Cannot install Transformers version >2.3.0 with pip on CentOS #2980
Labels
Comments
LysandreJik
added
Core: Tokenization
|
Which version of Rust are you on? Looking at the trace PyO3 requires at least 1.37.0-nightly 2019-07-19. |
|
I am on version |
|
Interestingly, their website says 1.37.x is the requirement, but on GitHub they say you need 1.42.0-nightly 2020-01-21. That's quite a harsh requirement, I think, but nothing you can do about it I suppose. (Except installing an older version of PyO3 from source.) Can you try either of those options and let us know? |
|
I went with the former option (installing a nightly build of rust). Here is what I tried:
Try to update No beans. Got the following stacktrace: |
|
I am not sure if this is a version issue or an incompatibility issue, cf. and related: https://stackoverflow.com/questions/55363823/redhat-centos-glibc-2-18-not-found Because of cross-platform compatibility, it might better to have |
|
Hi @JohnGiorgi, Thanks for reporting this issue and sorry you're having trouble with transformers. As @BramVanroy mentioned pyo3 package (which allows us to build the Python binding to Rust) requires some features only available in nightly release. However 1.42 is only required for the version I did try to reproduce the error you linked above on a Fedora machine I've at hand and wasn't able to. Can you provide some more information regarding this machine ? On which platform is it running ? x86_64 ? POWER9 ? Also if you can include some Any information you might be able to provide to us will help us track down this build issue. Many thanks |
|
Looking at the trace, it seems @mfuntowicz From the CentOS fora, it seems that 2.18 will never come to CentOS 7 so I fear this is just an incompatibility. That adds to my point that I would suggest that |
|
@BramVanroy The nightly features can take many release cycles before landing on a stable version. So these features are probably still part of the nightly only, even in the stable Anyway, we provide |
|
@n1t0 Ah, that makes sense. Thanks for the clarification. @JohnGiorgi Can you try to install |
|
Output of Output of Output of I am working on a Compute Canada cluster, so information about it can also be found here. Here is the output of |
|
I'm also having these same errors. It gets past the install, but then when running the tests `error: Can not find Rust compilerERROR: Failed building wheel for tokenizers |
|
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions. |
|
Hey guys, @JohnGiorgi @Shane-Neeley , I think I figured out what was happening to you (to me also). At the time of your question, I think there was a high possibility that in the "setup.py" of transformers' source code, there was a line writing "tokenizers=0.5.2" ; but in fact this old version will bother the update of transformers. And now they just corrected this line in the setup.py. So I just suggest you to
Then voilà ! You'll get a brand new smoothy transformers. |
I am installing transformers from source but I am getting the error. |
|
these lines worked for me |
Подскажите, пожалуйста, куда именно Вы вводите данные команды? Please tell me where exactly you enter these commands? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
🐛 Bug
I cannot install
pip install transformersfor a release newer than2.3.0. The install errors out when trying to installtokenizers. This is similar to another issue, except I have a Rust Compiler in my environment so I do not see:"error: can not find Rust Compiler".Information
Model I am using (Bert, XLNet ...):
N/A
Language I am using the model on (English, Chinese ...):
N/A
The problem arises when using:
The tasks I am working on is:
To reproduce
Steps to reproduce the behavior:
which leads to the following error:
Expected behavior
A successful install of
transformers.Environment info
transformersversion:>2.3.03.7.41.4.0The text was updated successfully, but these errors were encountered: