new version of kenlm

13 files changed, 329 insertions, 102 deletions

diff --git a/klm/lm/binary_format.cc b/klm/lm/binary_format.cc
index 2a075b6b..69a06355 100644
--- a/klm/lm/binary_format.cc
+++ b/klm/lm/binary_format.cc
@@ -141,7 +141,6 @@ uint8_t *SetupBinary(const Config &config, const Parameters &params, std::size_t
 }
 
 uint8_t *SetupZeroed(const Config &config, ModelType model_type, const std::vector<uint64_t> &counts, std::size_t memory_size, Backing &backing) {
-  if (config.probing_multiplier <= 1.0) UTIL_THROW(FormatLoadException, "probing multiplier must be > 1.0");
   if (config.write_mmap) {
     std::size_t total_map = TotalHeaderSize(counts.size()) + memory_size;
     // Write out an mmap file.  
diff --git a/klm/lm/binary_format.hh b/klm/lm/binary_format.hh
index f95f05f7..a43c883c 100644
--- a/klm/lm/binary_format.hh
+++ b/klm/lm/binary_format.hh
@@ -67,9 +67,12 @@ template <class To> void LoadLM(const char *file, const Config &config, To &to)
     if (detail::IsBinaryFormat(backing.file.get())) {
       detail::ReadHeader(backing.file.get(), params);
       detail::MatchCheck(To::kModelType, params);
-      std::size_t memory_size = To::Size(params.counts, config);
-      uint8_t *start = detail::SetupBinary(config, params, memory_size, backing);
-      to.InitializeFromBinary(start, params, config, backing.file.get());
+      // Replace the probing_multiplier.  
+      Config new_config(config);
+      new_config.probing_multiplier = params.fixed.probing_multiplier;
+      std::size_t memory_size = To::Size(params.counts, new_config);
+      uint8_t *start = detail::SetupBinary(new_config, params, memory_size, backing);
+      to.InitializeFromBinary(start, params, new_config, backing.file.get());
     } else {
       detail::ComplainAboutARPA(config, To::kModelType);
       util::FilePiece f(backing.file.release(), file, config.messages);
diff --git a/klm/lm/build_binary.cc b/klm/lm/build_binary.cc
index 4db631a2..ec034640 100644
--- a/klm/lm/build_binary.cc
+++ b/klm/lm/build_binary.cc
@@ -1,13 +1,113 @@
 #include "lm/model.hh"
+#include "util/file_piece.hh"
 
 #include <iostream>
+#include <iomanip>
+
+#include <math.h>
+#include <stdlib.h>
+#include <unistd.h>
+
+namespace lm {
+namespace ngram {
+namespace {
+
+void Usage(const char *name) {
+  std::cerr << "Usage: " << name << " [-u unknown_probability] [-p probing_multiplier] [-t trie_temporary] [-m trie_building_megabytes] [type] input.arpa output.mmap\n\n"
+"Where type is one of probing, trie, or sorted:\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"
+"trie is a straightforward trie with bit-level packing.  It uses the least\n"
+"memory and is still faster than SRI or IRST.  Building the trie format uses an\n"
+"on-disk sort to save memory.\n"
+"-t is the temporary directory prefix.  Default is the output file name.\n"
+"-m is the amount of memory to use, in MB.  Default is 1024MB (1GB).\n\n"
+"sorted is like probing but uses a sorted uniform map instead of a hash table.\n"
+"It uses more memory than trie and is also slower, so there's no real reason to\n"
+"use it.\n\n"
+"See http://kheafield.com/code/kenlm/benchmark/ for data structure benchmarks.\n"
+"Passing only an input file will print memory usage of each data structure.\n"
+"If the ARPA file does not have <unk>, -u sets <unk>'s probability; default 0.0.\n";
+  exit(1);
+}
+
+// I could really use boost::lexical_cast right about now.  
+float ParseFloat(const char *from) {
+  char *end;
+  float ret = strtod(from, &end);
+  if (*end) throw util::ParseNumberException(from);
+  return ret;
+}
+unsigned long int ParseUInt(const char *from) {
+  char *end;
+  unsigned long int ret = strtoul(from, &end, 10);
+  if (*end) throw util::ParseNumberException(from);
+  return ret;
+}
+
+void ShowSizes(const char *file, const lm::ngram::Config &config) {
+  std::vector<uint64_t> counts;
+  util::FilePiece f(file);
+  lm::ReadARPACounts(f, counts);
+  std::size_t probing_size = ProbingModel::Size(counts, config);
+  // probing is always largest so use it to determine number of columns.  
+  long int length = std::max<long int>(5, lrint(ceil(log10(probing_size))));
+  std::cout << "Memory usage:\ntype    ";
+  // right align bytes.  
+  for (long int i = 0; i < length - 5; ++i) std::cout << ' ';
+  std::cout << "bytes\n"
+    "probing " << std::setw(length) << probing_size << " assuming -p " << config.probing_multiplier << "\n"
+    "trie    " << std::setw(length) << TrieModel::Size(counts, config) << "\n"
+    "sorted  " << std::setw(length) << SortedModel::Size(counts, config) << "\n";
+}
+
+} // namespace ngram
+} // namespace lm
+} // namespace
 
