mattdl/continualprototypeevolution

Codebase for Continual Prototype Evolution (CoPE) to attain perpetually representative prototypes for online and non-stationary datastreams. Includes implementation of the Pseudo-Prototypical Proxy (PPP) loss.

Matthias De Lange ~ Professional Portfolio

continualprototypeevolution's Issues

Missing file ?

Hi,

I'm trying to include your method as a baseline for a research project. I tried running the code following the .sh file, and I'm getting the following error

lpagec@dp-gpu2[ContinualPrototypeEvolution](main)$python main.py --tasks_to_preserve 5 --save_path results/cifar10/ --data_path ./data --log_every 100 --samples_per_task 10000 --data_file cifar10.pt --cuda yes --model prototypical.CoPE --batch_size 10 --lr 0.005 --loss_T 0.1 --p_momentum 0.99 --n_memories 100 --n_outputs 256 --n_iter 1 --n_seeds 5 demo_CIFAR10
Traceback (most recent call last):
  File "main.py", line 12, in <module>
    import metrics.plot as plot
  File "/home/ml/users/lpagec/pytorch/ContinualPrototypeEvolution/metrics/plot.py", line 4, in <module>
    import utils
ModuleNotFoundError: No module named 'utils'

Is this a file you created ? Or is there a package called utils that you are using ?

Thanks,
Lucas

Looking for appendices mentioned in the paper for more details

Hi Authors,

Your work looks very interesting and I would like to dive more into the details, however, I was not able to refer the appendices mentioned in your paper. Can you please direct me to the version where I can find the Appendices.

Regards,
tomar-s