Séb Arnold

Closing since fixed.

Closing since we won't provide weights for all possible configs (see [get_pretrained_backbones](https://learn2learn.net/docs/learn2learn.vision/#learn2learn.vision.models.get_pretrained_backbone) for supervised pretraining ones).

Closing: fixed in #387.

Closing: inactive. Note that this error is quite generic, and it typically stems from non-standard computation graphs (as @xuliwalker was attempting). In this case, my advice is to skip the...

I'm not sure batching tasks will be very helpful because we're already running OOM when looping through individual tasks with MAML. I'll leave this open for now and we can...

Closing: will be fixed as soon as we merge #400.

Closing: inactive.

Closing: it seems the error came from decaying the meta learning rate, which is now fixed.

This tutorial should guide you step-by-step to build your own taskset: https://learn2learn.net/tutorials/task_transform_tutorial/transform_tutorial/

Closing since inactive.