 int main(int argc, char *argv[]) {
-  if (argc != 3) {
-    std::cerr << "Usage: " << argv[0] << " input.arpa output.mmap" << std::endl;
-    return 1;
-  }
+  using namespace lm::ngram;
+
   lm::ngram::Config config;
-  config.write_mmap = argv[2];
-  lm::ngram::Model(argv[1], config);
+  int opt;
+  while ((opt = getopt(argc, argv, "u:p:t:m:")) != -1) {
+    switch(opt) {
+      case 'u':
+        config.unknown_missing_prob = ParseFloat(optarg);
+        break;
+      case 'p':
+        config.probing_multiplier = ParseFloat(optarg);
+        break;
+      case 't':
+        config.temporary_directory_prefix = optarg;
+        break;
+      case 'm':
+        config.building_memory = ParseUInt(optarg) * 1048576;
+        break;
+      default:
+        Usage(argv[0]);
+    }
+  }
+  if (optind + 1 == argc) {
+    ShowSizes(argv[optind], config);
+  } else if (optind + 2 == argc) {
+    config.write_mmap = argv[optind + 1];
+    ProbingModel(argv[optind], config);
+  } else if (optind + 3 == argc) {
+    const char *model_type = argv[optind];
+    const char *from_file = argv[optind + 1];
+    config.write_mmap = argv[optind + 2];
+    if (!strcmp(model_type, "probing")) {
+      ProbingModel(from_file, config);
+    } else if (!strcmp(model_type, "sorted")) {
+      SortedModel(from_file, config);
+    } else if (!strcmp(model_type, "trie")) {
+      TrieModel(from_file, config);
+    } else {
+      Usage(argv[0]);
+    }
+  } else {
+    Usage(argv[0]);
+  }
+  return 0;
 }
diff --git a/klm/lm/enumerate_vocab.hh b/klm/lm/enumerate_vocab.hh
index 7a2f7d12..e734316b 100644
--- a/klm/lm/enumerate_vocab.hh
+++ b/klm/lm/enumerate_vocab.hh
@@ -8,9 +8,10 @@ namespace lm {
 namespace ngram {
 
 /* If you need the actual strings in the vocabulary, inherit from this class
- * and implement Add.  Then put a pointer in Config.enumerate_vocab.  
- * Add is called once per n-gram.  index starts at 0 and increases by 1 each
- * time.  
+ * and implement Add.  Then put a pointer in Config.enumerate_vocab; it does
+ * not take ownership.  Add is called once per vocab word.  index starts at 0
+ * and increases by 1 each time.  This is only used by the Model constructor;
+ * the pointer is not retained by the class.  
  */
 class EnumerateVocab {
   public:
diff --git a/klm/lm/lm_exception.cc b/klm/lm/lm_exception.cc
index ab2ec52f..473849d1 100644
--- a/klm/lm/lm_exception.cc
+++ b/klm/lm/lm_exception.cc
@@ -5,14 +5,18 @@
 
 namespace lm {
 
+ConfigException::ConfigException() throw() {}
+ConfigException::~ConfigException() throw() {}
+
 LoadException::LoadException() throw() {}
 LoadException::~LoadException() throw() {}
-VocabLoadException::VocabLoadException() throw() {}
-VocabLoadException::~VocabLoadException() throw() {}
 
 FormatLoadException::FormatLoadException() throw() {}
 FormatLoadException::~FormatLoadException() throw() {}
 
+VocabLoadException::VocabLoadException() throw() {}
+VocabLoadException::~VocabLoadException() throw() {}
+
 SpecialWordMissingException::SpecialWordMissingException(StringPiece which) throw() {
   *this << "Missing special word " << which;
 }
diff --git a/klm/lm/lm_exception.hh b/klm/lm/lm_exception.hh
index 1216c4c7..3773c572 100644
--- a/klm/lm/lm_exception.hh
+++ b/klm/lm/lm_exception.hh
@@ -11,6 +11,12 @@
 
 namespace lm {
 
+class ConfigException : public util::Exception {
+  public:
+    ConfigException() throw();
+    ~ConfigException() throw();
+};
+
 class LoadException : public util::Exception {
    public:
       virtual ~LoadException() throw();
@@ -19,18 +25,18 @@ class LoadException : public util::Exception {
       LoadException() throw();
 };
 
-class VocabLoadException : public LoadException {
-  public:
-    virtual ~VocabLoadException() throw();
-    VocabLoadException() throw();
-};
-
 class FormatLoadException : public LoadException {
   public:
     FormatLoadException() throw();
     ~FormatLoadException() throw();
 };
 
+class VocabLoadException : public LoadException {
+  public:
+    virtual ~VocabLoadException() throw();
+    VocabLoadException() throw();
+};
+
 class SpecialWordMissingException : public VocabLoadException {
   public:
     explicit SpecialWordMissingException(StringPiece which) throw();
diff --git a/klm/lm/model.cc b/klm/lm/model.cc
index 6921d4d9..421e72fa 100644
--- a/klm/lm/model.cc
+++ b/klm/lm/model.cc
@@ -23,6 +23,7 @@ namespace detail {
 template <class Search, class VocabularyT> size_t GenericModel<Search, VocabularyT>::Size(const std::vector<uint64_t> &counts, const Config &config) {
   if (counts.size() > kMaxOrder) UTIL_THROW(FormatLoadException, "This model has order " << counts.size() << ".  Edit ngram.hh's kMaxOrder to at least this value and recompile.");
   if (counts.size() < 2) UTIL_THROW(FormatLoadException, "This ngram implementation assumes at least a bigram model.");
+  if (config.probing_multiplier <= 1.0) UTIL_THROW(ConfigException, "probing multiplier must be > 1.0");
   return VocabularyT::Size(counts[0], config) + Search::Size(counts, config);
 }
 
diff --git a/klm/lm/model.hh b/klm/lm/model.hh
index e0eeee17..53e5773d 100644
--- a/klm/lm/model.hh
+++ b/klm/lm/model.hh
@@ -12,6 +12,8 @@
 #include <algorithm>
 #include <vector>
 
+#include <string.h>
+
 namespace util { class FilePiece; }
 
 namespace lm {
@@ -21,9 +23,10 @@ namespace ngram {
 // Having this limit means that State can be
 // (kMaxOrder - 1) * sizeof(float) bytes instead of
 // sizeof(float*) + (kMaxOrder - 1) * sizeof(float) + malloc overhead
-const std::size_t kMaxOrder = 6;
+const unsigned char kMaxOrder = 6;
 
-// This is a POD.  
+// This is a POD but if you want memcmp to return the same as operator==, call
+// ZeroRemaining first.    
 class State {
   public:
     bool operator==(const State &other) const {
@@ -37,6 +40,22 @@ class State {
       return true;
     }
 
+    // Three way comparison function.  
+    int Compare(const State &other) const {
+      if (valid_length_ == other.valid_length_) {
+        return memcmp(history_, other.history_, valid_length_ * sizeof(WordIndex));
+      }
+      return (valid_length_ < other.valid_length_) ? -1 : 1;
+    }
+
+    // Call this before using raw memcmp.  
+    void ZeroRemaining() {
+      for (unsigned char i = valid_length_; i < kMaxOrder - 1; ++i) {
+        history_[i] = 0;
+        backoff_[i] = 0.0;
+      }
+    }
+
     // You shouldn't need to touch anything below this line, but the members are public so FullState will qualify as a POD.  
     // This order minimizes total size of the struct if WordIndex is 64 bit, float is 32 bit, and alignment of 64 bit integers is 64 bit.  
     WordIndex history_[kMaxOrder - 1];
diff --git a/klm/lm/model_test.cc b/klm/lm/model_test.cc
index 159628d4..b5125a95 100644
--- a/klm/lm/model_test.cc
+++ b/klm/lm/model_test.cc
@@ -4,6 +4,7 @@
 
 #define BOOST_TEST_MODULE ModelTest
 #include <boost/test/unit_test.hpp>
+#include <boost/test/floating_point_comparison.hpp>
 
 namespace lm {
 namespace ngram {
@@ -123,7 +124,7 @@ class ExpectEnumerateVocab : public EnumerateVocab {
     }
 
     void Check(const base::Vocabulary &vocab) {
-      BOOST_CHECK_EQUAL(34, seen.size());
+      BOOST_CHECK_EQUAL(34ULL, seen.size());
       BOOST_REQUIRE(!seen.empty());
       BOOST_CHECK_EQUAL("<unk>", seen[0]);
       for (WordIndex i = 0; i < seen.size(); ++i) {
@@ -144,6 +145,7 @@ template <class ModelT> void LoadingTest() {
   config.messages = NULL;
   ExpectEnumerateVocab enumerate;
   config.enumerate_vocab = &enumerate;
+  config.probing_multiplier = 2.0;
   ModelT m("test.arpa", config);
   enumerate.Check(m.GetVocabulary());
   Starters(m);
diff --git a/klm/lm/read_arpa.cc b/klm/lm/read_arpa.cc
index 8e9a770d..262a9c6a 100644
--- a/klm/lm/read_arpa.cc
+++ b/klm/lm/read_arpa.cc
@@ -49,7 +49,7 @@ template <class F> void GenericReadNGramHeader(F &in, unsigned int length) {
   while (IsEntirelyWhiteSpace(line = in.ReadLine())) {}
   std::stringstream expected;
   expected << '\\' << length << "-grams:";
-  if (line != expected.str()) UTIL_THROW(FormatLoadException, "Was expecting n-gram header " << expected.str() << " but got " << line << " instead.  ");
+  if (line != expected.str()) UTIL_THROW(FormatLoadException, "Was expecting n-gram header " << expected.str() << " but got " << line << " instead");
 }
 
