It is a reproduced version of maml, which is implemented with PyTorch 1.6.0 and support Multi-GPUs both in Meta-training phase and Meta-testing phase.
All the hyper-parameters and tricks, e.g. gradient clip, are strictly consistent with the original paper Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
HERE and its Tensorflow Implementation
HERE.
- python: 3.7
- Pytorch: 1.6.0
- Downloading
MiniImagenet
dataset - Changing
dataset_dir
inrun_meta.py
with your own root of the MiniImagenet dataset - python run_meta.py
5-way 1-shot | 5-way 5-shot | |
---|---|---|
MAML | 48.7% | 63.1% |
Ours | 48.9% | #TODO# |
https://github.com/cbfinn/maml
https://github.com/dragen1860/MAML-Pytorch