Minor fixes to seq2seq logging

.idea/deployment.xml

-<?xml version="1.0" encoding="UTF-8"?>
-<project version="4">
-  <component name="PublishConfigData" autoUpload="Always" serverName="guppi" createEmptyFolders="true" persistUploadOnCheckin="false" autoUploadExternalChanges="true">
-    <serverData>
-      <paths name="guppi">
-        <serverdata>
-          <mappings>
-            <mapping deploy="/projects/clef18" local="$PROJECT_DIR$" web="/" />
-          </mappings>
-        </serverdata>
-      </paths>
-      <paths name="sonic">
-        <serverdata>
-          <mappings>
-            <mapping local="$PROJECT_DIR$" web="/" />
-          </mappings>
-        </serverdata>
-      </paths>
-      <paths name="vistTriton">
-        <serverdata>
-          <mappings>
-            <mapping deploy="/clef18" local="$PROJECT_DIR$" web="/" />
-          </mappings>
-        </serverdata>
-      </paths>
-    </serverData>
-    <option name="myAutoUpload" value="ALWAYS" />
-  </component>
-</project>
\ No newline at end of file

code_jurica/classificationICD10.py

@@ -2,6 +2,7 @@
 # experiment=Experiment(api_key="hSd9vTj0EfMu72569YnVEvtvj")
 
 # from loader import *
+import keras.backend as K
 from util import *
 import numpy as np
 import random
@@ -27,14 +28,17 @@ config=tf.ConfigProto()
 config.gpu_options.allow_growth=True
 config.gpu_options.allocator_type='BFC'
 
+sess = tf.Session(graph=tf.get_default_graph(), config=config)
+K.set_session(sess)
+
 callbacks_list=[
     EarlyStopping(
         monitor='val_loss',
-        patience=2,
+        patience=50,
     ),
     ModelCheckpoint(
         filepath='models/icd10Classification.h5',
-        monitor='val_loss',
+        monitor='loss',
         save_best_only=True,
     ),
     CSVLogger(
@@ -44,7 +48,7 @@ callbacks_list=[
 ]
 
 latent_dim = 512
-epochs = 100
+epochs = 500
 batch_size = 1000
 
 tokenizer=TokenizePreprocessor()
@@ -115,7 +119,8 @@ try:
           batch_size=batch_size,
           epochs=epochs,
           callbacks=callbacks_list,
-          validation_split=0.25
+          validation_data=[X_test, Y_test]
+          #verbose=0
     )
 
 except Exception as e:

code_jurica/seq2seq.py

@@ -48,10 +48,10 @@ except OSError:
     from classificationICD10 import *
     icd10_model = keras_load_model('models/icd10Classification_attention_extended.h5')
 
-with open('models/icd10_tokenizer_extended.p', 'rb') as handle:
+with open('models/icd10_tokenizer.p', 'rb') as handle:
     icd10Tokenizer = pickle.load(handle)
 
-with open('models/icd10_mappings_extended.p', 'rb') as handle:
+with open('models/icd10_mappings.p', 'rb') as handle:
     encoded_Y = pickle.load(handle)
 # LOAD ICD 10 CLASSIFICATION MODEL
 

code_jurica/seq2seq_attention.py

@@ -45,17 +45,17 @@ except OSError:
     from classificationICD10 import *
     icd10_model = keras_load_model('models/icd10Classification_attention_extended.h5')
 
-with open('models/icd10_tokenizer_extended.p', 'rb') as handle:
+with open('models/icd10_tokenizer.p', 'rb') as handle:
     icd10Tokenizer = pickle.load(handle)
 
-with open('models/icd10_mappings_extended.p', 'rb') as handle:
+with open('models/icd10_mappings.p', 'rb') as handle:
     encoded_Y = pickle.load(handle)
 # LOAD ICD 10 CLASSIFICATION MODEL
 
 callbacks_list = [
     EarlyStopping(
         monitor='val_loss',
-        patience=2,
+        patience=5,
         # min_delta=0.001
     ),
     ModelCheckpoint(
@@ -71,7 +71,7 @@ callbacks_list = [
 
 latent_dim = 256
 batch_size = 400
-epochs = 1
+epochs = 2000
 
 train_data_generator = KerasBatchGenerator(batch_size,
                                            source_train,
@@ -237,7 +237,7 @@ run_pipeline_prediction(source_val, decode_seq, icd10_model, encoded_Y, labels_v
                         source_kerasTokenizer, source_max_sequence_tokenizer,
                         icd10Tokenizer, max_icd10_length,
                         target_index_to_word_dict, target_val,
-                        'logs/seq2seq')
+                        'logs/seq2seq-att')
 
 # y_true = []
 # y_pred = []

code_jurica/seq2seq_base.py

@@ -55,13 +55,13 @@ def run_pipeline_prediction(sentences: List[str], decode_seq_fnc: Callable, icd1
         pred_text = " ".join(translated_sent)
         pred_texts.append(pred_text)
 
-        pred_indexes = " ".join(translated_index)
+        pred_indexes = " ".join([str(i) for i in translated_index])
         pred_ids.append(pred_indexes)
 
         gold_indexes = np.trim_zeros(gold_target_indexes[seq_index], 'b')[1:-1]
-        gold_ids.append(" ".join(gold_indexes))
+        gold_ids.append(" ".join([str(i) for i in gold_indexes]))
 
-        gold_text = " ".join([target_index_to_word_dict[x] for x in gold_indexes])
+        gold_text = " ".join([target_index_to_word_dict[x] for x in gold_indexes if x in target_index_to_word_dict])
         gold_texts.append(gold_text)
 
         print('Target indexes:', gold_indexes)