 template <class F> void GenericReadEnd(F &in) {
@@ -110,7 +110,7 @@ void ReadBackoff(util::FilePiece &in, Prob &/*weights*/) {
       {
         float got = in.ReadFloat();
         if (got != 0.0)
-          UTIL_THROW(FormatLoadException, "Non-zero backoff " << got << " provided for an n-gram that should have no backoff.");
+          UTIL_THROW(FormatLoadException, "Non-zero backoff " << got << " provided for an n-gram that should have no backoff");
       }
       break;
     case '\n':
diff --git a/klm/lm/read_arpa.hh b/klm/lm/read_arpa.hh
index cabdb195..571fcbc5 100644
--- a/klm/lm/read_arpa.hh
+++ b/klm/lm/read_arpa.hh
@@ -54,7 +54,7 @@ template <class Voc, class Weights> void ReadNGram(util::FilePiece &f, const uns
     }
     ReadBackoff(f, weights);
   } catch(util::Exception &e) {
-    e << " in the " <<  n << "-gram at byte " << f.Offset();
+    e << " in the " << static_cast<unsigned int>(n) << "-gram at byte " << f.Offset();
     throw;
   }
 }
diff --git a/klm/lm/search_trie.cc b/klm/lm/search_trie.cc
index 182e27f5..12294682 100644
--- a/klm/lm/search_trie.cc
+++ b/klm/lm/search_trie.cc
@@ -1,3 +1,4 @@
+/* This is where the trie is built.  It's on-disk.  */
 #include "lm/search_trie.hh"
 
 #include "lm/lm_exception.hh"
@@ -8,6 +9,7 @@
 #include "lm/word_index.hh"
 #include "util/ersatz_progress.hh"
 #include "util/file_piece.hh"
+#include "util/proxy_iterator.hh"
 #include "util/scoped.hh"
 
 #include <algorithm>
@@ -30,43 +32,119 @@ namespace ngram {
 namespace trie {
 namespace {
 
-template <unsigned char Order> class FullEntry {
+/* An entry is a n-gram with probability.  It consists of:
+ * WordIndex[order]
+ * float probability
+ * backoff probability (omitted for highest order n-gram)
+ * These are stored consecutively in memory.  We want to sort them.  
+ *
+ * The problem is the length depends on order (but all n-grams being compared
+ * have the same order).  Allocating each entry on the heap (i.e. std::vector
+ * or std::string) then sorting pointers is the normal solution.  But that's
+ * too memory inefficient.  A lot of this code is just here to force std::sort
+ * to work with records where length is specified at runtime (and avoid using
+ * Boost for LM code).  I could have used qsort, but the point is to also
+ * support __gnu_cxx:parallel_sort which doesn't have a qsort version.  
+ */
+
+class EntryIterator {
   public:
-    typedef ProbBackoff Weights;
-    static const unsigned char kOrder = Order;
+    EntryIterator() {}
 
-    // reverse order
-    WordIndex words[Order];
-    Weights weights;
+    EntryIterator(void *ptr, std::size_t size) : ptr_(static_cast<uint8_t*>(ptr)), size_(size) {}
 
-    bool operator<(const FullEntry<Order> &other) const {
-      for (const WordIndex *i = words, *j = other.words; i != words + Order; ++i, ++j) {
-        if (*i < *j) return true;
-        if (*i > *j) return false;
-      }
-      return false;
+    bool operator==(const EntryIterator &other) const {
+      return ptr_ == other.ptr_;
+    }
+    bool operator<(const EntryIterator &other) const {
+      return ptr_ < other.ptr_;
+    }
+    EntryIterator &operator+=(std::ptrdiff_t amount) {
+      ptr_ += amount * size_;
+      return *this;
+    }
+    std::ptrdiff_t operator-(const EntryIterator &other) const {
+      return (ptr_ - other.ptr_) / size_;
     }
+
+    const void *Data() const { return ptr_; }
+    void *Data() { return ptr_; }
+    std::size_t EntrySize() const { return size_; }
+    
+  private:
+    uint8_t *ptr_;
+    std::size_t size_;
 };
 
-template <unsigned char Order> class ProbEntry {
+class EntryProxy {
   public:
-    typedef Prob Weights;
-    static const unsigned char kOrder = Order;
+    EntryProxy() {}
+
+    EntryProxy(void *ptr, std::size_t size) : inner_(ptr, size) {}
+
+    operator std::string() const {
+      return std::string(reinterpret_cast<const char*>(inner_.Data()), inner_.EntrySize());
+    }
+
+    EntryProxy &operator=(const EntryProxy &from) {
+      memcpy(inner_.Data(), from.inner_.Data(), inner_.EntrySize());
+      return *this;
+    }
+
+    EntryProxy &operator=(const std::string &from) {
+      memcpy(inner_.Data(), from.data(), inner_.EntrySize());
+      return *this;
+    }
+
+    const WordIndex *Indices() const {
+      return static_cast<const WordIndex*>(inner_.Data());
+    }
+
+  private:
+    friend class util::ProxyIterator<EntryProxy>;
+
+    typedef std::string value_type;
 
