online-continual-learning
online-continual-learning copied to clipboard
RaptorMai
A collection of online continual learning paper implementations and tricks for computer vision in PyTorch, including our ASER(AAAI-21), SCR(CVPR21-W) and an online continual learning survey (Neurocomp...
Hello author, Thanks for the release of the code of your paper. After reading your paper and code, I wonder if the best 'lambda' for EWC ++ is 0 when...
In the evaluate() function, on agents/base.py, after line 171: ` _, pred_label = dists.min(1)` Shouldn't there be a: `pred_label = torch.tensor(self.old_labels, dtype=batch_y.dtype, device=batch_y.device)[pred_label]` Right after 171? Otherwise the error_analysis part...
As I was reading the code, I noticed that in the file utils/buffer/reservoir_update.py, in case the IF statement on line 23 fails (but not the one in line 13), some...
Thanks for your nice work. I have a small question: in `experiments/run.py`, there exist two similar functions: `def multiple_run_tune(default_params, tune_params, save_path)` and `def multiple_run_tune_separate(default_params, tune_params, save_path)`. In my opinion, both...
First, thank you for your great effort in your survey and code implementation : ) I am using it well for my research and I have a question. I tried...
I was wondering, how do we save the model in this framework? Is it same as saving torch model? And can we test the model after the training cycle is...
Greetings! I just made a correction to the calculation of the Average Forgetting. The final task is excluded from the calculation as per Eq (4). This error doesn't have much...
i want to experiment on Experience replay with a buffer which is class balanced. Is there any functionality in the code which can help me do these experiments?
Hi, Thanks for the amazing work. I enjoyed reading your paper. Running experiments I wanted to train ImageNet-1k using some baseline methods you have kindly shared the implementation for. However,...
