4 ms·
> It's likely that GPT-4 is on the optimal spot between cost and performance and there won't be significant improvements on performance in a near future. Given
by Tepix 4y ago
> It's likely that GPT-4 is on the optimal spot between cost and performance and there won't be significant improvements on performance in a near future.
Given the amount of resources being thrown at AI right now, i consider this to be very unlikely indeed.
- hnfong 4y agoI think GP's comment is probably apt if understood together with the qualification of "unless we find a better modeling architecture than Transformer".