Closed stoerr closed 1 year ago
We currently make word count limitations on the request, which is pretty inexact to make sure the requests aren't too long. There is already a function to count tokens - we can use that to replace this by limiting tokens.
Was done in https://github.com/ist-dresden/composum-AI/pull/7 .
We currently make word count limitations on the request, which is pretty inexact to make sure the requests aren't too long. There is already a function to count tokens - we can use that to replace this by limiting tokens.