Open DumoeDss opened 2 months ago
Thank you for your kind words!😄 Since we haven't conducted experiments on multilingual data, I don't have a definitive answer, but I think ProX could work better given proper SFT data. Please note that we provided fairly detailed prompts in the appendix(around page 21~24), and I believe applying these prompts directly to data in other languages's to generate SFT data might be a viable approach.
Great and insightful work! If the refinement model does not support multilingualism, will it work for multilingual datasets?