avalanche

avalanche copied to clipboard

Avalanche doesn't work well in Google Colab

1

I tried to run my code in colab using this code '!pip install avalanche-lib' but I got this error:  There is something wrong with...

soorasa

IncrementalClassifiers inside MultiHeadClassifiers are adapted on all experiences

2

🐛 **Describe the bug** MultiHead classifier contains a dictionnary of Incremental classifiers, both MultiHeadClassifier and IncrementalClassifier are DynamicModules, so they get adapted through the following loop: ```{python} def avalanche_model_adaptation(model: nn.Module,...

AlbinSou

Generative Replay with weighted loss for replayed data

3

First attempt to add support for Generative Replay with weighted loss for replayed data. Added three arguments: ```python is_weighted_replay=True, weight_replay_loss_factor=2.0, weight_replay_loss=0.001, ```

gogamid

make_train_dataloader discards custom collate function passed as kwarg

5

**Describe the bug** Calling ```python cl_strategy.train( experience, eval_streams=[val_exp], num_workers=4, collate_fn=my_custom_collate, ) ``` should respect all of the keyword arguments I pass in. **In this case, `my_custom_collate` is discarded.** **To Reproduce**...

niniack

The three NLP examples in the example folder don't seem to work. There are issues with initializing CLExperience, CLStream, and CLSceneario that I am having a hard time fixing.

akakayan

The coverage badges are ignoring a lot of the tests we are running, especially checkpointing and distributed tests. @lrzpellegrini do you know if there is any way to make them...

AntonioCarta

Checkpointing issues in 0.4.0

2

When trying to save a checkpoint the, `torch.save(checkpoint_data, fname, pickle_module=dill)` call in `save_checkpoint()` function fails. It complains about a recursive definition when calling `torch.save()` inside the respective function. **Additional context:**...

TuKo

Unnecessarily strict requirement for batch size in MASPlugin

1

**Describe the bug** In the MASPlugin, the following check seems to be unnecessarily strict: [Link to code](https://github.com/ContinualAI/avalanche/blob/afebe34d3f927a0ecefc11a884c7b0282899bc2e/avalanche/training/plugins/mas.py#L84) ```python for _, batch in enumerate(dataloader): # Get batch if len(batch) == 2...

niniack

tutorial missing docs

2

https://github.com/ContinualAI/avalanche/blob/master/examples/icarl.py Tutorials about icarl are missing

Yanyi-Zhang

Usage of function ni_benchmark

1

Dear Avalanche, I would like to ask how to use the function ni_benchmark properly. Here I am using my own dataset defined as: ``` class DistilledCore50(Dataset): def __init__(self, images, labels,...

hailuu684