Commits on Mar 10, 2024

1. whisper : improve beam search candidate diversity …

   As of ggerganov#1486, whisper.cpp uses a unified KV cache with KQ masking.
   As a result, depending on their location in the batch,
   identical sequences in a batch can have slightly different outputs
   due to floating point rounding errors during reduction.
   See the discussion in ggerganov#1941 for more details.
   
   The beam search code used "has identical sum of log probabilities"
   as a shorthand for "is an identical token sequence". However, per above,
   identical tokens do not necessarily result in identical probabilities.
   
   Instead, explicitly compare on sequences.
   This is linear in cost when they are identical,
   but the lengths are always small and the comparisons are cheap.
   
   This increases diversity during beam search.
   
   This improves output quality for some short samples I've been working
   with, at no detectable performance cost.
   I haven't checked against larger corpuses.
   
   Fixes ggerganov#1941

   

   josharian committed Mar 10, 2024
   Configuration menu

   • View commit details
   • Copy full SHA for f99b649
   • Browse repository at this point

   Copy the full SHA

   f99b649 View commit details

   Browse the repository at this point in the history