3 ms·
> That's the trick to doing computations with LLMs, the computation per token is constant so you want to split the computation over many tokens. Could you expl
by amelius 4y ago
> That's the trick to doing computations with LLMs, the computation per token is constant so you want to split the computation over many tokens.
Could you explain this a bit more? What are LLMs? Why is the number of tokens different in your approach?
- Gh0stRAT 4y agoLLM = Large Language Model. Asking it to show its work will cause it to generate more tokens (you could think of it as the logical leaps between each step being smaller the more detail you go into)