What is the logprobs in line 108 of the ShowTellModel? #1
Comments
|
It should be correct. What error did you get. |
|
You're right. It is defined at every step. The code checker alerts that something is wrong here since it is not defined ahead(maybe it is not allowed in python3). I am clear about this matter. Thanks for your reply. |
Closed
ruotianluo
added a commit
that referenced
this issue
Apr 9, 2019
* transformer: Further towards AttModel: remove the use of get_logprobs clean up the code, and make transformer as a subclass of AttModel. Fix #1 Update readme for transformer [0.5]Match the new arange behaior. Add reduce on plateau. Use the original options to represent hyperparameters in transofmer Update to previous framework of sample and sample_beam uncomment the original sample. formatting and remove variable. Add noamopt options. Add transformer # Conflicts: # misc/utils.py # models/__init__.py # train.py
ruotianluo
added a commit
that referenced
this issue
Apr 10, 2019
* transformer: Further towards AttModel: remove the use of get_logprobs clean up the code, and make transformer as a subclass of AttModel. Fix #1 Update readme for transformer [0.5]Match the new arange behaior. Add reduce on plateau. Use the original options to represent hyperparameters in transofmer Update to previous framework of sample and sample_beam uncomment the original sample. formatting and remove variable. Add noamopt options. Add transformer # Conflicts: # misc/utils.py # models/__init__.py # train.py
ruotianluo
added a commit
that referenced
this issue
Apr 11, 2019
* transformer: Further towards AttModel: remove the use of get_logprobs clean up the code, and make transformer as a subclass of AttModel. Fix #1 Update readme for transformer [0.5]Match the new arange behaior. Add reduce on plateau. Use the original options to represent hyperparameters in transofmer Update to previous framework of sample and sample_beam uncomment the original sample. formatting and remove variable. Add noamopt options. Add transformer # Conflicts: # misc/utils.py # models/__init__.py # train.py
linzhlalala
pushed a commit
to linzhlalala/self-critical.pytorch
that referenced
this issue
Feb 23, 2021
linzhlalala
pushed a commit
to linzhlalala/self-critical.pytorch
that referenced
this issue
Feb 23, 2021
* transformer: Further towards AttModel: remove the use of get_logprobs clean up the code, and make transformer as a subclass of AttModel. Fix ruotianluo#1 Update readme for transformer [0.5]Match the new arange behaior. Add reduce on plateau. Use the original options to represent hyperparameters in transofmer Update to previous framework of sample and sample_beam uncomment the original sample. formatting and remove variable. Add noamopt options. Add transformer # Conflicts: # misc/utils.py # models/__init__.py # train.py
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Hi, @ruotianluo
I got stuck in this line since I cannot find the logprobs neither in
torch.autogradnor the current file.Is it a typo or am I wrong?
Best regards,
Lerner
The text was updated successfully, but these errors were encountered: