QingruZhang / AdaLoRA

AdaLoRA: Adaptive Budget Allocation for Parameter-Efficient Fine-Tuning (ICLR 2023).
MIT License
231 stars 23 forks source link

Random seeds used in your experiments #23

Open Car-pe opened 4 months ago

Car-pe commented 4 months ago

Hi authors!

I read that you report the mean of the performance from 5 random seeds. Could I know what random seeds are used? Since when I try different random seeds, the results are quite different and I can not reproduce your results.

Thank you!

QingruZhang commented 3 months ago

Hello, can you tell me which experiments you are working on?

Car-pe commented 3 months ago

Sure, the model is Divert_v3_base, task is NLU glue.