-
Notifications
You must be signed in to change notification settings - Fork 2.2k
Switch from tensorboard
to tensorboard-pytorch
dependency
#835
Comments
This is due to some issues with the |
And if you want a newer version of allennlp, that hasn't been released yet, you need to give pip a particular git commit, or clone the repo and use that. |
So, I ran the upgrade and
There is no |
Those flags are in the current master, not in version 0.3. They will be included in the 0.4 release that's coming up in the next week or so. The tutorials are versioned, and you're looking at the tutorials for the master branch, not for the 0.3 release. |
so I need to git pull the repository to install the 0.4 release, correct? |
You could also do |
Yep, I git pull and install and it has the flag now. However, I fail some verification test. Should I be worried? |
You probably shouldn't be worried, but what tests fail? |
I am not sure if this is an allennlp problem, but this is the output
|
Hmm, I've never seen that one, but it looks like your machine doesn't have enough RAM. I'm not sure how much RAM our tests take, but I would be concerned that you might not have enough RAM to train any significant model if you're getting a memory error just running the tests. |
I see, thank you! |
Um, I am still about how to get my own package recognized by |
if you are using the 0.3.0 pip release, you have to make your own copy of https://github.com/allenai/allennlp/blob/v0.3.0/allennlp/run.py import your packages in that file, and use that as your entry point. See: -- (in the newest version there's just a command line flag to include extra packages, but that hasn't been released on pip yet.) |
I have the same issue mentioned in the beginning: Based on what Matt mentioned #835 (comment) it's Side note: I am still trying to figure how Joel's #835 (comment) fits in the picture :-) |
Joel's comment above was answering a different question. It looks to me like the right thing to do is what @nelson-liu suggested (#762 (comment)), switching to |
Sure, I can try to do it. To be clear: the chief purpose of the original |
Yes, I'm pretty sure that's true. @DeNeutoy, @matt-peters, you agree? |
Sounds fine with me as long as whatever library we do use has an easy way to view the resulting log files. |
@nelson-liu, the idea is that you'll install |
Yes! I don't think it makes sense to have |
tensorboard
to tensorboard-pytorch
dependency
This was in the 0.4 release. Thanks @nelson-liu ! |
I tried to update the allennlp by using
pip install --upgrade allennlp
, but I received the error message like the following when I attempted to runpython -m allennlp.run predict -h
:I thought I should use the
--upgrade
flag, so I reinstall the miniconda3 and allennlp. However, the error message is still printed out.I execute those lines in
CentOS 7 Linux
run on a computing cluster's login node. I presume this should be an allennlp error ? Also, just curious, what is the correct way to update allennlp without pulling again from the git? (since this is unclear in the allennlp README)The text was updated successfully, but these errors were encountered: