updated maca_trans_parser/src/context.h to accomodate for options specific to maca_trans_parser_nn

CMakeLists.txt

@@ -8,6 +8,10 @@ SET(CMAKE_C_COMPILER g++)
 SET(CMAKE_CXX_COMPILER g++)
 
 
+SET( CMAKE_CXX_FLAGS  "${CMAKE_CXX_FLAGS} -Ofast -DUSE_CBLAS")
+SET( CMAKE_EXE_LINKER_FLAGS  "${CMAKE_EXE_LINKER_FLAGS} -lm -lopenblas" )
+
+
 if (${CMAKE_C_COMPILER_VERSION} VERSION_LESS 5.3)
 	set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -std=gnu11")
 	# better, but needs CMake >= 3.0

maca_trans_parser/src/context.h

@@ -24,6 +24,14 @@
 #define DEFAULT_VOCABS_TAGPARSER_FILENAME "maca_trans_tagparser.vocab" 
 #define DEFAULT_MODEL_TAGPARSER_FILENAME  "maca_trans_tagparser.model" 
 
+#define DEFAULT_MULTI_COL_DESC_PARSER_NN_FILENAME "maca_trans_parser_nn.mcd" 
+#define DEFAULT_FEATURES_MODEL_PARSER_NN_FILENAME "maca_trans_parser_nn.fm" 
+#define DEFAULT_VOCABS_PARSER_NN_FILENAME "maca_trans_parser_nn.vocab" 
+#define DEFAULT_MODEL_PARSER_NN_FILENAME  "maca_trans_parser_nn.weights" 
+#define DEFAULT_JSON_PARSER_NN_FILENAME  "maca_trans_parser_nn.json" 
+
+
+
 #define DEFAULT_F2P_FILENAME "fP" 
 #define DEFAULT_FPLM_FILENAME "fplm" 
 

maca_trans_parser/src/maca_trans_parser_nn.cc

@@ -44,41 +44,40 @@ void maca_trans_parser_nn_check_options(context *ctx){
   }
 }
 
-
-
 void set_linguistic_resources_filenames_parser(context *ctx)
 {
   char absolute_filename[500];
 
-  if(!ctx->perc_model_filename){
+  if(!ctx->dnn_model_filename){
     strcpy(absolute_filename, ctx->maca_data_path);
-    strcat(absolute_filename, DEFAULT_MODEL_FILENAME);
-    ctx->perc_model_filename = strdup(absolute_filename);
+    strcat(absolute_filename, DEFAULT_MODEL_PARSER_NN_FILENAME);
+    ctx->dnn_model_filename = strdup(absolute_filename);
   }
 
-  if(!ctx->vocabs_filename){
+  if(!ctx->json_filename){
     strcpy(absolute_filename, ctx->maca_data_path);
-    strcat(absolute_filename, DEFAULT_VOCABS_FILENAME);
-    ctx->vocabs_filename = strdup(absolute_filename);
+    strcat(absolute_filename, DEFAULT_JSON_PARSER_NN_FILENAME);
+    ctx->json_filename = strdup(absolute_filename);
   }
 
-  /*  if(!ctx->mcd_filename){
+  if(!ctx->vocabs_filename){
     strcpy(absolute_filename, ctx->maca_data_path);
-    strcat(absolute_filename, DEFAULT_MULTI_COL_DESC_FILENAME);
-    ctx->mcd_filename = strdup(absolute_filename);
-    }*/
+    strcat(absolute_filename, DEFAULT_VOCABS_PARSER_NN_FILENAME);
+    ctx->vocabs_filename = strdup(absolute_filename);
+  }
 
   if(!ctx->features_model_filename){
     strcpy(absolute_filename, ctx->maca_data_path);
-    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_FILENAME);
+    strcat(absolute_filename, DEFAULT_FEATURES_MODEL_PARSER_NN_FILENAME);
     ctx->features_model_filename = strdup(absolute_filename);
   }
 
   if(ctx->verbose){
-    fprintf(stderr, "perc_model_filename = %s\n", ctx->perc_model_filename);
-    fprintf(stderr, "vocabs_filename = %s\n", ctx->vocabs_filename);
-    fprintf(stderr, "mcd_filename = %s\n", ctx->mcd_filename);
-    fprintf(stderr, "perc_features_model_filename = %s\n", ctx->features_model_filename);
+    fprintf(stderr, "dnn_model = %s\n",      ctx->dnn_model_filename);
+    fprintf(stderr, "json = %s\n",           ctx->json_filename);
+    fprintf(stderr, "vocabs %s\n",           ctx->vocabs_filename);
+    fprintf(stderr, "mcd = %s\n",            ctx->mcd_filename);
+    fprintf(stderr, "features_model = %s\n", ctx->features_model_filename);
   }
 }
 
@@ -184,16 +183,15 @@ std::vector<Matrix<float> > config2keras_vec(feat_model *fm, config *c)
 void simple_decoder_parser_arc_eager_nn(context *ctx, Model &model)
 {
   FILE *f = (ctx->input_filename)? myfopen(ctx->input_filename, "r") : stdin;
-  feature_table *ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);
+  //  feature_table *ft = feature_table_load(ctx->perc_model_filename, ctx->verbose);
   int root_label;
   int mvt_code;
   int mvt_type;
   int mvt_label;
-  feat_vec *fv = feat_vec_new(10);
   config *c = NULL;
   int result;
   std::vector<Matrix<float> > keras_vec;
-  
+  int sentence_nb = 0;  
   root_label = dico_string2int(ctx->dico_labels, ctx->root_label);
   if(root_label == -1) root_label = 0;
   
@@ -223,11 +221,7 @@ void simple_decoder_parser_arc_eager_nn(context *ctx, Model &model)
       Matrix<float> argmax = y[0].argmax();
       mvt_code = argmax.at(0, 0);
 
-      // printf("mvt code = %d\n", mvt_code);
-      
-      /* config2feat_vec_cff(ctx->features_model, c, ctx->d_perceptron_features, fv, LOOKUP_MODE); */
-      /* mvt_code = feature_table_argmax(fv, ft, &max); */
-
+      //      fprintf(stderr,"mvt code = %d\n", mvt_code);
       mvt_type = movement_parser_type(mvt_code);
       mvt_label = movement_parser_label(mvt_code);
             
@@ -247,6 +241,9 @@ void simple_decoder_parser_arc_eager_nn(context *ctx, Model &model)
 	break;
       case MVT_PARSER_EOS:
 	result = movement_parser_eos(c);
+	sentence_nb++;
+	if((sentence_nb % 2) == 0)
+	  fprintf(stderr, "\rsentence %d", sentence_nb);
 	break;
       case MVT_PARSER_SHIFT:
 	result = movement_parser_shift(c);
@@ -262,14 +259,12 @@ void simple_decoder_parser_arc_eager_nn(context *ctx, Model &model)
 	}
       }
     }
-  }
     
-  if(!ctx->trace_mode)
+  }
+  fprintf(stderr, "\n");  
   print_word_buffer(c, ctx->dico_labels, ctx->mcd_struct);
   
   config_free(c); 
-  feat_vec_free(fv);
-  feature_table_free(ft);
   if(ctx->input_filename)
     fclose(f);
 }
@@ -297,10 +292,9 @@ int main(int argc, char *argv[])
     fprintf(stderr, "cannot find label names\n");
     return 1;
   }
-
   simple_decoder_parser_arc_eager_nn(ctx, model);
   
-  context_free(ctx);
+  //  context_free(ctx);
   return 0;
 }