How is context-from-following-tokens implemented in the pure OAI API? I assumed there must be specialized models that have two separate context windows.
- they are working on a stateful API
- they are working on a cheaper version of GPT-4
Most probably this is driven by their use of it in ChatGPT, which is on fire from PMF. Clearly they're experimenting with the cheaper GPT-4 in ChatGPT right now as it's fairly turbo now, as discussed earlier today.
I've been programming in Cairo 1 in the past few weeks and must say that the developer experience is very good for web3, where it's usually very bad.
I also believe that it's the solution to blockchain scaling – this finally lets us have the entire 'web3' have more than the computing power of one single machine.
they pushed the rocket to the limits in this flight