train/dict updates and move seq2seq (#153)

examples/build_dict.py

@@ -14,7 +14,9 @@
 import os
 
 def build_dict(opt):
-    if 'dict_file' not in opt:
+    if not opt.get('dict_file'):
+        print('Tried to build dictionary but `--dict-file` is not set. Set ' +
+              'this param so the dictionary can be saved.')
         return
     print('[ setting up dictionary. ]')
     if os.path.isfile(opt['dict_file']):

examples/train_model.py

@@ -3,25 +3,25 @@
 # This source code is licensed under the BSD-style license found in the
 # LICENSE file in the root directory of this source tree. An additional grant
 # of patent rights can be found in the PATENTS file in the same directory.
-"""Train a model.
+'''Train a model.
 
 After training, computes validation and test error.
 
 Run with, e.g.:
 
-python examples/train_model.py -m ir_baseline -t dialog_babi:Task:1 -mf "/tmp/model"
+python examples/train_model.py -m ir_baseline -t dialog_babi:Task:1 -mf '/tmp/model'
 
 ..or..
 
-python examples/train_model.py -m rnn_baselines/seq2seq -t babi:Task10k:1 -mf "/tmp/model" -bs 32 -lr 0.5 -hs 128
+python examples/train_model.py -m rnn_baselines/seq2seq -t babi:Task10k:1 -mf '/tmp/model' -bs 32 -lr 0.5 -hs 128
 
 ..or..
 
-python examples/train_model.py -m drqa -t babi:Task10k:1 -mf "/tmp/model" -bs 10
+python examples/train_model.py -m drqa -t babi:Task10k:1 -mf '/tmp/model' -bs 10
 
 TODO List:
 - More logging (e.g. to files), make things prettier.
-"""
+'''
 
 from parlai.core.agents import create_agent
 from parlai.core.worlds import create_task
@@ -36,32 +36,37 @@
 
 def run_eval(agent, opt, datatype, still_training=False):
     ''' Eval on validation/test data. '''
-    print("[ running eval: " + datatype + " ]")
+    print('[ running eval: ' + datatype + ' ]')
     opt['datatype'] = datatype
+    if opt.get('evaltask'):
+        opt['task'] = opt['evaltask']
     valid_world = create_task(opt, agent)
     for i in range(len(valid_world)):
         valid_world.parley()
         if i == 1 and opt['display_examples']:
-            print(valid_world.display() + "\n~~")
+            print(valid_world.display() + '\n~~')
             print(valid_world.report())
         if valid_world.epoch_done():
             break
     valid_world.shutdown()
     valid_report = valid_world.report()
-    metrics = datatype + ":" + str(valid_report)
+    metrics = datatype + ':' + str(valid_report)
     print(metrics)
     if still_training:
         return valid_report
     else:
         if opt['model_file']:
             # Write out metrics
-            f = open(opt['model_file'] + '.' + datatype, "a+")
+            f = open(opt['model_file'] + '.' + datatype, 'a+')
             f.write(metrics + '\n')
             f.close()
 
 def main():
     # Get command line arguments
     parser = ParlaiParser(True, True)
+    parser.add_argument('-et', '--evaltask',
+                        help=('task to use for valid/test (defaults to the ' +
+                              'one used for training if not set)'))
     parser.add_argument('-d', '--display-examples',
                         type='bool', default=False)
     parser.add_argument('-e', '--num-epochs', type=int, default=1)
@@ -70,18 +75,18 @@ def main():
     parser.add_argument('-ltim', '--log-every-n-secs',
                         type=float, default=1)
     parser.add_argument('-vtim', '--validation-every-n-secs',
-                        type=float, default=False)
+                        type=float, default=0)
     parser.add_argument('-vp', '--validation-patience',
                         type=int, default=5,
                         help=('number of iterations of validation where result '
                               + 'does not improve before we stop training'))
-    parser.add_argument('-dbf', '--dict_build_first',
+    parser.add_argument('-dbf', '--dict-build-first',
                         type='bool', default=True,
                         help='build dictionary first before training agent')
     opt = parser.parse_args()
     # Possibly build a dictionary (not all models do this).
-    if opt['dict_build_first']:
-        if 'dict_file' not in opt and 'model_file' in opt:
+    if opt['dict_build_first'] and 'dict_file' in opt:
+        if opt['dict_file'] is None and opt.get('model_file'):
             opt['dict_file'] = opt['model_file'] + '.dict'
         build_dict.build_dict(opt)
     # Create model and assign it to the specified task
@@ -91,7 +96,7 @@ def main():
     train_time = Timer()
     validate_time = Timer()
     log_time = Timer()
-    print("[ training... ]")
+    print('[ training... ]')
     parleys = 0
     num_parleys = opt['num_epochs'] * int(len(world) / opt['batchsize'])
     best_accuracy = 0
@@ -101,17 +106,17 @@ def main():
         world.parley()
         parleys = parleys + 1
         if train_time.time() > opt['max_train_time']:
-            print("[ max_train_time elapsed: " + str(train_time.time()) + " ]")
+            print('[ max_train_time elapsed: ' + str(train_time.time()) + ' ]')
             break
         if log_time.time() > opt['log_every_n_secs']:
             if opt['display_examples']:
-                print(world.display() + "\n~~")
+                print(world.display() + '\n~~')
             parleys_per_sec =  train_time.time() / parleys
             time_left = (num_parleys - parleys) * parleys_per_sec
-            log = ("[ time:" + str(math.floor(train_time.time()))
-                  + "s parleys:" + str(parleys)
-                  + " time_left:"
-                  + str(math.floor(time_left))  + "s ]")
+            log = ('[ time:' + str(math.floor(train_time.time()))
+                  + 's parleys:' + str(parleys)
+                  + ' time_left:'
+                  + str(math.floor(time_left))  + 's ]')
             if hasattr(agent, 'report'):
                 log = log + str(agent.report())
             else:
@@ -125,7 +130,7 @@ def main():
             if valid_report['accuracy'] > best_accuracy:
                 best_accuracy = valid_report['accuracy']
                 impatience = 0
-                print("[ new best accuracy: " + str(best_accuracy) +  " ]")
+                print('[ new best accuracy: ' + str(best_accuracy) +  ' ]')
                 if opt['model_file']:
                     agent.save(opt['model_file'])
                     saved = True
@@ -134,8 +139,8 @@ def main():
                     break
             else:
                 impatience += 1
-                print("[ did not beat best accuracy: " + str(best_accuracy) +
-                      " impatience: " + str(impatience)  + " ]")
+                print('[ did not beat best accuracy: ' + str(best_accuracy) +
+                      ' impatience: ' + str(impatience)  + ' ]')
             validate_time.reset()
             if impatience >= opt['validation_patience']:
                 print('[ ran out of patience! stopping. ]')

