PyCIL
PyCIL copied to clipboard
G-U-N
PyCIL: A Python Toolbox for Class-Incremental Learning
Hi, I am trying to run pre-trained ResNet18 with CIFAR100 and FOSTER method but I am getting this specific error: `RuntimeError: Error(s) in loading state_dict for ResNet: Missing key(s) in...
When i run beef with below configs(b0 20steps and b0 10steps) on imagenet100, the loss is nan. But it can be run in b0 5steps normally. Is beef need special...
https://github.com/G-U-N/PyCIL/blob/07c9114d8b9be1534185ab1d9b36f1709867c7f0/convs/resnet.py#L174 This setting is usually applied to low-resolution datasets such as cifar100 to avoid too small feature maps. If the imagenet dataset also uses this setting, it will result in...
When i run codes on GPU1, it seems that .cuda() will put data on default GPU and then bugs occurred.
When i run codes on GPU1, it seems that .cuda() will put data on default GPU and then bugs occurred. File "/home/ubuntu/lzc/PyCIL/utils/inc_net.py", line 605, in forward fc_weight = torch.cat([self.old_fc.weight,torch.zeros((new_task_size,self.feature_dim-self.out_dim)).cuda()],dim=0) RuntimeError:...
No significant modifications to the code. Test run failed with the exps/gem.json provided. File "/scratch/zar00002/PyCIL/main.py", line 31, in main() File "/scratch/zar00002/PyCIL/main.py", line 12, in main train(args) File "/scratch/zar00002/PyCIL/trainer.py", line 21,...
Hi Contributors, I am trying to reproduce the results of SSRE and FeTrIL on tiny-imagenet-200 datasets. But getting the following TypeError. ``` python main.py --config=./exps/fetril.json ``` These are my parameters...
Hi, thank you for sharing this project, I was studying the code of beefios.py and I found that the loss_kd in the expansion phase's and the fusion phase's loss_fe and...
Hello, I've found some bugs about your code computing the accuracy after each task. After training on the new task, the learner is expected to evaluate on all seen data,...
1. ACIL: Analytic Class-Incremental Learning with Absolute Memorization and Privacy Protection. NeurIPS 2022. 2. DS-AL: A Dual-Stream Analytic Learning for Exemplar-Free Class-Incremental Learning. AAAI 2024. This commit fixes #70. I...
In trainer.py in the function "_set_device", the comparison to determine whether to use cpu or gpu is done inside the loop with the variable "device_type". If "device_type" is -1, "cpu"...
