Skip to content

Conversation

@niruc100
Copy link

Hey Bill,

I did a small reproducibility study of your work for my master seminar @hhu Düsseldorf. For this, I tried to run the code fully without GPU and noticed some small issues that I wanted to fix. Now the whole code can be executed on a CPU. I also couldn't resist to change minor parts of the README.md to ensure all commands work properly and that no original results are overwritten by the new ones.

Overall, very nice code and also a nice paper, I really appreciated that :)

Kind regards,
Nick

P.S. All predictions are reproducible (1 single case is different for GPT on "all") and I can also verify that fine-tuning helps.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant