some minor updates, added debug() instructions

BackoffLmKen.cpp

@@ -103,6 +103,7 @@ int BackoffLmKen::GetSentenceIds(WordID *&wid, const string &sentence, bool bos,
       nw++;
     }
   }
+  debug1(" parsing found %d words\n", nw);
 
     // end sentence with EOS ?
   if (eos) {
@@ -111,6 +112,7 @@ int BackoffLmKen::GetSentenceIds(WordID *&wid, const string &sentence, bool bos,
   }
 
   wid = &(wid_vect.front());
+  debug4("* split sent with %d words into %d-grams (bos=%d, eos=%d):\n", nw, ken_ngram->Order(), map_ken2wid[ken_vocab->BeginSentence()], map_ken2wid[ken_vocab->EndSentence()]);
   return nw;
 }
 
@@ -138,7 +140,9 @@ REAL BackoffLmKen::BoffLnPw(char **ctxt, char *w, int req_order)
   for (int i = 0; i < (req_order - 1); i++) {
     ken_ngram->Score(state, ken_vocab->Index(ctxt[i]), out_state);
     state = out_state;
+    debug2(" - context position ken=%d, ken_idx=%d\n", i, ken_vocab->Index(ctxt[i]));
   }
+  debug2(" - predict ken_idx=%d, log10P=%e\n", ken_vocab->Index(w), ken_ngram->Score(state, ken_vocab->Index(w), out_state));
 
     // we need to convert from log_10 to ln
   return M_LN10 * ken_ngram->Score(state, ken_vocab->Index(w), out_state);
@@ -168,7 +172,9 @@ REAL BackoffLmKen::BoffLnPid(REAL *ctxt, WordID predw, int req_order)
   for (int i = 0; i < (req_order - 1); i++) {
     ken_ngram->Score(state, map_cslm2ken[(WordID) ctxt[i]], out_state);
     state = out_state;
+    debug2(" - context position ken=%d, ken_idx=%d\n", i, map_cslm2ken[(WordID) ctxt[i]]);
   }
+  debug3(" - predict cslm_id=%d, ken_idx=%d, log10P=%e\n", predw, map_cslm2ken[predw], ken_ngram->Score(state, map_cslm2ken[predw], out_state));
 
     // we need to convert from log_10 to ln
   return M_LN10 * ken_ngram->Score(state, map_cslm2ken[predw], out_state);
@@ -198,7 +204,9 @@ REAL BackoffLmKen::BoffLnStd(WordID *ctxt, WordID predw, int req_order)
   for (int i = 0; i < (req_order - 1); i++) {
     ken_ngram->Score(state, ctxt[i], out_state);
     state = out_state;
+    debug2(" - context position ken=%d, ken_idx=%d\n", i, ctxt[i]);
   }
+  debug3(" - predict cslm_id=%d, ken_idx=%d, log10P=%e\n", predw, predw, ken_ngram->Score(state, predw, out_state));
 
     // we need to convert from log_10 to ln
   return M_LN10 * ken_ngram->Score(state, predw, out_state);

BackoffLmSri.cpp

@@ -121,6 +121,7 @@ BackoffLmSri::BackoffLmSri(char *p_fname, int p_max_order, const WordList &wlist
   WordList::const_iterator iter = wlist.Begin(), end = wlist.End();
   for (size_t ci=0; iter!=end; iter++, ci++) {
     VocabIndex vi = sri_vocab->getIndex(iter->word);
+    //debug3("'%s' bin=%d -> sri=%d\n", iter->word, ci, vi);
     if (vi == Vocab_None) {
       fprintf(stderr,"word %s not found at pos %zu\n", iter->word, ci );
     }
@@ -156,6 +157,7 @@ int BackoffLmSri::GetSentenceIds(WordID *&wid, const string &sentence, bool bos,
   strcpy(str,sentence.c_str()); // we need to copy since parseWords() modifies the string
   nw = sri_vocab->parseWords(str, vstr, max_words - 1);
   if (nw >= max_words-1) Error("too many words in one hypothesis\n");
+  debug1(" parsing found %d words\n", nw);
 
   int b=0;
     // start sentence with BOS ?
@@ -171,5 +173,6 @@ int BackoffLmSri::GetSentenceIds(WordID *&wid, const string &sentence, bool bos,
   if (eos) wid_table[nw++]=sri_vocab->seIndex();
 
   wid = wid_table;
+  debug4("* split sent with %d words into %d-grams (bos=%d, eos=%d):\n", nw, sri_order, sri_vocab->ssIndex(), sri_vocab->seIndex());
   return nw;
 }

BackoffLmSri.h

@@ -119,8 +119,10 @@ class BackoffLmSri : public BackoffLm {
     for (i=0; i<req_order-1; i++) {	// build context vector in REVERSE order
       int j=req_order-i-2;
       sri_context_idxs[i] = ctxt[j];
+      //debug4(" - context position cslm=%d -> sri=%d, sri_idx=%d word=%s\n", j, i, sri_context_idxs[i], sri_vocab->getWord(sri_context_idxs[i]) );
     }
     sri_context_idxs[i]=Vocab_None; // terminate, this is needed to specify the length of the context
+    //debug3(" - predict cslm_id=%d, sri_idx=%d word=%s\n", predw, predw, sri_vocab->getWord(predw) );
 
 #ifdef DEBUG
     printf(" - SRI %d-gram context: ",req_order);

Blas.h

@@ -151,6 +151,7 @@ inline void call_gemv (REAL *dest, REAL *matrix, REAL *source, REAL *bias,
 	//         m x n
 
 
+	debug0("-mkl- call gemv\n");
 #ifdef BLAS_CUDA
 	COPY(dim_dest,bias,inc,dest,inc);  // TODO: verify
   	GEMV(trans, dim_dest, dim_src, fact, matrix, dim_dest, source, inc, fact, dest, inc);

Data.cpp

@@ -78,6 +78,7 @@ Data::Data(const char *p_fname, Data *other_data, bool use_class)
    sr_wl_idx(-1), tg_wl_idx(-1), current_df(0), idx(-1),
    mem_cdf(NULL), mem_inp(NULL), mem_trg(NULL), input(NULL), target(NULL), aux(NULL)
 {
+  debug0("* constructor Data\n");
   ReadFile(other_data, use_class);
 }
 
@@ -354,6 +355,7 @@ void Data::ReadFile(Data *other_data, bool use_class)
 
 Data::~Data()
 {
+  debug0("* destructor Data\n");
   if (preload) {
     delete [] mem_cdf;
     delete [] mem_inp;
@@ -454,12 +456,14 @@ void Data::Preload()
     int odim1=(*itf)[0]->GetOdim();
 
     while (++n < maxn) {
+      debug1("getting example %d\n",idx);
       mem_cdf[idx] = cdf;
 
       bool ok=false;
       while (!ok) {
           // advance synchronously all factors until ok
         for (vector<DataFile*>::iterator it = (*itf).begin(); it!=(*itf).end(); ++it) {
+          debug1("  next factor %ld\n", it-(*itf).begin());
           if (! (*it)->Next()) (*it)->Rewind(); // TODO: deadlock if file empty
         }
 
@@ -469,12 +473,14 @@ void Data::Preload()
 		ok = (drand48() < (*itf)[0]->GetResamplCoef());
 	}
 
