Closed ha36d closed 11 months ago
Does it work after bumping auto-gptq?
Does it work after bumping auto-gptq?
Yes.
Done.
@hamed thanks for the contribution
@ha36d the issue still persist when i run. Can you provide the commands to run properly?
@ha36d the issue still persist when i run. Can you provide the commands to run properly?
@neokd for me, it is working with the following command:
docker build -f Dockerfile . -t neogpt
It takes time, but finishes successfully.
Is there any requirement to upgrade python ? in the container?
It says auto-gptq 0.4.2 not found and asks me to upgrade pip. I have updated pip in the dockerfile and when ran again i get the #82 issue
It seems some cache issues. can you please delete the docker images and layers related to neogpt?
also I see that dev-requirement.txt
is renamed to dev-requirements.txt
without chaning the references in dockerfile.
there aren't any images that i have built successfully till now
I can add a docker test/lint/build
job with Github Actions if you have a docker repository to push the image.
No currently we don't have a docker repository. Is there any other way to run it locally?
No, I guess there is an environment issue, as the pip is up to date, the version 0.4.2 of auto-gptq exists.
How can we fix it?
As I said, I think it is your local env that has an issue. I proposed to have a docker build (even without push) pipeline to show you that it works well in Github Actions.
Yeah you can create an issue and work on it
Yeah you can create an issue and work on it
resolves #79 Last version of torch is not compatible with python >= 3.12, therefore the version of python is fixed to 3.11. Bump auto-gptq to the latest version, as there was bugs.