What settings or preprocess steps are people using on text docs?

[jasan-s]

Currently I'm simply using the following:
RecursiveCharacterTextSplitter({chunkSize: 1000})
to split text before vector DB upload. however, I want to learn a more optimized method. Please share your experience that yielded best results for you, e.g best chunkSize, overlap. preprocessing step etc.

[UncIeRick]

Hi jasan-s,

I wonder if you have any insights about this question now? Thank you!