+	debug1("  %s\n", ok ? "keep" : "skip");
       }
 
         // copy all factors sequentially in memory
       REAL *adr_inp=mem_inp+idx*idim;
       REAL *adr_trg=mem_trg+idx*odim;
       for (vector<DataFile*>::iterator it = (*itf).begin(); it!=(*itf).end(); ++it) {
+        debug2("  load factor %ld to address %p\n", it-(*itf).begin(), adr_inp);
         memcpy(adr_inp, (*it)->input, idim1*sizeof(REAL));
 	adr_inp+=idim1;
         if (odim1 > 0) {
@@ -494,6 +500,7 @@ void Data::Preload()
       REAL m=0, *mptr;
       for (e=0, mptr=mem_inp+i; e<idx; e++, mptr+=idim) m+=*mptr;
       m = m/idx; // mean
+      debug2("mean[%d]=%f\n", i, m);
       for (e=0, mptr=mem_inp+i; e<idx; e++, mptr+=idim) *mptr -= m;
     }
   }
@@ -506,6 +513,7 @@ void Data::Preload()
       for (e=0, mptr=mem_inp+i; e<idx; e++, mptr+=idim) { m+=*mptr; m2+=*mptr * *mptr; }
       m = m/idx;  // mean
       m2 = m2/idx - m; // var = 1/n sum_i x_i^2  -  mu^2
+      debug3(" mean, var[%d]=%f %f\n", i, m, m2);
       if (m2>0)
         for (e=0, mptr=mem_inp+i; e<idx; e++, mptr+=idim)
           *mptr = (*mptr - m) / m2;
@@ -530,6 +538,7 @@ void Data::Preload()
 
 void Data::Rewind()
 {
+  debug0("** Data::Rewind()\n");
   if (preload) {
        // clear all data, resample and shuffle again
     Preload();
@@ -541,6 +550,7 @@ void Data::Rewind()
         (*it)->Rewind();
   }
   idx = -1;
+  debug0("** Data::Rewind() done\n");
 }
 
 /**************************

DataAscii.cpp

@@ -31,6 +31,7 @@ const char* DATA_FILE_ASCII="DataAscii";
 DataAscii::DataAscii(char *p_prefix, ifstream &ifs, int p_aux_dim, const string& p_aux_ext, int p_nb_SentSc, const string& p_SentSc_ext,int p_betweenSentCtxt , DataAscii *prev_df)
  : DataFile::DataFile(p_prefix, ifs, p_aux_dim, p_aux_ext, p_nb_SentSc, p_SentSc_ext, p_betweenSentCtxt, prev_df)
 {
+  debug0("** constructor DataAscii\n");
 
   char full_fname[max_word_len]="";
 
@@ -70,6 +71,7 @@ DataAscii::DataAscii(char *p_prefix, ifstream &ifs, int p_aux_dim, const string&
 
 DataAscii::~DataAscii()
 {
+  debug0("** destructor DataAscii\n");
   dfs.close();
   if (idim>0) delete [] input;
   if (odim>0) delete [] target_vect;
@@ -82,6 +84,7 @@ DataAscii::~DataAscii()
 
 void DataAscii::Rewind()
 {
+  debug0("*** DataAscii::Rewind()\n");
   dfs.seekg(0, dfs.beg);
   char buf[DATA_LINE_LEN];
   dfs.getline(buf,DATA_LINE_LEN);
@@ -95,6 +98,7 @@ void DataAscii::Rewind()
 
 bool DataAscii::Next()
 {
+//  debug0("*** DataAscii::Next() "); cout<<idx<< ", fpos=" << dfs.tellg() << endl;
   char line[DATA_LINE_LEN];
   dfs.getline(line, DATA_LINE_LEN);
   if (dfs.eof()) return false;

DataAsciiClass.cpp

@@ -31,6 +31,7 @@ const char* DATA_FILE_ASCIICLASS="DataAsciiClass";
 DataAsciiClass::DataAsciiClass(char *p_prefix, ifstream &ifs, int p_aux_dim, const string& p_aux_ext, int p_nb_SentSc, const string& p_SentSc_ext,int p_betweenSentCtxt, DataAsciiClass *prev_df)
  : DataAscii::DataAscii(p_prefix, ifs, p_aux_dim, p_aux_ext, p_nb_SentSc, p_SentSc_ext, p_betweenSentCtxt, prev_df)
 {
+  debug0("** constructor DataAsciiClass\n");
 
   if (prev_df) {
     tgt0=prev_df->tgt0;
@@ -50,6 +51,7 @@ DataAsciiClass::DataAsciiClass(char *p_prefix, ifstream &ifs, int p_aux_dim, con
 
 bool DataAsciiClass::Next()
 {
+//  debug0("*** DataAsciiClass::Next() "); cout<<idx<< ", fpos=" << dfs.tellg() << endl;
   char line[DATA_LINE_LEN];
   dfs.getline(line, DATA_LINE_LEN);
   if (dfs.eof()) return false;

