Source code for NeurIPS 2020 paper "Meta-Learning with Adaptive Hyperparameters"
Source code for NeurIPS 2020 paper "Meta-Learning with Adaptive Hyperparameters" (previously titled "Adaptive Learning for Fast Adaptation")
This repository is the implementation of ALFA. The code is based off the public code of MAML++, where their reimplementation of MAML is used as the baseline.
To install requirements, first download Anaconda3 and then run the following:
bash install.sh
For miniIamgenet, the dataset can be downloaded from the link provided from MAML++ public code. make a directory named 'datasets' and place the downloaded miniImagnet under the 'datasets' directory.
To train the model(s) in the paper, run this command in experiment_scripts folder:
For single GPU
bash alfa+maml.sh 0
where 0 represent GPU_ID.
After training is finished, the same command is run to evaluate:
For single GPU:
bash alfa+maml.sh 0