Skip to content

Latest commit

 

History

History
11 lines (7 loc) · 430 Bytes

TODO.rst

File metadata and controls

11 lines (7 loc) · 430 Bytes
  • Implement vocabulary mapper which simply transform text into vocabulary-id given ngram language model or counts
  • Implement --min-ngram to perform filtering duing indexing (for bigrams and upward)
  • Implement mini-expgram, which supports only loading.
  • Support for ngram-suffix traversal in the backward ngram. Use an additional bit to indicate the end of context?
  • Port for other batch-queue systems, such as Torque.