You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If there's not much chunks say, 1K number of chunks, probably, the bigger the chunk is the better the answer is.
If there's much chunks say, 10K number of chunks, much more chunks will be retrieved, it's realy hard to distinguish which chunk is more relevant to quesition that we need to sent to LLM, if we have long chunks.
JinHai-CN
changed the title
[Question]: layout mode law, sze chunks missconception
[Question]: layout mode law, size chunks missconception
Dec 13, 2024
Describe your problem
I have read several articles arround chunking methods for law documents. In addition I have visit different specoaliced embbeding modelos
Articles always speak arround long embbedings and embbedings models has more a more context capacity.
Muy expiments goed in the same direction , longs chunks improve the results
But when I use ragflow law layout, It always create small chunks about a pararagth.
What is ny blind?
The text was updated successfully, but these errors were encountered: