diff options
| author | Kenneth Heafield <github@kheafield.com> | 2012-08-16 17:02:56 -0400 |
|---|---|---|
| committer | Kenneth Heafield <github@kheafield.com> | 2012-08-16 17:07:51 -0400 |
| commit | f82833385b7cf3c01dc2f92830119dfe3ebc573e (patch) | |
| tree | 6a8c04bb723d2504c461a398135d9645135a730a /klm/lm/build_binary.cc | |
| parent | a68aaea190197f17f308b91f66ceff222a784460 (diff) | |
KenLM update. Remove a couple of segfaults for weird input. Other oddball stuff.
Diffstat (limited to 'klm/lm/build_binary.cc')
| -rw-r--r-- | klm/lm/build_binary.cc | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/klm/lm/build_binary.cc b/klm/lm/build_binary.cc index c4a01cb4..49901c9e 100644 --- a/klm/lm/build_binary.cc +++ b/klm/lm/build_binary.cc @@ -25,7 +25,11 @@ void Usage(const char *name) { "-i allows buggy models from IRSTLM by mapping positive log probability to 0.\n" "-w mmap|after determines how writing is done.\n" " mmap maps the binary file and writes to it. Default for trie.\n" -" after allocates anonymous memory, builds, and writes. Default for probing.\n\n" +" after allocates anonymous memory, builds, and writes. Default for probing.\n" +"-r \"order1.arpa order2 order3 order4\" adds lower-order rest costs from these\n" +" model files. order1.arpa must be an ARPA file. All others may be ARPA or\n" +" the same data structure as being built. All files must have the same\n" +" vocabulary. For probing, the unigrams must be in the same order.\n\n" "type is either probing or trie. Default is probing.\n\n" "probing uses a probing hash table. It is the fastest but uses the most memory.\n" "-p sets the space multiplier and must be >1.0. The default is 1.5.\n\n" @@ -111,7 +115,7 @@ void ShowSizes(const char *file, const lm::ngram::Config &config) { for (long int i = 0; i < length - 2; ++i) std::cout << ' '; std::cout << prefix << "B\n" "probing " << std::setw(length) << (sizes[0] / divide) << " assuming -p " << config.probing_multiplier << "\n" - "probing " << std::setw(length) << (sizes[1] / divide) << " assuming -r -p " << config.probing_multiplier << "\n" + "probing " << std::setw(length) << (sizes[1] / divide) << " assuming -r models -p " << config.probing_multiplier << "\n" "trie " << std::setw(length) << (sizes[2] / divide) << " without quantization\n" "trie " << std::setw(length) << (sizes[3] / divide) << " assuming -q " << (unsigned)config.prob_bits << " -b " << (unsigned)config.backoff_bits << " quantization \n" "trie " << std::setw(length) << (sizes[4] / divide) << " assuming -a " << (unsigned)config.pointer_bhiksha_bits << " array pointer compression\n" |