Hi Zhanghao,
As I was trying to reproduce the results in your paper based on the code you have provided, I have encountered the following errors when I try to execute the command bash configs/iwslt14.de-en/prepare.sh in README.md,
Traceback (most recent call last):
File "/usr/local/bin/fairseq-preprocess", line 11, in <module>
load_entry_point('fairseq', 'console_scripts', 'fairseq-preprocess')()
File "/usr/local/lib/python3.6/dist-packages/pkg_resources/__init__.py", line 487, in load_entry_point
return get_distribution(dist).load_entry_point(group, name)
File "/usr/local/lib/python3.6/dist-packages/pkg_resources/__init__.py", line 2728, in load_entry_point
return ep.load()
File "/usr/local/lib/python3.6/dist-packages/pkg_resources/__init__.py", line 2346, in load
return self.resolve()
File "/usr/local/lib/python3.6/dist-packages/pkg_resources/__init__.py", line 2352, in resolve
module = __import__(self.module_name, fromlist=['__name__'], level=0)
File "/home/ubuntu/lite-transformer/fairseq_cli/preprocess.py", line 1
../preprocess.py
^
SyntaxError: invalid syntax
I was running the script on the following machine:
Deep Learning Base AMI (Ubuntu 18.04) Version 25.0
Nvidia driver version: 440.33.01
CUDA versions available: cuda-10.0 cuda-10.1 cuda-10.2 cuda-9.0 cuda-9.2
Default CUDA version is 10.0
Libraries: cuDNN, NCCL, Intel MKL-DNN
I found out that it may be the file fairseq_cli/preprocess.py that is resulting in this error. Given the code you have provided, you might want to redirect the script to ../preprocess.py. But the current code doesn't seem to be working on all platforms.
So I am proposing the following changes.
Furthermore, I have analyzed code in your repo and confirmed that there is actually no other reference to the ./fairseq_cli/ files. So maybe we can just remove them and redirect everything to its real position.
By the way, I have also noticed some others wishing to have similar change on your code, for example, Akramz in their commit af6f3be.
Please point out if there is anything not correct or if there are other considerations about the issue.
Thank you for proposing the changes! The fairseq_cli folder is from the original pytorch/fairseq repository, but I think your changes are reasonable. I merged the changes. Thank you again for your contribution. :)
Hi Zhanghao, As I was trying to reproduce the results in your paper based on the code you have provided, I have encountered the following errors when I try to execute the command
bash configs/iwslt14.de-en/prepare.sh
inREADME.md
,I was running the script on the following machine:
I found out that it may be the file
fairseq_cli/preprocess.py
that is resulting in this error. Given the code you have provided, you might want to redirect the script to../preprocess.py
. But the current code doesn't seem to be working on all platforms. So I am proposing the following changes.Furthermore, I have analyzed code in your repo and confirmed that there is actually no other reference to the
./fairseq_cli/
files. So maybe we can just remove them and redirect everything to its real position.By the way, I have also noticed some others wishing to have similar change on your code, for example, Akramz in their commit af6f3be.
Please point out if there is anything not correct or if there are other considerations about the issue.
Thanks!