On: Ask HN: People who switched from GPT to their own models. How was it?

[visarga]

> GPT 4 isn't that great, in general

same here, it doesn't adhere to explicit instructions, maybe one or two simple instructions are ok but not more complex ones

[exploringBytes → visarga]

More complex instructions should be split into multiple prompts, solving the challenge step by step

[visarga → exploringBytes]

Yes, but sometimes the two parts are also correlated so it's better if they are done at the same time.

[quickgist → exploringBytes]

Done this way, you also re-incur any input token costs for every additional step. If you're using a large context, this can be significant.

[theshrike79 → visarga]

YMMV, it depends very heavily on what you're asking it to do.

[MaxMatti → visarga]

What model are you using for more complex prompts?