An Implementation of Model-Agnostic Meta-Learning in PyTorch with Torchmeta
An implementation of Model-Agnostic Meta-Learning (MAML) in PyTorch with Torchmeta.
To avoid any conflict with your existing Python setup, it is suggested to work in a virtual environment with
virtualenv. To install
virtualenv:
bash pip install --upgrade virtualenvCreate a virtual environment, activate it and install the requirements in
requirements.txt.
bash virtualenv venv source venv/bin/activate pip install -r requirements.txt
train.pyto meta-train your model with MAML. For example, to run MAML on Omniglot 1-shot 5-way with default parameters from the original paper:
bash python train.py /path/to/data --dataset omniglot --num-ways 5 --num-shots 1 --use-cuda --step-size 0.4 --batch-size 32 --num-workers 8 --num-epochs 600 --output-folder /path/to/resultsThe meta-training script creates a configuration file you can use to meta-test your model. You can use
test.pyto meta-test your model:
bash python test.py /path/to/results/config.json
The code available in this repository is mainly based on the paper
Chelsea Finn, Pieter Abbeel, and Sergey Levine. Model-agnostic meta-learning for fast adaptation of deep networks. International Conference on Machine Learning (ICML), 2017 [ArXiv]
If you want to cite this paper
@article{finn17maml, author = {Chelsea Finn and Pieter Abbeel and Sergey Levine}, title = {Model-{A}gnostic {M}eta-{L}earning for {F}ast {A}daptation of {D}eep {N}etworks}, journal = {International Conference on Machine Learning (ICML)}, year = {2017}, url = {http://arxiv.org/abs/1703.03400} }