From 54bcfb835232d190a5ab6f0bd825de8a50dae126 Mon Sep 17 00:00:00 2001
From: Chris Dyer <cdyer@cs.cmu.edu>
Date: Wed, 29 Feb 2012 01:12:40 -0500
Subject: cleanup, mpi-ify lblmodel

---
 utils/corpus_tools.cc | 62 +++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 62 insertions(+)
 create mode 100644 utils/corpus_tools.cc

(limited to 'utils/corpus_tools.cc')

diff --git a/utils/corpus_tools.cc b/utils/corpus_tools.cc
new file mode 100644
index 00000000..a0542b6e
--- /dev/null
+++ b/utils/corpus_tools.cc
@@ -0,0 +1,62 @@
+#include "corpus_tools.h"
+
+#include <iostream>
+
+#include "tdict.h"
+#include "filelib.h"
+#include "verbose.h"
+
+using namespace std;
+
+void CorpusTools::ReadFromFile(const string& filename,
+                           vector<vector<WordID> >* src,
+                           set<WordID>* src_vocab,
+                           vector<vector<WordID> >* trg,
+                           set<WordID>* trg_vocab,
+                           int rank,
+                           int size) {
+  assert(rank >= 0);
+  assert(size > 0);
+  assert(rank < size);
+  if (src) src->clear();
+  if (src_vocab) src_vocab->clear();
+  if (trg) trg->clear();
+  if (trg_vocab) trg_vocab->clear();
+  const int expected_fields = 1 + (trg == NULL ? 0 : 1);
+  if (!SILENT) cerr << "Reading from " << filename << " ...\n";
+  ReadFile rf(filename);
+  istream& in = *rf.stream();
+  string line;
+  int lc = 0;
+  static const WordID kDIV = TD::Convert("|||");
+  vector<WordID> tmp;
+  while(getline(in, line)) {
+    const bool skip = (lc % size != rank);
+    ++lc;
+    if (skip) continue;
+    TD::ConvertSentence(line, &tmp);
+    src->push_back(vector<WordID>());
+    vector<WordID>* d = &src->back();
+    set<WordID>* v = src_vocab;
+    int s = 0;
+    for (unsigned i = 0; i < tmp.size(); ++i) {
+      if (tmp[i] == kDIV) {
+        ++s;
+        if (s > 1) { cerr << "Unexpected format in line " << lc << ": " << line << endl; abort(); }
+        assert(trg);
+        trg->push_back(vector<WordID>());
+        d = &trg->back();
+        v = trg_vocab;
+      } else {
+        d->push_back(tmp[i]);
+        if (v) v->insert(tmp[i]);
+      }
+    }
+    ++s;
+    if (expected_fields != s) {
+      cerr << "Wrong number of fields in line " << lc << ": " << line << endl; abort();
+    }
+  }
+}
+
+
-- 
cgit v1.2.3


From a872f46ce1212703b8bed562c894ea1a932c0746 Mon Sep 17 00:00:00 2001
From: Chris Dyer <prguest11@taipan.cs>
Date: Wed, 29 Feb 2012 07:00:49 +0000
Subject: mpi fixes

---
 training/lbl_model.cc | 54 +++++++++++++++++++++++++++++++++++++++++----------
 utils/corpus_tools.cc | 16 +++++++++------
 2 files changed, 54 insertions(+), 16 deletions(-)

(limited to 'utils/corpus_tools.cc')

diff --git a/training/lbl_model.cc b/training/lbl_model.cc
index def5075a..a46ce33c 100644
--- a/training/lbl_model.cc
+++ b/training/lbl_model.cc
@@ -15,6 +15,7 @@
 #ifdef HAVE_MPI
 #include <boost/mpi/timer.hpp>
 #include <boost/mpi.hpp>
+#include <boost/archive/text_oarchive.hpp>
 namespace mpi = boost::mpi;
 #endif
 #include <boost/math/special_functions/fpclassify.hpp>
@@ -34,12 +35,26 @@ namespace mpi = boost::mpi;
 namespace po = boost::program_options;
 using namespace std;
 
-#define kDIMENSIONS 100
+#define kDIMENSIONS 10
 typedef Eigen::Matrix<double, kDIMENSIONS, 1> RVector;
 typedef Eigen::Matrix<double, 1, kDIMENSIONS> RTVector;
 typedef Eigen::Matrix<double, kDIMENSIONS, kDIMENSIONS> TMatrix;
 vector<RVector> r_src, r_trg;
 
