Baum Welch training for HMMs

1 files changed, 316 insertions, 0 deletions

diff --git a/training/crf/mpi_baum_welch.cc b/training/crf/mpi_baum_welch.cc
new file mode 100644
index 00000000..d69b1769
--- /dev/null
+++ b/training/crf/mpi_baum_welch.cc
@@ -0,0 +1,316 @@
+#include <sstream>
+#include <iostream>
+#include <vector>
+#include <cassert>
+#include <cmath>
+
+#include "config.h"
+#ifdef HAVE_MPI
+#include <boost/mpi/timer.hpp>
+#include <boost/mpi.hpp>
+namespace mpi = boost::mpi;
+#endif
+
+#include <boost/unordered_map.hpp>
+#include <boost/functional/hash.hpp>
+#include <boost/shared_ptr.hpp>
+#include <boost/program_options.hpp>
+#include <boost/program_options/variables_map.hpp>
+
+#include "sentence_metadata.h"
+#include "verbose.h"
+#include "hg.h"
+#include "prob.h"
+#include "inside_outside.h"
+#include "ff_register.h"
+#include "decoder.h"
+#include "filelib.h"
+#include "stringlib.h"
+#include "fdict.h"
+#include "weights.h"
+#include "sparse_vector.h"
+
+using namespace std;
+namespace po = boost::program_options;
+
+bool InitCommandLine(int argc, char** argv, po::variables_map* conf) {
+  po::options_description opts("Configuration options");
+  opts.add_options()
+        ("input_weights,w",po::value<string>(),"Input feature weights file")
+        ("iterations,n",po::value<unsigned>()->default_value(50), "Number of training iterations")
+        ("training_data,t",po::value<string>(),"Training data")
+        ("decoder_config,c",po::value<string>(),"Decoder configuration file");
+  po::options_description clo("Command line options");
+  clo.add_options()
+        ("config", po::value<string>(), "Configuration file")
+        ("help,h", "Print this help message and exit");
+  po::options_description dconfig_options, dcmdline_options;
+  dconfig_options.add(opts);
+  dcmdline_options.add(opts).add(clo);
+  
+  po::store(parse_command_line(argc, argv, dcmdline_options), *conf);
+  if (conf->count("config")) {
+    ifstream config((*conf)["config"].as<string>().c_str());
+    po::store(po::parse_config_file(config, dconfig_options), *conf);
+  }
+  po::notify(*conf);
+
+  if (conf->count("help") || !conf->count("input_weights") || !(conf->count("training_data")) || !conf->count("decoder_config")) {
+    cerr << dcmdline_options << endl;
+    return false;
+  }
+  return true;
+}
+
+void ReadTrainingCorpus(const string& fname, int rank, int size, vector<string>* c) {
+  ReadFile rf(fname);
+  istream& in = *rf.stream();
+  string line;
+  int lc = 0;
+  while(in) {
+    getline(in, line);
+    if (!in) break;
+    if (lc % size == rank) c->push_back(line);
+    ++lc;
+  }
+}
+
+static const double kMINUS_EPSILON = -1e-6;
+
+struct TrainingObserver : public DecoderObserver {
+  void Reset() {
+    acc_grad.clear();
+    acc_obj = 0;
+    total_complete = 0;
+    trg_words = 0;
+  } 
+
+  void SetLocalGradientAndObjective(vector<double>* g, double* o) const {
+    *o = acc_obj;
+    for (SparseVector<double>::const_iterator it = acc_grad.begin(); it != acc_grad.end(); ++it)
+      (*g)[it->first] = it->second;
+  }
+
+  virtual void NotifyDecodingStart(const SentenceMetadata& smeta) {
+    state = 1;
+  }
+
+  // compute model expectations, denominator of objective
+  virtual void NotifyTranslationForest(const SentenceMetadata& smeta, Hypergraph* hg) {
+    assert(state == 1);
+    trg_words += smeta.GetSourceLength();
+    state = 2;
+    SparseVector<prob_t> exps;
+    const prob_t z = InsideOutside<prob_t,
+                                   EdgeProb,
+                                   SparseVector<prob_t>,
+                                   EdgeFeaturesAndProbWeightFunction>(*hg, &exps);
+    exps /= z;
+    for (SparseVector<prob_t>::iterator it = exps.begin(); it != exps.end(); ++it)
+      acc_grad.add_value(it->first, it->second.as_float());
+
+    acc_obj += log(z);
+  }
+
+  // compute "empirical" expectations, numerator of objective
+  virtual void NotifyAlignmentForest(const SentenceMetadata& smeta, Hypergraph* hg) {
+    cerr << "Shouldn't get an alignment forest!\n";
+    abort();
+  }
+
+  virtual void NotifyDecodingComplete(const SentenceMetadata& smeta) {
+    ++total_complete;
+  }
+
+  int total_complete;
+  SparseVector<double> acc_grad;
+  double acc_obj;
+  unsigned trg_words;
+  int state;
+};
+
+void ReadConfig(const string& ini, vector<string>* out) {
+  ReadFile rf(ini);
+  istream& in = *rf.stream();
+  while(in) {
+    string line;
+    getline(in, line);
+    if (!in) continue;
+    out->push_back(line);
+  }
+}
+
+void StoreConfig(const vector<string>& cfg, istringstream* o) {
+  ostringstream os;
+  for (int i = 0; i < cfg.size(); ++i) { os << cfg[i] << endl; }
+  o->str(os.str());
+}
+
+#if 0
+template <typename T>
+struct VectorPlus : public binary_function<vector<T>, vector<T>, vector<T> >  {
+  vector<T> operator()(const vector<int>& a, const vector<int>& b) const {
+    assert(a.size() == b.size());
+    vector<T> v(a.size());
+    transform(a.begin(), a.end(), b.begin(), v.begin(), plus<T>()); 
+    return v;
+  } 
+}; 
+#endif
+
+int main(int argc, char** argv) {
+#ifdef HAVE_MPI
+  mpi::environment env(argc, argv);
+  mpi::communicator world;
+  const int size = world.size(); 
+  const int rank = world.rank();
+#else
+  const int size = 1;
+  const int rank = 0;
+#endif
+  SetSilent(true);  // turn off verbose decoder output
+  register_feature_functions();
+
+  po::variables_map conf;
+  if (!InitCommandLine(argc, argv, &conf)) return 1;
+  const unsigned iterations = conf["iterations"].as<unsigned>();
+
+  // load cdec.ini and set up decoder
+  vector<string> cdec_ini;
+  ReadConfig(conf["decoder_config"].as<string>(), &cdec_ini);
+  istringstream ini;
+  StoreConfig(cdec_ini, &ini);
+  Decoder* decoder = new Decoder(&ini);
+  if (decoder->GetConf()["input"].as<string>() != "-") {
+    cerr << "cdec.ini must not set an input file\n";
+    return 1;
+  }
+
+  // load initial weights
+  if (rank == 0) { cerr << "Loading weights...\n"; }
+  vector<weight_t>& lambdas = decoder->CurrentWeightVector();
+  Weights::InitFromFile(conf["input_weights"].as<string>(), &lambdas);
+  if (rank == 0) { cerr << "Done loading weights.\n"; }
+
+  // freeze feature set (should be optional?)
+  const bool freeze_feature_set = true;
+  if (freeze_feature_set) FD::Freeze();
+
+  const int num_feats = FD::NumFeats();
+  if (rank == 0) cerr << "Number of features: " << num_feats << endl;
+  lambdas.resize(num_feats);
+
+  vector<double> gradient(num_feats, 0.0);
+  vector<double> rcv_grad;
+  rcv_grad.clear();
+  bool converged = false;
+
+  vector<string> corpus, test_corpus;
+  ReadTrainingCorpus(conf["training_data"].as<string>(), rank, size, &corpus);
+  assert(corpus.size() > 0);
+  if (conf.count("test_data"))
+    ReadTrainingCorpus(conf["test_data"].as<string>(), rank, size, &test_corpus);
+
+  // build map from feature id to the accumulator that should normalize
+  boost::unordered_map<std::string, boost::unordered_map<int, double>, boost::hash<std::string> > ccs;
+  vector<boost::unordered_map<int, double>* > cpd_to_acc;
+  if (rank == 0) {
+    cpd_to_acc.resize(num_feats);
+    for (unsigned f = 1; f < num_feats; ++f) {
+      string normalizer;
+      //0 ||| 7 9 ||| Bi:BOS_7=1 Bi:7_9=1 Bi:9_EOS=1 Id:a:7=1 Uni:7=1 Id:b:9=1 Uni:9=1 ||| 0
+      const string& fstr = FD::Convert(f);
