3 ms·
It is a debut of their thinking mode iirc. Unfortunately LLMs are shifting compute time to test time instead of train time. I don't really like this and frankl
by bangaladore 2y ago
It is a debut of their thinking mode iirc.
Unfortunately LLMs are shifting compute time to test time instead of train time. I don't really like this and frankly it shows a stalling of the architectures, data sets, etc...
- minihat 2y agoAnother take is that the base models are now good enough that spending more money for more intelligence is viable at test time. A threshold has been crossed.
- bangaladore 2y agoI guess I'd always thought the direct opposite. Naively, I feel to be useful, the goal of LLMs should be to more power efficient. So that eventually all devices can be smarter. Power efficiency can be gained through less time-time, or more "intelligence" or some combination of the two. I'm not convinced these SOTA models are doing much more than increasing test-time.
- holoduke 2y agoBiggest impacts on power efficiency will be the advances in node size and transistor type like nanosheet or forksheet. Algorithm will help just a little.