Skip to content

Reinforcement Learning

AmazingDelectableFrenchbulldog-max-1mb

ChubbyConventionalBrontosaurus-size_restricted

IncomparableWebbedHen-size_restricted

yaw_in_place

featured

Examples (MAML)

Dependency

  • python: 3.x

  • Pytorch: 0.4+

Usage

python maml_rl.py

or 

python main.py --zero_order --approx_delta=0.3

Improve it

Based Paper:Multi-Step Model-Agnostic Meta-Learning: Convergence and Improved Algorithms

1667717075389

Rank-One Matrix Factorization

  • matrix_rank_train.py: The training file of rank-one matrix factorization problem
  • meta_matrix_rank.py: The meta configure file of rank-one matrix factorization problem
  • linear_matrix_rank.py: linear network file
python matrix_rank_train.py --approx_method=zero_order --approx_delta=1e-4

Regression

  • regression_train.py: The training file of regression problem
  • meta_regression.py: The meta configure file of regression problem
  • MLP.py: MLP network file
python regression_train.py --approx_method=first_approx --approx_delta=1e-7