+      if (fstr.find("Bi:") == 0) {
+        size_t pos = fstr.rfind('_');
+        if (pos < fstr.size())
+          normalizer = fstr.substr(0, pos);
+      } else if (fstr.find("Id:") == 0) {
+        size_t pos = fstr.rfind(':');
+        if (pos < fstr.size()) {
+          normalizer = "Emit:";
+          normalizer += fstr.substr(pos);
+        }
+      }
+      if (normalizer.size() > 0) {
+        boost::unordered_map<int, double>& acc = ccs[normalizer];
+        cpd_to_acc[f] = &acc;
+      }
+    }
+  }
+
+  TrainingObserver observer;
+  int iteration = 0;
+  while (!converged) {
+    ++iteration;
+    observer.Reset();
+#ifdef HAVE_MPI
+    mpi::timer timer;
+    world.barrier();
+#endif
+    if (rank == 0) {
+      cerr << "Starting decoding... (~" << corpus.size() << " sentences / proc)\n";
+      cerr << "  Testset size: " << test_corpus.size() << " sentences / proc)\n";
+      for(boost::unordered_map<string, boost::unordered_map<int,double>, boost::hash<string> >::iterator it = ccs.begin(); it != ccs.end(); ++it)
+        it->second.clear();
+    }
+    for (int i = 0; i < corpus.size(); ++i)
+      decoder->Decode(corpus[i], &observer);
+    cerr << "  process " << rank << '/' << size << " done\n";
+    fill(gradient.begin(), gradient.end(), 0);
+    double objective = 0;
+    observer.SetLocalGradientAndObjective(&gradient, &objective);
+
+    unsigned total_words = 0;
+#ifdef HAVE_MPI
+    double to = 0;
+    rcv_grad.resize(num_feats, 0.0);
+    mpi::reduce(world, &gradient[0], gradient.size(), &rcv_grad[0], plus<double>(), 0);
+    swap(gradient, rcv_grad);
+    rcv_grad.clear();
+
+    reduce(world, observer.trg_words, total_words, std::plus<unsigned>(), 0);
+    mpi::reduce(world, objective, to, plus<double>(), 0);
+    objective = to;
+#else
+    total_words = observer.trg_words;
+#endif
+    if (rank == 0) {  // run optimizer only on rank=0 node
+      cerr << "TRAINING CORPUS: ln p(x)=" << objective << "\t log_2 p(x) = " << (objective/log(2)) << "\t cross entropy = " << (objective/log(2) / total_words) << "\t ppl = " << pow(2, (-objective/log(2) / total_words)) << endl;
+      for (unsigned f = 1; f < num_feats; ++f) {
+        boost::unordered_map<int, double>* m = cpd_to_acc[f];
+        if (m && gradient[f]) {
+          (*m)[f] += gradient[f];
+        }
+        for(boost::unordered_map<string, boost::unordered_map<int,double>, boost::hash<string> >::iterator it = ccs.begin(); it != ccs.end(); ++it) {
+          const boost::unordered_map<int,double>& ccs = it->second;
+          double z = 0;
+          for (boost::unordered_map<int,double>::const_iterator ci = ccs.begin(); ci != ccs.end(); ++ci)
+            z += ci->second + 1e-09;
+          double lz = log(z);
+          for (boost::unordered_map<int,double>::const_iterator ci = ccs.begin(); ci != ccs.end(); ++ci)
+            lambdas[ci->first] = log(ci->second + 1e-09) - lz;
+        }
+      }
+      Weights::SanityCheck(lambdas);
+      Weights::ShowLargestFeatures(lambdas);
+
+      converged = (iteration == iterations);
+
+      string fname = "weights.cur.gz";
+      if (converged) { fname = "weights.final.gz"; }
+      ostringstream vv;
+      vv << "Objective = " << objective << "  (eval count=" << iteration << ")";
+      const string svv = vv.str();
+      Weights::WriteToFile(fname, lambdas, true, &svv);
+    }  // rank == 0
+    int cint = converged;
+#ifdef HAVE_MPI
+    mpi::broadcast(world, &lambdas[0], lambdas.size(), 0);
+    mpi::broadcast(world, cint, 0);
+    if (rank == 0) { cerr << "  ELAPSED TIME THIS ITERATION=" << timer.elapsed() << endl; }
+#endif
+    converged = cint;
+  }
+  return 0;
+}
+