Best way to match writing tone and style

API
,
1

Hi everyone!

I’m trying to fine tune a GPT3.5-Turbo model to match my writing tone and style.
I got good results using the 16k tokens model, but I want to know what would be the ideal way to fine tune the model with more examples that surpass the 16k token limit.

Any ideas or suggestions? Happy to hear your experience!

Thanks in advance!

2

Fine tuning is the the best way to ensure that the output matches your writing tone and style. If you have samples worth more than the limit, you could do the finetuning on chunks of data, then retune the already FT model