Spikosauropod
Parliamentarian
Registered: 06-2007
|
|
Reply | Quote
|
|
Image GPT
We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting.
Last edited by Spikosauropod, 6/17/2020, 11:45 pm
|
|
6/17/2020, 11:44 pm
|
Link to this post
PM Spikosauropod
|
greendocnowciv
Registered: 11-2017
|
|
Reply | Quote
|
|
Re: Image GPT
Fantasy up to incredibly high quality VR has started, and this helps it along.
Just tell that not too far future Siri to make a model or movie star or whoever.
Choosey guys will become impossible connoisseurs of the perfect face - telling Siri to add a bit of this star, a bit of this model...
Regular, real beauties will just not make the grade.
I expect to still be alive when we get to this point - this ain't no "maybe in 20 years" fusion energy issue.
I hope I'm strong enough to not be one of the many who stick stuck in VR all the time at that point.
|
|
6/21/2020, 7:41 pm
|
Link to this post
PM greendocnowciv
Blog
|
