From 59737f22fccb9c2ab8744a719f4dbb95eedf7943 Mon Sep 17 00:00:00 2001 From: Kenneth Heafield Date: Fri, 14 Dec 2012 12:48:26 -0800 Subject: Updated kenlm --- klm/lm/config.hh | 59 +++++++++++++++++++++++++++++--------------------------- 1 file changed, 31 insertions(+), 28 deletions(-) (limited to 'klm/lm/config.hh') diff --git a/klm/lm/config.hh b/klm/lm/config.hh index 739cee9c..0de7b7c6 100644 --- a/klm/lm/config.hh +++ b/klm/lm/config.hh @@ -11,46 +11,52 @@ /* Configuration for ngram model. Separate header to reduce pollution. */ namespace lm { - + class EnumerateVocab; namespace ngram { struct Config { - // EFFECTIVE FOR BOTH ARPA AND BINARY READS + // EFFECTIVE FOR BOTH ARPA AND BINARY READS + + // (default true) print progress bar to messages + bool show_progress; // Where to log messages including the progress bar. Set to NULL for // silence. std::ostream *messages; + std::ostream *ProgressMessages() const { + return show_progress ? messages : 0; + } + // This will be called with every string in the vocabulary. See // enumerate_vocab.hh for more detail. Config does not take ownership; you - // are still responsible for deleting it (or stack allocating). + // are still responsible for deleting it (or stack allocating). EnumerateVocab *enumerate_vocab; - // ONLY EFFECTIVE WHEN READING ARPA - // What to do when isn't in the provided model. + // What to do when isn't in the provided model. WarningAction unknown_missing; - // What to do when or is missing from the model. - // If THROW_UP, the exception will be of type util::SpecialWordMissingException. + // What to do when or is missing from the model. + // If THROW_UP, the exception will be of type util::SpecialWordMissingException. WarningAction sentence_marker_missing; // What to do with a positive log probability. For COMPLAIN and SILENT, map - // to 0. + // to 0. WarningAction positive_log_probability; - // The probability to substitute for if it's missing from the model. + // The probability to substitute for if it's missing from the model. // No effect if the model has or unknown_missing == THROW_UP. float unknown_missing_logprob; // Size multiplier for probing hash table. Must be > 1. Space is linear in // this. Time is probing_multiplier / (probing_multiplier - 1). No effect - // for sorted variant. + // for sorted variant. // If you find yourself setting this to a low number, consider using the - // TrieModel which has lower memory consumption. + // TrieModel which has lower memory consumption. float probing_multiplier; // Amount of memory to use for building. The actual memory usage will be @@ -58,10 +64,10 @@ struct Config { // models. std::size_t building_memory; - // Template for temporary directory appropriate for passing to mkdtemp. + // Template for temporary directory appropriate for passing to mkdtemp. // The characters XXXXXX are appended before passing to mkdtemp. Only // applies to trie. If NULL, defaults to write_mmap. If that's NULL, - // defaults to input file name. + // defaults to input file name. const char *temporary_directory_prefix; // Level of complaining to do when loading from ARPA instead of binary format. @@ -69,49 +75,46 @@ struct Config { ARPALoadComplain arpa_complain; // While loading an ARPA file, also write out this binary format file. Set - // to NULL to disable. + // to NULL to disable. const char *write_mmap; enum WriteMethod { - WRITE_MMAP, // Map the file directly. - WRITE_AFTER // Write after we're done. + WRITE_MMAP, // Map the file directly. + WRITE_AFTER // Write after we're done. }; WriteMethod write_method; - // Include the vocab in the binary file? Only effective if write_mmap != NULL. + // Include the vocab in the binary file? Only effective if write_mmap != NULL. bool include_vocab; - // Left rest options. Only used when the model includes rest costs. + // Left rest options. Only used when the model includes rest costs. enum RestFunction { REST_MAX, // Maximum of any score to the left - REST_LOWER, // Use lower-order files given below. + REST_LOWER, // Use lower-order files given below. }; RestFunction rest_function; - // Only used for REST_LOWER. + // Only used for REST_LOWER. std::vector rest_lower_files; - // Quantization options. Only effective for QuantTrieModel. One value is // reserved for each of prob and backoff, so 2^bits - 1 buckets will be used - // to quantize (and one of the remaining backoffs will be 0). + // to quantize (and one of the remaining backoffs will be 0). uint8_t prob_bits, backoff_bits; // Bhiksha compression (simple form). Only works with trie. uint8_t pointer_bhiksha_bits; - - + // ONLY EFFECTIVE WHEN READING BINARY - + // How to get the giant array into memory: lazy mmap, populate, read etc. - // See util/mmap.hh for details of MapMethod. + // See util/mmap.hh for details of MapMethod. util::LoadMethod load_method; - - // Set defaults. + // Set defaults. Config(); }; -- cgit v1.2.3