-    // reverse order
-    WordIndex words[Order];
-    Weights weights;
+    typedef EntryIterator InnerIterator;
+    InnerIterator &Inner() { return inner_; }
+    const InnerIterator &Inner() const { return inner_; } 
+    InnerIterator inner_;
+};
 
-    bool operator<(const ProbEntry<Order> &other) const {
-      for (const WordIndex *i = words, *j = other.words; i != words + Order; ++i, ++j) {
-        if (*i < *j) return true;
-        if (*i > *j) return false;
+typedef util::ProxyIterator<EntryProxy> NGramIter;
+
+class CompareRecords : public std::binary_function<const EntryProxy &, const EntryProxy &, bool> {
+  public:
+    explicit CompareRecords(unsigned char order) : order_(order) {}
+
+    bool operator()(const EntryProxy &first, const EntryProxy &second) const {
+      return Compare(first.Indices(), second.Indices());
+    }
+    bool operator()(const EntryProxy &first, const std::string &second) const {
+      return Compare(first.Indices(), reinterpret_cast<const WordIndex*>(second.data()));
+    }
+    bool operator()(const std::string &first, const EntryProxy &second) const {
+      return Compare(reinterpret_cast<const WordIndex*>(first.data()), second.Indices());
+    }
+    bool operator()(const std::string &first, const std::string &second) const {
+      return Compare(reinterpret_cast<const WordIndex*>(first.data()), reinterpret_cast<const WordIndex*>(first.data()));
+    }
+    
+  private:
+    bool Compare(const WordIndex *first, const WordIndex *second) const {
+      const WordIndex *end = first + order_;
+      for (; first != end; ++first, ++second) {
+        if (*first < *second) return true;
+        if (*first > *second) return false;
       }
       return false;
     }
+
+    unsigned char order_;
 };
 
 void WriteOrThrow(FILE *to, const void *data, size_t size) {
+  assert(size);
   if (1 != std::fwrite(data, size, 1, to)) UTIL_THROW(util::ErrnoException, "Short write; requested size " << size);
 }
 
@@ -84,21 +162,24 @@ void CopyOrThrow(FILE *from, FILE *to, size_t size) {
   }
 }
 
-template <class Entry> std::string DiskFlush(const Entry *begin, const Entry *end, const std::string &file_prefix, std::size_t batch) {
+std::string DiskFlush(const void *mem_begin, const void *mem_end, const std::string &file_prefix, std::size_t batch, unsigned char order, std::size_t weights_size) {
+  const std::size_t entry_size = sizeof(WordIndex) * order + weights_size;
+  const std::size_t prefix_size = sizeof(WordIndex) * (order - 1);
   std::stringstream assembled;
-  assembled << file_prefix << static_cast<unsigned int>(Entry::kOrder) << '_' << batch;
+  assembled << file_prefix << static_cast<unsigned int>(order) << '_' << batch;
   std::string ret(assembled.str());
   util::scoped_FILE out(fopen(ret.c_str(), "w"));
   if (!out.get()) UTIL_THROW(util::ErrnoException, "Couldn't open " << assembled.str().c_str() << " for writing");
-  for (const Entry *group_begin = begin; group_begin != end;) {
-    const Entry *group_end = group_begin;
-    for (++group_end; (group_end != end) && !memcmp(group_begin->words, group_end->words, sizeof(WordIndex) * (Entry::kOrder - 1)); ++group_end) {}
-    WriteOrThrow(out.get(), group_begin->words, sizeof(WordIndex) * (Entry::kOrder - 1));
-    WordIndex group_size = group_end - group_begin;
+  // Compress entries that being with the same (order-1) words.
+  for (const uint8_t *group_begin = static_cast<const uint8_t*>(mem_begin); group_begin != static_cast<const uint8_t*>(mem_end);) {
+    const uint8_t *group_end = group_begin;
+    for (group_end += entry_size; (group_end != static_cast<const uint8_t*>(mem_end)) && !memcmp(group_begin, group_end, prefix_size); group_end += entry_size) {}
+    WriteOrThrow(out.get(), group_begin, prefix_size);
+    WordIndex group_size = (group_end - group_begin) / entry_size;
     WriteOrThrow(out.get(), &group_size, sizeof(group_size));
-    for (const Entry *i = group_begin; i != group_end; ++i) {
-      WriteOrThrow(out.get(), &i->words[Entry::kOrder - 1], sizeof(WordIndex));
-      WriteOrThrow(out.get(), &i->weights, sizeof(typename Entry::Weights));
+    for (const uint8_t *i = group_begin; i != group_end; i += entry_size) {
+      WriteOrThrow(out.get(), i + prefix_size, sizeof(WordIndex));
+      WriteOrThrow(out.get(), i + sizeof(WordIndex) * order, weights_size);
     }
     group_begin = group_end;
   }
@@ -219,25 +300,37 @@ void MergeSortedFiles(const char *first_name, const char *second_name, const cha
   }
 }
 
-template <class Entry> void ConvertToSorted(util::FilePiece &f, const SortedVocabulary &vocab, const std::vector<uint64_t> &counts, util::scoped_memory &mem, const std::string &file_prefix) {
-  ConvertToSorted<FullEntry<Entry::kOrder - 1> >(f, vocab, counts, mem, file_prefix);
-
-  ReadNGramHeader(f, Entry::kOrder);
-  const size_t count = counts[Entry::kOrder - 1];
-  const size_t batch_size = std::min(count, mem.size() / sizeof(Entry));
-  Entry *const begin = reinterpret_cast<Entry*>(mem.get());
+void ConvertToSorted(util::FilePiece &f, const SortedVocabulary &vocab, const std::vector<uint64_t> &counts, util::scoped_memory &mem, const std::string &file_prefix, unsigned char order) {
+  if (order == 1) return;
+  ConvertToSorted(f, vocab, counts, mem, file_prefix, order - 1);
+
+  ReadNGramHeader(f, order);
+  const size_t count = counts[order - 1];
+  // Size of weights.  Does it include backoff?  
+  const size_t words_size = sizeof(WordIndex) * order;
+  const size_t weights_size = sizeof(float) + ((order == counts.size()) ? 0 : sizeof(float));
+  const size_t entry_size = words_size + weights_size;
+  const size_t batch_size = std::min(count, mem.size() / entry_size);
+  uint8_t *const begin = reinterpret_cast<uint8_t*>(mem.get());
   std::deque<std::string> files;
   for (std::size_t batch = 0, done = 0; done < count; ++batch) {
-    Entry *out = begin;
-    Entry *out_end = out + std::min(count - done, batch_size);
-    for (; out != out_end; ++out) {
-      ReadNGram(f, Entry::kOrder, vocab, out->words, out->weights);
+    uint8_t *out = begin;
+    uint8_t *out_end = out + std::min(count - done, batch_size) * entry_size;
+    if (order == counts.size()) {
+      for (; out != out_end; out += entry_size) {
+        ReadNGram(f, order, vocab, reinterpret_cast<WordIndex*>(out), *reinterpret_cast<Prob*>(out + words_size));
+      }
+    } else {
+      for (; out != out_end; out += entry_size) {
+        ReadNGram(f, order, vocab, reinterpret_cast<WordIndex*>(out), *reinterpret_cast<ProbBackoff*>(out + words_size));
+      }
     }
-    //__gnu_parallel::sort(begin, out_end);
-    std::sort(begin, out_end);
+    // TODO: __gnu_parallel::sort here.
+    EntryProxy proxy_begin(begin, entry_size), proxy_end(out_end, entry_size);
+    std::sort(NGramIter(proxy_begin), NGramIter(proxy_end), CompareRecords(order));
     
-    files.push_back(DiskFlush(begin, out_end, file_prefix, batch));
-    done += out_end - begin;
+    files.push_back(DiskFlush(begin, out_end, file_prefix, batch, order, weights_size));
+    done += (out_end - begin) / entry_size;
   }
 
   // All individual files created.  Merge them.  
@@ -245,9 +338,9 @@ template <class Entry> void ConvertToSorted(util::FilePiece &f, const SortedVoca
   std::size_t merge_count = 0;
   while (files.size() > 1) {
     std::stringstream assembled;
-    assembled << file_prefix << static_cast<unsigned int>(Entry::kOrder) << "_merge_" << (merge_count++);
+    assembled << file_prefix << static_cast<unsigned int>(order) << "_merge_" << (merge_count++);
     files.push_back(assembled.str());
-    MergeSortedFiles(files[0].c_str(), files[1].c_str(), files.back().c_str(), sizeof(typename Entry::Weights), Entry::kOrder);
+    MergeSortedFiles(files[0].c_str(), files[1].c_str(), files.back().c_str(), weights_size, order);
     if (std::remove(files[0].c_str())) UTIL_THROW(util::ErrnoException, "Could not remove " << files[0]);
     files.pop_front();
     if (std::remove(files[0].c_str())) UTIL_THROW(util::ErrnoException, "Could not remove " << files[0]);
@@ -255,14 +348,12 @@ template <class Entry> void ConvertToSorted(util::FilePiece &f, const SortedVoca
   }
   if (!files.empty()) {
     std::stringstream assembled;
-    assembled << file_prefix << static_cast<unsigned int>(Entry::kOrder) << "_merged";
+    assembled << file_prefix << static_cast<unsigned int>(order) << "_merged";
     std::string merged_name(assembled.str());
     if (std::rename(files[0].c_str(), merged_name.c_str())) UTIL_THROW(util::ErrnoException, "Could not rename " << files[0].c_str() << " to " << merged_name.c_str());
   }
 }
 
