externally call each script for each cross-validation iteration

build_data.py

@@ -4,7 +4,7 @@
     export_trimmed_word2vec_vectors, get_processing_word
 
 
-def main(**kwargs):
+def main():
     """Procedure to build data
 
     You MUST RUN this procedure. It iterates over the whole dataset (train,
@@ -21,7 +21,7 @@ def main(**kwargs):
     """
     # get config and processing of words
 
-    config = Config(**dict(kwargs, load=False))
+    config = Config(load=False)
     processing_word = get_processing_word(lowercase=False)
 
     # Generators

cross-validation.py

@@ -1,9 +1,10 @@
 import codecs
 import numpy as np
+from pip._vendor.progress import counter
 from sklearn.model_selection import ShuffleSplit
-from build_data import main as build
-from train import main as train
-from evaluate import main as eval
+from shutil import copyfile
+import subprocess
+import sys
 
 """
 
@@ -15,6 +16,7 @@
 sentences = []
 sentence = []
 
+
 def write(path, sts):
     f = open(path, 'w')
     for s in sts:
@@ -24,6 +26,7 @@ def write(path, sts):
         f.write('\n')
     f.close()
 
+
 for line in codecs.open('data/celikkaya2013/input.txt', 'r', 'utf8'):
     line = line.rstrip()
     if not line:
@@ -74,21 +77,41 @@ def write(path, sts):
 
     print("Created train, dev and test sets of iteration: %i" % count)
 
+    copyfile(filename_train, 'data/celikkaya2013/tr.train.iobes')
+    copyfile(filename_dev, 'data/celikkaya2013/tr.testa.iobes')
+    copyfile(filename_test, 'data/celikkaya2013/tr.testb.iobes')
+
     # Build
-    kwargs = {
-        "filename_train": filename_train,
-        "filename_dev": filename_dev,
-        "filename_test": filename_test
-    }
-    build(**kwargs)
+    with open('output.log', 'a+') as out:
+        out.write("Beginning building for CV iteration:{}".format(str(counter)))
+        p = subprocess.Popen('python3 build_data.py', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
+        for line in p.stdout.readlines():
+            out.write(line + '\n')
+        retval = p.wait()
+        out.write("Finished building. exit code:{}\n".format(str(retval)))
+        out.flush()
     print("Built model.")
 
     # Train
-    train(**kwargs)
+    with open('output.log', 'a+') as out:
+        out.write("Beginning training for CV iteration:{}".format(str(counter)))
+        p = subprocess.Popen('python3 train.py', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
+        for line in p.stdout.readlines():
+            out.write(line + '\n')
+        retval = p.wait()
+        out.write("Finished training. exit code:{}\n".format(str(retval)))
+        out.flush()
     print("Trained model.")
 
     # Evaluate
-    eval(interactive=False, **kwargs)
+    with open('output.log', 'a+') as out:
+        out.write("Beginning eval for CV iteration:{}".format(str(counter)))
+        p = subprocess.Popen('python3 evaluate.py', shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
+        for line in p.stdout.readlines():
+            out.write(line + '\n')
+        retval = p.wait()
+        out.write("Finished eval. exit code:{}\n".format(str(retval)))
+        out.flush()
     print("Evaluated model.")
 
     count -= 1

data/extract_word_list.py

@@ -4,12 +4,26 @@
 # https://groups.google.com/forum/#!topic/gensim/JRYhCt10AMw
 # https://radimrehurek.com/gensim/models/keyedvectors.html
 # https://stackoverflow.com/questions/44693241/how-to-extract-a-word-vector-from-the-google-pre-trained-model-for-word2vec
-
-model = KeyedVectors.load_word2vec_format('../other/TweetNER/TweetNER/newfile.txt', binary=True, limit=200000)
+"""
+model = KeyedVectors.load_word2vec_format('../other/TweetNER/TweetNER/newfile.txt', binary=True, limit=200000, encoding='utf-8', unicode_errors='ignore')
 print(model['*UNKNOWN*'])
 with open('../other/TweetNER/TweetNER/word-list.txt', 'w+', encoding='utf-8') as f:
     for key in model.vocab.keys():
         f.write(key + "\n")
+"""
+
+print("Building vocab...")
+count = 0
+with open('../other/TweetNER/TweetNER/newfile.txt', encoding='utf-8') as f:
+    with open('../other/TweetNER/TweetNER/word-list.txt', 'w+', encoding='utf-8') as f2:
+        for line in f:
+            if count == 200001:
+                break
+            word = line.strip().split(' ')[0]
+            f2.write(word + "\n")
+            count += 1
+
+
 
 """
 with open('../other/TweetNER/TweetNER/pretrained.txt', 'r', encoding='utf-8') as f:

evaluate.py

@@ -66,9 +66,9 @@ def interactive_shell(model):
             model.logger.info(seq)
 
 
-def main(interactive=True, **kwargs):
+def main(interactive=False):
     # create instance of config
-    config = Config(**kwargs)
+    config = Config()
 
     # build model
     model = NERModel(config)

model/config.py

@@ -7,18 +7,14 @@
 
 
 class Config():
-    def __init__(self, load=True, **kwargs):
+    def __init__(self, load=True):
         """Initialize hyperparameters and load vocabs
 
         Args:
             load_embeddings: (bool) if True, load embeddings into
                 np array, else None
 
         """
-        # update default parameters!
-        for key, value in kwargs.items():
-            setattr(self, key, value)
-
         # directory for training outputs
         if not os.path.exists(self.dir_output):
             os.makedirs(self.dir_output)

train.py

@@ -3,9 +3,9 @@
 from model.config import Config
 
 
-def main(**kwargs):
+def main():
     # create instance of config
-    config = Config(**kwargs)
+    config = Config()
 
     # build model
     model = NERModel(config)
@@ -22,5 +22,6 @@ def main(**kwargs):
     # train model
     model.train(train, dev)
 
+
 if __name__ == "__main__":
     main()