both legal and productivity reasons.
I think some people confuse plagiarism and innovation, especially when we start to wrap in "intellectual property" into it.
Plagiarism is a shortcut used to fake skills/credentials.
Innovation is a real skill, though could be debated I am sure.
Intellectual property value is up for debate.
People who are cheating/faking their way, lying about their value/skills harms both employers and students.
Just don't let people debating about plagiarism try to sneak in innovation/building-upon as a means of a straw man.
We're talking copy and paste here. Maybe some synonym swaps.
The differences are pretty big, but the simplest way to illustrate is to try to use gazebo, isaac...etc, and then try to build a whole physically interactive kitchen.
First off, it's gonna take you 3 months to author that thing, if you don't ragequit along the way.
Then, when you go to run, your "50 million steps per second" sim becomes 500 steps per second.
The reason we have robots doing backflips and acrobatics instead of actually useful stuff like picking up your house is making the scenes and getting the data is tough. It requires sensors like cameras and rendering, vs purely proprioceptive-only envs with a flat ground plane and no other physics interactions.
Right now, the industry is doing manual teleop to collect data because it's straight up easier than trying to build these sorts of things in simulators.
This is why we're building Lucky.