-template <> void ConvertToSorted<FullEntry<1> >(util::FilePiece &/*f*/, const SortedVocabulary &/*vocab*/, const std::vector<uint64_t> &/*counts*/, util::scoped_memory &/*mem*/, const std::string &/*file_prefix*/) {}
-
 void ARPAToSortedFiles(util::FilePiece &f, const std::vector<uint64_t> &counts, std::size_t buffer, const std::string &file_prefix, SortedVocabulary &vocab) {
   {
     std::string unigram_name = file_prefix + "unigrams";
@@ -275,7 +366,7 @@ void ARPAToSortedFiles(util::FilePiece &f, const std::vector<uint64_t> &counts,
   util::scoped_memory mem;
   mem.reset(malloc(buffer), buffer, util::scoped_memory::ARRAY_ALLOCATED);
   if (!mem.get()) UTIL_THROW(util::ErrnoException, "malloc failed for sort buffer size " << buffer);
-  ConvertToSorted<ProbEntry<5> >(f, vocab, counts, mem, file_prefix);
+  ConvertToSorted(f, vocab, counts, mem, file_prefix, counts.size());
   ReadEnd(f);
 }
 
@@ -390,7 +481,8 @@ void TrieSearch::InitializeFromARPA(const char *file, util::FilePiece &f, const
   temporary_directory.resize(strlen(temporary_directory.c_str()));
   // Add directory delimiter.  Assumes a real operating system.  
   temporary_directory += '/';
-  ARPAToSortedFiles(f, counts, config.building_memory, temporary_directory.c_str(), vocab);
+  // At least 1MB sorting memory.  
+  ARPAToSortedFiles(f, counts, std::max<size_t>(config.building_memory, 1048576), temporary_directory.c_str(), vocab);
   BuildTrie(temporary_directory.c_str(), counts, config.messages, *this);
   if (rmdir(temporary_directory.c_str())) {
     std::cerr << "Failed to delete " << temporary_directory << std::endl;
diff --git a/klm/lm/trie.cc b/klm/lm/trie.cc
index 8ed7b2a2..04bd2079 100644
--- a/klm/lm/trie.cc
+++ b/klm/lm/trie.cc
@@ -15,21 +15,21 @@ namespace {
 // Assumes key is first.  
 class JustKeyProxy {
   public:
-    JustKeyProxy() : inner_(), base_(), key_mask_(), total_bits_() {}
+    JustKeyProxy() : inner_(), base_(), key_mask_(), key_bits_(), total_bits_() {}
 
     operator uint64_t() const { return GetKey(); }
 
     uint64_t GetKey() const {
       uint64_t bit_off = inner_ * static_cast<uint64_t>(total_bits_);
-      return util::ReadInt57(base_ + bit_off / 8, bit_off & 7, key_mask_);
+      return util::ReadInt57(base_ + bit_off / 8, bit_off & 7, key_bits_, key_mask_);
     }
 
   private:
     friend class util::ProxyIterator<JustKeyProxy>;
-    friend bool FindBitPacked(const void *base, uint64_t key_mask, uint8_t total_bits, uint64_t begin_index, uint64_t end_index, WordIndex key, uint64_t &at_index);
+    friend bool FindBitPacked(const void *base, uint64_t key_mask, uint8_t key_bits, uint8_t total_bits, uint64_t begin_index, uint64_t end_index, WordIndex key, uint64_t &at_index);
 
-    JustKeyProxy(const void *base, uint64_t index, uint64_t key_mask, uint8_t total_bits)
-      : inner_(index), base_(static_cast<const uint8_t*>(base)), key_mask_(key_mask), total_bits_(total_bits) {}
+    JustKeyProxy(const void *base, uint64_t index, uint64_t key_mask, uint8_t key_bits, uint8_t total_bits)
+      : inner_(index), base_(static_cast<const uint8_t*>(base)), key_mask_(key_mask), key_bits_(key_bits), total_bits_(total_bits) {}
 
     // This is a read-only iterator.  
     JustKeyProxy &operator=(const JustKeyProxy &other);
@@ -44,12 +44,12 @@ class JustKeyProxy {
     uint64_t inner_;
     const uint8_t *const base_;
     const uint64_t key_mask_;
-    const uint8_t total_bits_;
+    const uint8_t key_bits_, total_bits_;
 };
 
