3 ms·
If it works, you could use the llms for the first few thousand cases, then use these annotations to train an efficient supervised model, and switch to that. Th
by joelthelion 1y ago
If it works, you could use the llms for the first few thousand cases, then use these annotations to train an efficient supervised model, and switch to that.
That way it would be both efficient and cost-effective.
- bee_rider 1y agoIt always is fraught to make analogies between human brains and these learning models, but this sounds a bit like muscle memory or something.