zlacker

You know, we are dealing with people and people think differently so they won't move altogether. Moving to a new company under Alt is not an easy choice, at least the new company:

- Does not yet have a big model (need $$$ and months to train, if code is ready)

- Does not have proprietary code OpenAI has right now

- Does not have labeled data ($$$ and time) and chatgpt logs

- Does not have ChatGpt brand...

>>azurez+(OP)
I thought GPT-4 was not trained on labeled data, but simply on a large volume of text / code. Most of it is publicly accessible: wikipedia, archives of scientific articles, books, github, plus probably purchased data from text-heavy sites like Reddit.

>>buffer+E1
Assuming it's a reference to RLHF? Not sure

>>buffer+E1
No it's reinforcement learning with human feedback, RLHF lots of labeling

>>buffer+E1
Whatever they've built this year presumably uses all the positive/negative feedback on ChatGPT that they have a year worth of data now...

Another examples is the Be My Eyes data - presumably the vision part of GPT-4 was trained on the archive of data the blind assistance app has, and that could be an exclusive deal with OpenAI.