131–140 of 165 posts
"OpenAI technology, just an HTTPS call away" 'an' is only mean to proceed a vowel. Should say "OpenAI technology, just a HTTPS call away"
Earlier quoted context omitted.
State of the art can be (and usually is) born in academic labs.
I don't think this is true. The ResNet was born at Microsoft, DQN was born at Deepmind, the Transformer was born at Google, and GPT2 was born at OpenAI. I'm obviously biased since I work at an industry AI lab, but we both have important roles to play.
Earlier quoted context omitted.
Why is it not justified, beyond a moral belief? Seems rational that people would strive to make as much money as possible.
Rational is building community. Rational is searching for inner peace. Rational is being the best you can, and helping others. Greed is not rational. It's just rationalized.
"OpenAI technology, just an HTTPS call away" 'an' is only mean to proceed a vowel. Should say "OpenAI technology, just a HTTPS call away"
It depends how you pronounce "H". If you pronounce it aitch instead of haitch then using "an" in this context is totally correct. https://blog.apastyle.org/apastyle/2012/04/using-a-or-an-wit... .
Earlier quoted context omitted.
It's only Open™️ if I can run the API on my own machines.
If big LM's are the future then even if you had the model you couldn't run it on your own machines without having a DGX or two laying around.
The essence of open source is that the resources are made available to you (without warranty). That isn’t the case here.
Earlier quoted context omitted.
$10m per training run gets me a lot of engineering time to build our own version of this system and lease it to other customers. Just skip one training run and I've got a pretty good team.
Putting aside the question of whether it would ever be a choice between spending $10M on a training run and hiring a team for $10M, GPT transformers were the end result of decades of language research and innovations. You’re making it sound as though you can build the next iteration past GPT-3 for $10M, which I don’t think is the case.
Is it a confidence problem? Are the OpenAI folks not confident on a single use case? Or did I miss the live demo somewhere?
Why are there no live examples on the page. All I see is video presentations and some cached API response. Is it a confidence problem? Are the OpenAI folks not confident on a single use case? Or did I miss the live demo somewhere?
Why are there no live examples on the page. All I see is video presentations and some cached API response. Is it a confidence problem? Are the OpenAI folks not confident on a single use case? Or did I miss the live demo somewhere?
You can use the API live in multiple products, such as AI Dungeon ( https://play.aidungeon.io/ )!
In NLP there is a very clear and powerful new paradigm: train a HUGE language model using vast amounts of raw text. Then to solve the problem of interest, either fine-tune the model by training on your specific dataset (usually quite small), or 0/1-shot the learning somehow. The crucial question is : is this paradigm viable for OTHER types of data? My hypothesis is YES. If you train a HUGE image model using vast quan…