DataFile.cpp

@@ -36,6 +36,7 @@ DataFile::DataFile(char *p_path_prefix, ifstream &ifs, int p_aux_dim, const stri
    nb_SentSc(p_nb_SentSc), betweenSent_ctxt(p_betweenSentCtxt), SentSc_ext(p_SentSc_ext), 
    idx(-1), input(NULL), target_vect(NULL), aux(NULL), target_id(0)
 {
+  debug0("** constructor DataFile\n");
   char p_fname[DATA_LINE_LEN];
 
   ifs >> p_fname;
@@ -69,6 +70,7 @@ DataFile::DataFile(char *p_path_prefix, char *p_fname, const float p_rcoeff)
  : idim(0), odim(0), auxdim(0), nbex(0), resampl_coeff(p_rcoeff), path_prefix(p_path_prefix), fname(NULL),
    idx(-1), input(NULL), target_vect(NULL), aux(NULL), target_id(0)
 {
+  debug0("** constructor DataFile with fname\n");
   if (NULL != p_fname)
     fname = strdup(p_fname);
 
@@ -78,6 +80,7 @@ DataFile::DataFile(char *p_path_prefix, char *p_fname, const float p_rcoeff)
 
 DataFile::~DataFile()
 {
+  debug0("** destructor DataFile\n");
   if (fname) free(fname);
   if (aux_fs.is_open())
     aux_fs.close();
@@ -153,6 +156,7 @@ int DataFile::Info(const char *txt)
 
 void DataFile::Rewind()
 {
+  debug0("*** DataFile::Rewind()\n");
   Error("DataFile::Rewind() should be overriden");
 }
 
@@ -174,6 +178,7 @@ bool DataFile::Next()
 
 int DataFile::Resampl()
 {
+  //debug0("*** DataFile::Resampl()\n");
   bool ok=false;
 
   while (!ok) {
@@ -184,6 +189,7 @@ int DataFile::Resampl()
 //cout << " ok=" << ok << endl;
   }
 
+  //debug0("*** DataFile::Resampl() end\n");
   return idx;
 }
 

DataMnist.cpp

@@ -48,6 +48,7 @@ uint DataMnist::read_iswap(int fd) {
 
   // swap integer Big Endian -> little Endian
   ps[0]=pi[3]; ps[1]=pi[2]; ps[2]=pi[1]; ps[3]=pi[0];
+  debug2("read=%4x, swap=%4x\n", i, s);
 
   return s;
 }
@@ -59,6 +60,7 @@ uint DataMnist::read_iswap(int fd) {
 DataMnist::DataMnist(char *p_prefix, ifstream &ifs, int p_aux_dim, const string& p_aux_ext, int p_nb_SentSc, string& p_SentSc_ext,int p_betweenSentCtxt, DataMnist *prev_df)
  : DataFile::DataFile(p_prefix, ifs, p_aux_dim, p_aux_ext, p_nb_SentSc, p_SentSc_ext, p_betweenSentCtxt, prev_df)
 {
+  debug0("** constructor DataMnist\n");
   char full_fname[max_word_len]="";
 
   printf(" - %s: MNIST data ", fname); fflush(stdout);
@@ -125,6 +127,7 @@ DataMnist::DataMnist(char *p_prefix, ifstream &ifs, int p_aux_dim, const string&
 
