Question about seed numbers.

microsoft / LoRA

Code for loralib, an implementation of "LoRA: Low-Rank Adaptation of Large Language Models"

https://arxiv.org/abs/2106.09685

MIT License

10.43k stars 669 forks source link

Question about seed numbers. #144

Open SEONHOK opened 11 months ago

SEONHOK commented 11 months ago

Hi Authors, I want to reproduce the results but the estimated performances in the paper come from several different random seeds. Can I know how you determined the random seed numbers? From the codes, I know at least one random seed but I wondered the other seeds.