princeton-nlp / LESS

[ICML 2024] LESS: Selecting Influential Data for Targeted Instruction Tuning
MIT License
306 stars 25 forks source link

data selection results #6

Closed wang-debug closed 5 months ago

wang-debug commented 5 months ago

Thank you very much for your work! i hope it's not too forward to ask if you could share the datasets obtained during the third step of data selection. I believe it could help reduce costs.

xiamengzhou commented 5 months ago

Hi! Sorry for the late reply. less-data.zip contains the selected data in the subfolder selected_data. Could you check it out?

wang-debug commented 5 months ago

Thank you for your sharing!