Skip to main content

Advertisement

Table 6 Peak memory usage in the first pass obtained on the BNSI evaluation set

From: Context-dependent factored language models

Vocabulary size Memory usage [MB]
  Good-Turing Knesser-Ney
  Bigram Trigram Bigram Trigram
60K 743 1287 745 1299
100K 917 1512 914 1530
200K 973 1776 1110 1815
300K 1252 1934 1273 1935