+#if HAVE_MPI
+namespace boost {
+namespace serialization {
+
+template<class Archive>
+void serialize(Archive & ar, RVector & v, const unsigned int version) {
+  for (unsigned i = 0; i < kDIMENSIONS; ++i)
+    ar & v[i];
+}
+
+} // namespace serialization
+} // namespace boost
+#endif
+
 bool InitCommandLine(int argc, char** argv, po::variables_map* conf) {
   po::options_description opts("Configuration options");
   opts.add_options()
@@ -224,7 +239,7 @@ int main(int argc, char** argv) {
     srand(seed);
   }
   
-  TMatrix t;
+  TMatrix t = TMatrix::Zero();
   if (rank == 0) {
     t = TMatrix::Random() / 50.0;
     for (unsigned i = 1; i < r_trg.size(); ++i) {
@@ -241,16 +256,18 @@ int main(int argc, char** argv) {
   TMatrix g = TMatrix::Zero();
   vector<TMatrix> exp_src;
   vector<double> z_src;
-  vector<double> flat_g, flat_t;
+  vector<double> flat_g, flat_t, rcv_grad;
   Flatten(t, &flat_t);
   bool converged = false;
-  // TODO broadcast embeddings
-  for (int iter = 0; !converged && iter < ITERATIONS; ++iter) {
-#ifdef HAVE_MPI
-    mpi::broadcast(world, &flat_t[0], flat_t.size(), 0);
+#if HAVE_MPI
+  mpi::broadcast(world, &flat_t[0], flat_t.size(), 0);
+  mpi::broadcast(world, r_trg, 0);
+  mpi::broadcast(world, r_src, 0);
 #endif
+  cerr << "rank=" << rank << ": " << r_trg[0][4] << endl;
+  for (int iter = 0; !converged && iter < ITERATIONS; ++iter) {
+    if (rank == 0) cerr << "ITERATION " << (iter + 1) << endl;
     Unflatten(flat_t, &t);
-    cerr << "ITERATION " << (iter + 1) << endl;
     double likelihood = 0;
     double denom = 0.0;
     lc = 0;
@@ -350,7 +367,22 @@ int main(int argc, char** argv) {
     if (!SGD) {
       Flatten(g, &flat_g);
       obj = -likelihood;
-      // TODO - reduce gradient
+#if HAVE_MPI
+      rcv_grad.resize(flat_g.size(), 0.0);
+      mpi::reduce(world, &flat_g[0], flat_g.size(), &rcv_grad[0], plus<double>(), 0);
+      swap(flat_g, rcv_grad);
+      rcv_grad.clear();
+
+      double to = 0;
+      mpi::reduce(world, obj, to, plus<double>(), 0);
+      obj = to;
+      double tlh = 0;
+      mpi::reduce(world, likelihood, tlh, plus<double>(), 0);
+      likelihood = tlh;
+      double td = 0;
+      mpi::reduce(world, denom, td, plus<double>(), 0);
+      denom = td;
+#endif
     }
 
     if (rank == 0) {
@@ -376,10 +408,12 @@ int main(int argc, char** argv) {
       }
     }
 #ifdef HAVE_MPI
+    mpi::broadcast(world, &flat_t[0], flat_t.size(), 0);
     mpi::broadcast(world, converged, 0);
 #endif
   }
-  cerr << "TRANSLATION MATRIX:" << endl << t << endl;
+  if (rank == 0)
+    cerr << "TRANSLATION MATRIX:" << endl << t << endl;
   return 0;
 }
 
diff --git a/utils/corpus_tools.cc b/utils/corpus_tools.cc
index a0542b6e..d17785af 100644
--- a/utils/corpus_tools.cc
+++ b/utils/corpus_tools.cc
@@ -33,10 +33,12 @@ void CorpusTools::ReadFromFile(const string& filename,
   while(getline(in, line)) {
     const bool skip = (lc % size != rank);
     ++lc;
-    if (skip) continue;
     TD::ConvertSentence(line, &tmp);
-    src->push_back(vector<WordID>());
-    vector<WordID>* d = &src->back();
+    vector<WordID>* d = NULL;
+    if (!skip) {
+      src->push_back(vector<WordID>());
+      d = &src->back();
+    }
     set<WordID>* v = src_vocab;
     int s = 0;
     for (unsigned i = 0; i < tmp.size(); ++i) {
@@ -44,11 +46,13 @@ void CorpusTools::ReadFromFile(const string& filename,
         ++s;
         if (s > 1) { cerr << "Unexpected format in line " << lc << ": " << line << endl; abort(); }
         assert(trg);
-        trg->push_back(vector<WordID>());
-        d = &trg->back();
+        if (!skip) {
+          trg->push_back(vector<WordID>());
+          d = &trg->back();
+        }
         v = trg_vocab;
       } else {
-        d->push_back(tmp[i]);
+        if (d) d->push_back(tmp[i]);
         if (v) v->insert(tmp[i]);
       }
     }
-- 
cgit v1.2.3