-bool FindBitPacked(const void *base, uint64_t key_mask, uint8_t total_bits, uint64_t begin_index, uint64_t end_index, WordIndex key, uint64_t &at_index) {
-  util::ProxyIterator<JustKeyProxy> begin_it(JustKeyProxy(base, begin_index, key_mask, total_bits));
-  util::ProxyIterator<JustKeyProxy> end_it(JustKeyProxy(base, end_index, key_mask, total_bits));
+bool FindBitPacked(const void *base, uint64_t key_mask, uint8_t key_bits, uint8_t total_bits, uint64_t begin_index, uint64_t end_index, WordIndex key, uint64_t &at_index) {
+  util::ProxyIterator<JustKeyProxy> begin_it(JustKeyProxy(base, begin_index, key_mask, key_bits, total_bits));
+  util::ProxyIterator<JustKeyProxy> end_it(JustKeyProxy(base, end_index, key_mask, key_bits, total_bits));
   util::ProxyIterator<JustKeyProxy> out;
   if (!util::SortedUniformFind(begin_it, end_it, key, out)) return false;
   at_index = out.Inner();
@@ -96,67 +96,67 @@ void BitPackedMiddle::Insert(WordIndex word, float prob, float backoff, uint64_t
   assert(next <= next_mask_);
   uint64_t at_pointer = insert_index_ * total_bits_;
 
-  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, word);
+  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, word_bits_, word);
   at_pointer += word_bits_;
   util::WriteNonPositiveFloat31(base_ + (at_pointer >> 3), at_pointer & 7, prob);
   at_pointer += prob_bits_;
   util::WriteFloat32(base_ + (at_pointer >> 3), at_pointer & 7, backoff);
   at_pointer += backoff_bits_;
-  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, next);
+  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_bits_, next);
 
   ++insert_index_;
 }
 
 bool BitPackedMiddle::Find(WordIndex word, float &prob, float &backoff, NodeRange &range) const {
   uint64_t at_pointer;
-  if (!FindBitPacked(base_, word_mask_, total_bits_, range.begin, range.end, word, at_pointer)) return false;
+  if (!FindBitPacked(base_, word_mask_, word_bits_, total_bits_, range.begin, range.end, word, at_pointer)) return false;
   at_pointer *= total_bits_;
   at_pointer += word_bits_;
   prob = util::ReadNonPositiveFloat31(base_ + (at_pointer >> 3), at_pointer & 7);
   at_pointer += prob_bits_;
   backoff = util::ReadFloat32(base_ + (at_pointer >> 3), at_pointer & 7);
   at_pointer += backoff_bits_;
-  range.begin = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_mask_);
+  range.begin = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_bits_, next_mask_);
   // Read the next entry's pointer.  
   at_pointer += total_bits_;
-  range.end = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_mask_);
+  range.end = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_bits_, next_mask_);
   return true;
 }
 
 bool BitPackedMiddle::FindNoProb(WordIndex word, float &backoff, NodeRange &range) const {
   uint64_t at_pointer;
-  if (!FindBitPacked(base_, word_mask_, total_bits_, range.begin, range.end, word, at_pointer)) return false;
+  if (!FindBitPacked(base_, word_mask_, word_bits_, total_bits_, range.begin, range.end, word, at_pointer)) return false;
   at_pointer *= total_bits_;
   at_pointer += word_bits_;
   at_pointer += prob_bits_;
   backoff = util::ReadFloat32(base_ + (at_pointer >> 3), at_pointer & 7);
   at_pointer += backoff_bits_;
-  range.begin = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_mask_);
+  range.begin = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_bits_, next_mask_);
   // Read the next entry's pointer.  
   at_pointer += total_bits_;
-  range.end = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_mask_);
+  range.end = util::ReadInt57(base_ + (at_pointer >> 3), at_pointer & 7, next_bits_, next_mask_);
   return true;
 }
 
 void BitPackedMiddle::FinishedLoading(uint64_t next_end) {
   assert(next_end <= next_mask_);
   uint64_t last_next_write = (insert_index_ + 1) * total_bits_ - next_bits_;
-  util::WriteInt57(base_ + (last_next_write >> 3), last_next_write & 7, next_end);
+  util::WriteInt57(base_ + (last_next_write >> 3), last_next_write & 7, next_bits_, next_end);
 }
 
 
 void BitPackedLongest::Insert(WordIndex index, float prob) {
   assert(index <= word_mask_);
   uint64_t at_pointer = insert_index_ * total_bits_;
-  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, index);
+  util::WriteInt57(base_ + (at_pointer >> 3), at_pointer & 7, word_bits_, index);
   at_pointer += word_bits_;
   util::WriteNonPositiveFloat31(base_ + (at_pointer >> 3), at_pointer & 7, prob);
   ++insert_index_;
 }
 
-bool BitPackedLongest::Find(WordIndex word, float &prob, const NodeRange &node) const {
+bool BitPackedLongest::Find(WordIndex word, float &prob, const NodeRange &range) const {
   uint64_t at_pointer;
-  if (!FindBitPacked(base_, word_mask_, total_bits_, node.begin, node.end, word, at_pointer)) return false;
+  if (!FindBitPacked(base_, word_mask_, word_bits_, total_bits_, range.begin, range.end, word, at_pointer)) return false;
   at_pointer = at_pointer * total_bits_ + word_bits_;
   prob = util::ReadNonPositiveFloat31(base_ + (at_pointer >> 3), at_pointer & 7);
   return true;