maj

maca_trans_parser/src/train_dynet.cpp

@@ -133,6 +133,9 @@ int main(int argc, char** argv)
 	
 	// Build model -----------------------------------------------------------------------------------
 	ParameterCollection model;
+	LookupParameter table_embedding = model.add_lookup_parameters(/*nb of words in cff (this is an example for testing)*/5027, 
+		/*dim of the vector embedding*/{300});
+	
 	// Use Adam optimizer
 	AdamTrainer trainer(model);
 	trainer.clip_threshold *= batch_size;
@@ -196,11 +199,17 @@ int main(int argc, char** argv)
 			cur_labels = vector<unsigned int>(bsize);
 			for (unsigned int idx = 0; idx < bsize; ++idx) 
 			{
-				cur_batch[idx] = input(cg, {5}, cff_train[id + idx]);
+				vector<Expression> vect_expr(NB_FEATS);
+				for(unsigned int i=0; i<NB_FEATS; ++i)
+					vect_expr[i] = const_lookup(cg, table_embedding, static_cast<unsigned>(cff_train[id + idx][i]) );
+				
+				//cur_batch[idx] = input(cg, {5}, cff_train[id + idx]);
+				cur_batch[idx] = concatenate(vect_expr, 300);
 				cur_labels[idx] = cff_train_labels[id + idx];
 			}
 			// Reshape as batch (not very intuitive yet)
-			Expression x_batch = reshape(concatenate_cols(cur_batch), Dim({5}, bsize));
+			/**Trouver comment faire les batchs avec les embeddings !*/
+			/**Expression x_batch = reshape(concatenate_cols(cur_batch), Dim({5}, bsize));*/
 			
 			// Get negative log likelihood on batch
 			Expression loss_expr = nn.get_nll(x_batch, cur_labels, cg);