isolating changes

mimick/make_dataset.py

@@ -72,7 +72,6 @@ def read_text_embs(files):
             in_vocab += 1
         training_instances.append(Instance(charseq(word, c2i), emb))
     training_char_count = len(c2i)
-
     print "Total in Embeddings vocabulary:", len(words)
     print "Training set character count: ", training_char_count
 

mimick/model.py

@@ -199,8 +199,7 @@ def old_save(self, file_name):
         members_to_save.append(self.lstm_to_rep_bias)
         members_to_save.append(self.mlp_out)
         members_to_save.append(self.mlp_out_bias)
-        #self.model.save(file_name, members_to_save)
-        self.model.save(file_name)
+        self.model.save(file_name, members_to_save)
 
         # character mapping saved separately
         cPickle.dump(self.c2i, open(file_name[:-4] + '.c2i', 'w'))
@@ -232,7 +231,6 @@ def dist(instance, vec):
     # Argument parsing
     # ===-----------------------------------------------------------------------===
     parser = argparse.ArgumentParser()
-
     parser.add_argument("--dataset", required=True, help=".pkl file to use")
     parser.add_argument("--vocab", required=True, help="total vocab to output")
     parser.add_argument("--output", help="file with all embeddings")
@@ -300,7 +298,6 @@ def dist(instance, vec):
 
     # Load words to write
     vocab_words = {}
-
     if populate_test_insts_from_vocab:
         train_words = [wordify(w, i2c) for w in training_instances]
     with codecs.open(options.vocab, "r", "utf-8") as vocab_file:
@@ -384,7 +381,6 @@ def dist(instance, vec):
 
         root_logger.info("\n")
         root_logger.info("Epoch {} complete".format(epoch + 1))
-
         # here used to be a learning rate update, no longer supported in dynet 2.0
         print trainer.status()
 
@@ -435,8 +431,7 @@ def dist(instance, vec):
                 if rand < showcase_size:
                     showcase[rand] = word
 
-    if (len(test_instances) > 0): root_logger.info("Average norm for trained: {}".format(inferred_vec_norms / len(test_instances)))
-
+    root_logger.info("Average norm for trained: {}".format(inferred_vec_norms / len(test_instances)))
 
     if options.debug:
         similar_words = {}

model.py

@@ -5,7 +5,7 @@
 from collections import Counter
 from _collections import defaultdict
 from evaluate_morphotags import Evaluator
-from sys import maxsize
+from sys import maxint
 
 import collections
 import argparse
@@ -14,7 +14,6 @@
 import logging
 import progressbar
 import os
-import codecs
 import dynet as dy
 import numpy as np
 
@@ -97,19 +96,6 @@ def __init__(self, tagset_sizes, num_lstm_layers, hidden_dim, word_embeddings, n
             self.mlp_out[att] = self.model.add_parameters((set_size, set_size), name=att+"O")
             self.mlp_out_bias[att] = self.model.add_parameters(set_size, name=att+"Ob")
 
-
-    def dump_embeddings(self, filename):
-        with codecs.open(filename, "w", "utf-8") as writer:
-            #writer.write("{} {}\n".format(self.vocab_size, self.word_embedding_dim))
-            writer.write("{} {}\n".format(self.words_lookup.shape()[0], self.words_lookup.shape()[1]))
-            for w,i in w2i.items():
-                wemb = dy.lookup(self.words_lookup, i, update=self.we_update)
-                writer.write(w + " ")
-                for i in wemb.npvalue():
-                    writer.write(str(i) + " ")
-                writer.write("\n")
-
-
     def word_rep(self, word, char_ids):
         '''
         :param word: index of word in lookup table
@@ -126,17 +112,10 @@ def word_rep(self, word, char_ids):
     def build_tagging_graph(self, sentence, word_chars):
         dy.renew_cg()
 
