Python3 didn't work at the end of pre-training And pre-model didn't save successfully

[chajiuqqq]

System information

OS Platform and Distribution (e.g., Linux Ubuntu 16.04): win10 x86
TensorFlow installed from (source or binary): anaconda
TensorFlow version: 1.14.0
Python version: 3.7.4

My problem

I downloaded the data files supported by you, when I began step4.2(Keywords Extraction and Genre Pattern Building), it always showed that 'gbk' can't decode \xxx in line xx. So I tried to change the read-file way in the source. Like this:
Before:

    fin = open(file_path, 'r')

After:

    fin = open(file_path, 'r',encoding='utf-8')

It works, errors have gone when I changed all the places where it gave a decoding error.

But at the end of pre-training, it showed that "Python3 didn't work" with a dialogue box. And the code showed at last are following, And there hasn't a fold named 'premodel' in WMPoetry\wm, can you help me?

...
...
不辞朝谒远                          # 唯要近慈恩                          # 唯有信慈亲                          #
epoch: 1, 8400/8607 97.607%, 0.282 s per iter
train loss: 3.806  ppl:44.96, l2 loss: 17243.167, lr:0.0010.
______________________
今朝复相见                          # 疑是葛仙翁                          # 留是后天翁                          #
epoch: 1, 8600/8607 99.930%, 0.284 s per iter
train loss: 3.773  ppl:43.52, l2 loss: 17536.656, lr:0.0010.
______________________
0.0010 to 0.0010
run validation...
validation epoch: 1  loss: 5.679  ppl: 292.55, l2 loss: 30140.641
saving model...
WARNING:tensorflow:*******************************************************
W0518 15:41:08.506486 15588 saver.py:1134] *******************************************************
WARNING:tensorflow:TensorFlow's V1 checkpoint format has been deprecated.
W0518 15:41:08.507480 15588 saver.py:1135] TensorFlow's V1 checkpoint format has been deprecated.
WARNING:tensorflow:Consider switching to the more efficient V2 format:
W0518 15:41:08.509477 15588 saver.py:1136] Consider switching to the more efficient V2 format:
WARNING:tensorflow:   `tf.train.Saver(write_version=tf.train.SaverDef.V2)`
W0518 15:41:08.510476 15588 saver.py:1137]    `tf.train.Saver(write_version=tf.train.SaverDef.V2)`
WARNING:tensorflow:now on by default.
W0518 15:41:08.511476 15588 saver.py:1138] now on by default.
WARNING:tensorflow:*******************************************************
W0518 15:41:08.512475 15588 saver.py:1139] *******************************************************
Windows fatal exception: access violation

Current thread 0x00003ce4 (most recent call first):
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 1429 in _call_tf_sessionrun
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 1341 in _run_fn
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 1356 in _do_call
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 1350 in _do_run
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 1173 in _run
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\client\session.py", line 950 in run
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\training\saver.py", line 1173 in save
  File "pretrain.py", line 244 in train
  File "pretrain.py", line 251 in main
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\absl\app.py", line 250 in _run_main
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\absl\app.py", line 299 in run
  File "C:\Users\Alice\Miniconda3\envs\tflearning\lib\site-packages\tensorflow\python\platform\app.py", line 40 in run
  File "pretrain.py", line 254 in <module>

[XiaoyuanYi]

@chajiuqqq Hi, there should be two directories, model/ and premodel/ in WMPoetry/wm/, which are used to save checkpoints of the wm model and the pre-trained model respectively, as shown in WMPoetry/wm/config.py. Have you created these directories? Since we run our model in Linux instead of windows, there may be some unexpected problems. Please check that if you correctly create model/ and premodel/. For any further questions, just reply in this issue.

[chajiuqqq]

@XiaoyuanYi Thank you for your reply, I haven't created these two directories before. And it succeeded to save model when I have done this operation. I have supposed these two directories will be created automatically.
THANK YOU!

[chajiuqqq]

@XiaoyuanYi I have another question when attempting to integrate WMPoetry with Flask, a framework of Python Web. If you have such experience in Python Web, hope you take a few time to view my question.

Flask receives http request and returns response after having done the designed service. It's natural to run gen_ui.py when receiving the request and finally return the output as response. But it takes time to do some same preparative work every time running gen_ui.py.

So, I'm curious if there is any way to keep gen_ui.py running and block it when hasn't request instead of restarting gen_ui.py every time.

One possible way is multi-thread. Is it possible? I'm a green hand in Python. Hope for your reply.

[XiaoyuanYi]

@chajiuqqq I would be glad to help but I'm in charge of model and algorithm design, not familiar with Flask. Sorry, I can't provide advice about this problem.

[chajiuqqq]

Never mind it. Thank you for contributing such a good open-source project. Hope you have a nice day! 获取 Outlook for iOS<https://aka.ms/o0ukef>

…

________________________________ 发件人: XiaoyuanYi <notifications@github.com> 发送时间: Thursday, May 21, 2020 8:18:34 AM 收件人: XiaoyuanYi/WMPoetry <WMPoetry@noreply.github.com> 抄送: chajiuqqq <chajiuqqq@outlook.com>; Mention <mention@noreply.github.com> 主题: Re: [XiaoyuanYi/WMPoetry] Python3 didn't work at the end of pre-training And pre-model didn't save successfully (#5) @chajiuqqq<https://github.com/chajiuqqq> I would be glad to help but I'm in charge of model and algorithm design, not familiar with Flask. Sorry, I can't provide advice about this problem. ― You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub<#5 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AL5SEDXXEDYHOVE4IG3XUSTRSRXNVANCNFSM4ND325ZA>.