sholtodouglas/learning_from_play

Issues

• Issue on collecting scripted data

  #21 opened a year ago
  1

• Fully implement learning from pixels

  #16 opened 3 years ago
  1

• Attempt to implement NVAE

  #15 opened 3 years ago
  0

• Fix VAE training stability

  #14 opened 3 years ago
  0

• Change actions to be relative & in end-effector space (not joint space)

  #13 opened 4 years ago
  0

• Investigate loss spikes during training

  #12 opened 4 years ago
  1

• Maximum Data Augmentation (Variable Seq lens, equivalent quaternions)

  #9 opened 3 years ago
  1

• Trial WnB Logging

  #8 opened 4 years ago
  1

• Solve gripper loss

  #7 opened 4 years ago
  1

• Dataloader pre-processing does not match original trajectory

  #5 opened 4 years ago
  0

• Fix ProtoBuf crash when running `from_tensor_slices` on large datasets

  #4 opened 4 years ago
  1

• Change rotation representation from Quaternion to 5D representation

  #3 opened 4 years ago
  0

• Add support for local logging of metrics

  #2 opened 4 years ago
  0

• Change from joint angle actions to xyz-rpy

  #1 opened 3 years ago
  1