-<<<<<<< HEAD
-        if word_chars is not None:
-            embeddings = [self.word_rep(w, chars) for w, chars in zip(sentence, word_chars)]
-        else:
-            embeddings = [self.word_rep(w, word_chars) for w in sentence]
-=======
         if word_chars == None:
             embeddings = [self.word_rep(w, None) for w in sentence]
         else:
             embeddings = [self.word_rep(w, chars) for w, chars in zip(sentence, word_chars)]
->>>>>>> upstream/master
 
         lstm_out = self.word_bi_lstm.transduce(embeddings)
 
@@ -224,8 +203,7 @@ def old_save(self, file_name):
         members_to_save.extend(utils.sortvals(self.lstm_to_tags_bias))
         members_to_save.extend(utils.sortvals(self.mlp_out))
         members_to_save.extend(utils.sortvals(self.mlp_out_bias))
-        #self.model.save(file_name, members_to_save)
-        self.model.save(file_name)
+        self.model.save(file_name, members_to_save)
 
         with open(file_name + "-atts", 'w') as attdict:
             attdict.write("\t".join(sorted(self.attributes)))
@@ -262,8 +240,8 @@ def get_word_chars(sentence, i2w, c2i):
     parser.add_argument("--num-epochs", default=20, dest="num_epochs", type=int, help="Number of full passes through training set (default - 20)")
     parser.add_argument("--num-lstm-layers", default=2, dest="lstm_layers", type=int, help="Number of LSTM layers (default - 2)")
     parser.add_argument("--hidden-dim", default=128, dest="hidden_dim", type=int, help="Size of LSTM hidden layers (default - 128)")
-    parser.add_argument("--training-sentence-size", default=maxsize, dest="training_sentence_size", type=int, help="Instance count of training set (default - unlimited)")
-    parser.add_argument("--token-size", default=maxsize, dest="token_size", type=int, help="Token count of training set (default - unlimited)")
+    parser.add_argument("--training-sentence-size", default=maxint, dest="training_sentence_size", type=int, help="Instance count of training set (default - unlimited)")
+    parser.add_argument("--token-size", default=maxint, dest="token_size", type=int, help="Token count of training set (default - unlimited)")
     parser.add_argument("--learning-rate", default=0.01, dest="learning_rate", type=float, help="Initial learning rate (default - 0.01)")
     parser.add_argument("--dropout", default=-1, dest="dropout", type=float, help="Amount of dropout to apply to LSTM part of graph (default - off)")
     parser.add_argument("--no-we-update", dest="no_we_update", action="store_true", help="Word Embeddings aren't updated")
@@ -305,7 +283,7 @@ def get_word_chars(sentence, i2w, c2i):
                options.training_sentence_size, options.token_size, options.learning_rate, options.dropout, options.loss_prop))
 
     if options.debug:
-        print("DEBUG MODE")
+        print "DEBUG MODE"
 
     # ===-----------------------------------------------------------------------===
     # Read in dataset
@@ -349,8 +327,6 @@ def get_word_chars(sentence, i2w, c2i):
     else:
         word_embeddings = None
 
-
-
     tag_set_sizes = { att: len(t2i) for att, t2i in t2is.items() }
 
     if options.loss_prop:
@@ -423,7 +399,6 @@ def get_word_chars(sentence, i2w, c2i):
         # log epoch's train phase
         logging.info("\n")
         logging.info("Epoch {} complete".format(epoch + 1))
-
         # here used to be a learning rate update, no longer supported in dynet 2.0
         print trainer.status()
 
@@ -525,11 +500,6 @@ def get_word_chars(sentence, i2w, c2i):
 
         # epoch loop ends
 
-
-    # dump embeddings after tagger training
-    emb_dump = options.log_dir + "/trained-embeddings.txt"
-    model.dump_embeddings(emb_dump)
-
     # evaluate test data (once)
     logging.info("\n")
     logging.info("Number test instances: {}".format(len(test_instances)))