Have you successfully fine-tuned an LLM for anything useful?

[johhns4]

I'm interested to see if people are fine-tuning the larger LLMs just for kicks or if there are situation where it makes sense (not including high-stakes decision making like medical/legal) and it is computationally efficient?

Smaller models that are cheap to host, sure, but in which cases does fine-tune a larger model (and host it) really shine? Oppose to just using RAG and a closed-source API.

Perhaps it makes sense if it is serving a huge customer base, and the tone of voice needs to be different, but the question is how much work it is to train it and if it is worth it.

I'm not against fine-tuning but curious what the actual use cases are, where it makes economic sense and how successful people/organisations have been.

[alexei_rudak]

I've trained LLM to make conversion from American English to British English and vise versa. To do that I've used 10k sentences for fine-tuning.

[muzani → alexei_rudak]

Similar, but I trained it on dialogue from Conan the Cimmerian. It makes some really epic speech.

I'm convinced this is the route to doing poetry, find a favorite poet and actually use samples of their writing rather than telling it "write poetry in the style of Robert E Howard"

[FezzikTheGiant → alexei_rudak]

how did you generate the data

[ldjkfkdsjnv]

Also curious about this, I suspect that fine tuning LLMs is more nuanced than people think