learn2learn
learn2learn copied to clipboard
A PyTorch Library for Meta-learning Research
My environment already has Pytorch(1.12.1) installed with conda. > pytorch 1.12.1 py3.10_cuda11.3_cudnn8.3.2_0 pytorch pytorch-cluster 1.6.0 py310_torch_1.12.0_cu113 pyg pytorch-mutex 1.0 cuda pytorch pytorch-scatter 2.1.0 py310_torch_1.12.0_cu113 pyg pytorch-sparse 0.6.16 py310_torch_1.12.0_cu113 pyg While...
### Description New design of the images added to the "docs/assets" directory.
In the pytorch-lightning2.2.1 environment, Maml is implemented by wrapping the meta_learn function with torch.enable_grad, which makes the model's outer losses uncomputed during validation, resulting in memory overflow. This can be...
Training the model for one epoch without learner.adapt(losses) takes about 20 minutes, but using learner.adapt(losses) increases the time to 8 hours and 30 minutes, making it highly inefficient. My GPU...
### Description There are incorrect Source urls on several functions defined in `utils`. They are linking to a `utils.py` file, which was turned into a subpackage, so now they are...
Hello, When trying to apply the Sine Wave example approach to a transformer based model I get the following output: File "/usr/local/lib/python3.10/dist-packages/torch/autograd/graph.py", line 767, in _engine_run_backward return Variable._execution_engine.run_backward( # Calls...
Error message: p.grad.data.mul_(1.0 / meta_batch_size) AttributeError: 'NoneType' object has no attribute 'data' For lines; # Average the accumulated gradients and optimize for p in maml.parameters(): print(p) p.grad.data.mul_(1.0 / meta_batch_size) opt.step()...
Hi. First of all, thank you very much for this project. We've been using it for several years in a meta-learning class. Unfortunately, for some reason the installation of learn2learn...
Hi i have been learning meta-rl recently. But i met a problem! and i tired for several days,but still didn't solve it.If anyone can help me, i would be very...