tutorial fails with latest release 1.1.2 because of wrong encoding for python3

[fac2003]

Running the tutorial with python3, I get:

Traceback (most recent call last):
File "/Library/Frameworks/Python.framework/Versions/3.5/bin/t2t-datagen", line 41, in
from tensor2tensor.data_generators import all_problems # pylint: disable=unused-import
File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/tensor2tensor/data_generators/all_problems.py", line 22, in
from tensor2tensor.data_generators import algorithmic
File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/tensor2tensor/data_generators/algorithmic.py", line 27, in
from tensor2tensor.data_generators import generator_utils as utils
File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/tensor2tensor/data_generators/generator_utils.py", line 35, in
from tensor2tensor.data_generators import text_encoder
File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/tensor2tensor/data_generators/text_encoder.py", line 56
_UNESCAPE_REGEX = re.compile(ur"\u|\\|\([0-9]+);")

This can be fixed by removing the u in front of the regexp. Python3 does not support it.

Note that there is another spot with the same problem:

from tensor2tensor.data_generators import text_encoder

File "/Library/Frameworks/Python.framework/Versions/3.5/lib/python3.5/site-packages/tensor2tensor/data_generators/text_encoder.py", line 222
c if c in alphabet and c != u"\n" else ur"%d;" % ord(c)

Remove u in front of ur"%d will solve this.

[rsepassi]

Fixed in 1.1.3! Sorry about that.

[lukaszkaiser]

Of course please reopen if you see any other problem in 1.1.3!

[colmantse]

running the tutorial t2t-trainer i got the following

Traceback (most recent call last):
File "t2t-trainer", line 82, in
tf.app.run()
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\platform\app.py", line 48, in run
_sys.exit(main(_sys.argv[:1] + flags_passthrough))
File "t2t-trainer", line 78, in main
schedule=FLAGS.schedule)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensor2tensor\utils\trainer_utils.py", line 284, in run
exp.train_and_evaluate()
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\experiment.py", line 495, in train_and_evaluate
self.train(delay_secs=0)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\experiment.py", line 275, in train
hooks=self._train_monitors + extra_hooks)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\experiment.py", line 665, in _call_train
monitors=hooks)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\python\util\deprecation.py", line 289, in new_func
return func(*args, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 455, in fit
loss = self._train_model(input_fn=input_fn, hooks=hooks)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 955, in _train_model
model_fn_ops = self._get_train_ops(features, labels)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 1162, in _get_train_ops
return self._call_model_fn(features, labels, model_fn_lib.ModeKeys.TRAIN)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensorflow\contrib\learn\python\learn\estimators\estimator.py", line 1133, in _call_model_fn
model_fn_results = self._model_fn(features, labels, **kwargs)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensor2tensor\utils\model_builder.py", line 218, in model_fn
len(my_hp.problems) - 1)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensor2tensor\utils\input_fn_builder.py", line 200, in cond_on_index
return fn(cur_idx)
File "C:\ProgramData\Anaconda3\lib\site-packages\tensor2tensor\utils\model_builder.py", line 198, in nth_model
for loss_key, loss_value in losses_dict.iteritems():
AttributeError: 'dict' object has no attribute 'iteritems'

[lukaszkaiser]

This was a python3 problem, already corrected in PR #210. We're in the process of adding automated tests for python3, that should prevent it from happening again. I'm closing this as it's already solved on master (will make a release tomorrow). Thanks for reporting!

[colmantse]

ah i saw the merge request, i thought it is already updated. so it will be fixed if i pull the master again tomorrow? thanks!

[lukaszkaiser]

Should be fixed if you pull the master now :). (Just remember to do pip install -e . in your pulled repo and/or pip uninstall tensor2tensor so your python will use your downloaded repo and not the pip release which wasn't updated yet.)

[colmantse]

thanks i think i got it running now !!