Skip to content

thomasbbrunner/tum-adlr-ws20-06

Repository files navigation

ADLR Project

  • run build.sh to install requirements

  • Results of both models with different DoFs are stored in src/figures/evaluation/results/

Hyper-parameters to tune

cVAE

  • learning rate (<= 0.01)
  • batch_size
  • amount of hidden layers per encoder/decoder (>= 1)
  • amount of neurons per hidden layer (>= 100)
  • variational beta
  • (loss weightings)

INN

  • learning rate (<= 0.001)
  • batch_size
  • amount of coupling layers (>= 2)
  • amount of hidden layers per subnetwork (>= 1)
  • amount of neurons per hidden layer (>= 100)
  • (loss weightings)

TODOs:

  • implement cVAE model
  • implement planar robot simulation with 2 and 3 DoF
  • implement paper's robot simulation
  • implement rejection sampling
  • implement random search for hyperparameter optimization (https://docs.ray.io/en/master/tune/)
  • implement basic INN model
  • implement backward training of INN model
  • debug MMD loss

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published