google-research / l2p

Learning to Prompt (L2P) for Continual Learning @ CVPR22 and DualPrompt: Complementary Prompting for Rehearsal-free Continual Learning @ ECCV22
https://arxiv.org/pdf/2112.08654.pdf
Apache License 2.0
408 stars 41 forks source link

reproducing 5-datasets with dualprompt #32

Open jcy132 opened 1 year ago

jcy132 commented 1 year ago

I applied the dual prompt for 5-datasets. Other settings are identical with the l2p case:

num_epochs=5 lenght=10 pull_constraint_coeff=0.5

The result was acc@1 75.5033 | forgetting 14.9538, which is much degraded than the result in the paper, acc@1 88.08 | forgetting 2.21

what are the settings for 5-datasets for dual prompt?