parlai/agents/drqa/config.py

@@ -82,7 +82,7 @@ def add_cmdline_args(parser):
 
 def set_defaults(opt):
     # Embeddings options
-    if 'embedding_file' in opt:
+    if opt.get('embedding_file'):
         if not os.path.isfile(opt['embedding_file']):
             raise IOError('No such file: %s' % args.embedding_file)
         with open(opt['embedding_file']) as f:

parlai/agents/drqa/drqa.py

@@ -59,7 +59,7 @@ def __init__(self, *args, **kwargs):
         super(SimpleDictionaryAgent, self).__init__(*args, **kwargs)
 
         # Index words in embedding file
-        if self.opt['pretrained_words'] and 'embedding_file' in self.opt:
+        if self.opt['pretrained_words'] and self.opt.get('embedding_file'):
             print('[ Indexing words with embeddings... ]')
             self.embedding_words = set()
             with open(self.opt['embedding_file']) as f:

parlai/agents/rnn_baselines/seq2seq.py → parlai/agents/seq2seq/seq2seq.py

@@ -12,7 +12,7 @@
 import torch.nn as nn
 import torch
 import copy
-import os 
+import os
 import random
 
 
@@ -74,7 +74,7 @@ def __init__(self, opt, shared=None):
             }
             if self.use_cuda:
                 self.cuda()
-            if 'model_file' in opt and os.path.isfile(opt['model_file']):
+            if opt.get('model_file') and os.path.isfile(opt['model_file']):
                 print('Loading existing model parameters from ' + opt['model_file'])
                 self.load(opt['model_file'])
 

parlai/core/dict.py

@@ -126,12 +126,14 @@ def __init__(self, opt, shared=None):
                 index = len(self.tok2ind)
                 self.tok2ind[self.unk_token] = index
                 self.ind2tok[index] = self.unk_token
-
-            if 'dict_file' in opt and os.path.isfile(opt['dict_file']):
-                self.load(opt['dict_file'])
-            elif 'dict_initpath' in opt:
+
+            if opt.get('dict_file') and os.path.isfile(opt['dict_file']):
                 # load pre-existing dictionary
+                self.load(opt['dict_file'])
+            elif opt.get('dict_initpath'):
+                # load seed dictionary
                 self.load(opt['dict_initpath'])
+
 
         # initialize tokenizers
         st_path = 'tokenizers/punkt/{0}.pickle'.format(opt['dict_language'])
@@ -157,7 +159,7 @@ def __init__(self, opt, shared=None):
                 # fix count for unknown token to one billion
                 self.freq[self.unk_token] = 1000000000
 
-            if 'dict_file' in opt:
+            if opt.get('dict_file'):
                 self.save_path = opt['dict_file']
 
     def __contains__(self, key):

parlai/core/params.py

@@ -110,7 +110,7 @@ def add_model_args(self, args=None):
             '-m', '--model', default='repeat_label',
             help='the model class name, should match parlai/agents/<model>')
         self.add_argument(
-            '-mf', '--model-file', default='',
+            '-mf', '--model-file', default=None,
             help='model file name for loading and saving models')
         # Find which model specified, and add its specific arguments.
         if args is None:
@@ -133,8 +133,7 @@ def parse_args(self, args=None, namespace=None, print_args=True):
         We specifically remove items with ``None`` as values in order to support
         the style ``opt.get(key, default)``, which would otherwise return ``None``.
         """
-        self.args = super().parse_args(args=args)
-        self.opt = {k: v for k, v in vars(self.args).items() if v is not None}
+        self.opt = vars(super().parse_args(args=args))
         self.opt['parlai_home'] = self.parlai_home
         if 'download_path' in self.opt:
             self.opt['download_path'] = self.opt['download_path']