 DataMnist::~DataMnist()
 {
+  debug0("** destructor DataMnist\n");
   close(dfd);
   close(lfd);
   if (idim>0) { delete [] input; delete [] ubuf; }
@@ -139,6 +142,7 @@ DataMnist::~DataMnist()
 
 void DataMnist::Rewind()
 {
+  debug0("*** DataMnist::Rewind()\n");
   lseek(dfd, 16, SEEK_SET);
   lseek(lfd, 8, SEEK_SET);
   if (aux_fs.is_open())
@@ -151,6 +155,7 @@ void DataMnist::Rewind()
 
 bool DataMnist::Next()
 {
+//  debug0("*** DataMnist::Next() "); cout<<idx<< " << endl;
 
     // read next image 
   int t=idim*sizeof(unsigned char);
@@ -182,6 +187,7 @@ bool DataMnist::Next()
     sprintf(msg, "no examples left in class file %s", cl_fname);
     Error(msg);
   }
+  debug1("class %d\n", ubuf[0]);
   target_id = (int) ubuf[0];
   if (target_id>=odim) {
     ErrorN("example %lu has a target of %d, but we have only %d classes\n", idx+1, target_id, odim);

DataNgramBin.cpp

@@ -166,6 +166,7 @@ DataNgramBin::DataNgramBin(char *p_prefix, ifstream &ifs, int p_aux_dim, const s
  : DataFile::DataFile(p_prefix, ifs, p_aux_dim, p_aux_ext, p_nb_SentSc, p_SentSc_ext,p_betweenSentCtxt, prev_df),
   order(0), tgpos(0), eospos(0), mode(0), nbw(0), nbs(0), nbu(0), nbi(0)
 {
+  debug0("*** constructor DataNgramBin\n");
     // DataNgramBin <file_name> <resampl_coeff> <order> [<tgpos>] <mode>
 
     // parse addtl params -> 
@@ -197,6 +198,7 @@ DataNgramBin::DataNgramBin(char *p_fname, float p_rcoeff, int p_order)
   : DataFile::DataFile(NULL, p_fname, p_rcoeff),
     order(p_order), tgpos(p_order - 1), eospos(0), mode(3), nbw(0), nbs(0), nbu(0), nbi(0)
 {
+  debug0("*** constructor DataNgramBin with fname\n");
 
   do_constructor_work();
     // skip counting for efficieny reasons
@@ -209,6 +211,7 @@ DataNgramBin::DataNgramBin(char *p_fname, float p_rcoeff, int p_order, int p_tgp
   : DataFile::DataFile(NULL, p_fname, p_rcoeff),
     order(p_order), tgpos(p_tgpos), mode(p_mode), nbw(0), nbs(0), nbu(0), nbi(0)
 {
+  debug0("*** constructor DataNgramBin with fname\n");
   if (tgpos<0 || tgpos>=order)
     ErrorN("wrong value of target position: %d not in [0,%d]\n",tgpos,order-1);
 
@@ -221,6 +224,7 @@ DataNgramBin::DataNgramBin(char *p_fname, float p_rcoeff, int p_order, int p_tgp
 
 DataNgramBin::~DataNgramBin()
 {
+  debug0("*** destructor DataNgramBin\n");
 
   close(fd);
   if (idim>0) {
@@ -238,6 +242,7 @@ DataNgramBin::~DataNgramBin()
  * */
 bool DataNgramBin::Next()
 {
+  //debug0("*** DataNgramBin::Next() \n");
   bool ok=false;
   string line_sc;
   int i;
@@ -353,12 +358,14 @@ bool DataNgramBin::Next()
 
 void DataNgramBin::Rewind()
 {
+  debug0("*** DataNgramBin::Rewind()\n");
   lseek(fd, header_len, SEEK_SET);
   if (aux_fs.is_open())
     aux_fs.seekg(0, aux_fs.beg);
   if(SentSc_fs.is_open())
     SentSc_fs.seekg(0, aux_fs.beg);
   idx=-1;
+  debug0("*** DataNgramBin::Rewind() done\n");
     // initialize read buffer
   buf_n=0; buf_pos=-1;
   eospos = 0;