Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

When i just run train.py,something was wrong #2

Open
Zengpeizhi opened this issue Mar 20, 2019 · 9 comments
Open

When i just run train.py,something was wrong #2

Zengpeizhi opened this issue Mar 20, 2019 · 9 comments

Comments

@Zengpeizhi
Copy link

Zengpeizhi commented Mar 20, 2019

CAE$OHL(1OO KSRO %0 `6U
@whywhs hello,could you please to help me to solve the problem?(i just follow you readme and when i run Train.py ,it was wrong)

@whywhs
Copy link
Owner

whywhs commented Mar 20, 2019

Of course. I just ran my program on computer but there was no error, so I am not sure what caused your problems. Could you please show me the complete error message? Or just show me where the error occurred in my program.

@Zengpeizhi
Copy link
Author

Zengpeizhi commented Mar 20, 2019 via email

@whywhs
Copy link
Owner

whywhs commented Mar 20, 2019

I see your pictures, but I can't see where the error occurred in my program. So, maybe you can take a complete screenshot of the error. Is your pytorch version 0.3?

@Zengpeizhi
Copy link
Author

I see your pictures, but I can't see where the error occurred in my program. So, maybe you can take a complete screenshot of the error. Is your pytorch version 0.3?

oh,sorry,my pytorch is 1.0.1 .So the error was because of that?If so,I try to use 0.3

@whywhs
Copy link
Owner

whywhs commented Mar 20, 2019

yes! I've been finished this project about 1 year ago, so the pytorch version is 0.3

@Zengpeizhi
Copy link
Author

yes! I've been finished this project about 1 year ago, so the pytorch version is 0.3

@whywhs Thank you very much!When I use pytorch0.3,it runs!

@whywhs
Copy link
Owner

whywhs commented Mar 21, 2019

Furthermore, you can write num_works in Train.py (line 128, line 135) if you want because this can make your training faster than before.

@Zengpeizhi
Copy link
Author

hello,thank you very much for your help! And if i want to improve the result ,what do you think i should do?Can you give me some advice?

@whywhs
Copy link
Owner

whywhs commented Apr 3, 2019

Sure. I think you can get a better result if you can do these:

  1. When you finished your training, you can fix the weight of the encoder and train the decoder separately, this will improve your result about 3-4%.
  2. If you want a better result, the batch_size is important. Meanwhile, you can also get a faster speed.
  3. The teacher-forcing ratio is also important, you can try some higher ratio if you are interested.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants