search

mingkaid / rl-prompt

Accompanying repo for the RLPrompt paper
MIT License
286 stars 52 forks source link

output data of your experiment #18

Closed li-jing-wen closed 1 year ago

li-jing-wen commented 1 year ago

I need to do some comparative experiments. Could you please provide the final output data of your experiment?

mingkaid commented 1 year ago

Thank you for your interest. What task and dataset are you trying to compare?

li-jing-wen commented 1 year ago

Thank you for your interest. What task and dataset are you trying to compare? Thank you for your help! I want to compare the results of the YELP dataset in the text style transfer task.

mingkaid commented 1 year ago

Sure. We might need a few days to reproduce the setting described in the paper. Once we're done we'll update our repo. I'll close this issue for now because the question has been addressed.

mingkaid commented 1 year ago

Hi, quick update that we have uploaded the final output for the Yelp dataset in the text style transfer task. You can find it at example/text-style-transfer/test_outputs.

We uploaded both the original GPT-2 format (clean) and the typical space-delimited format (raw). We evaluated everything with the clean format, but if you were comparing with other outputs/references in the typical format, raw may be more consistent.

Let us know if you have any questions. Best of luck with your project!