Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 5, 2019

• Add transfer learning script (HumanCompatibleAI#106)

  Show description for 744209e
  shwang

  authored
  744209e

• Support custom arguments to VecNormalize (HumanCompatibleAI#111)

  Show description for daeac66
  

  AdamGleave

  authored and

  shwang

  committed
  daeac66

• Imit table (HumanCompatibleAI#105)

  Show description for 279b7f5
  shwang

  authored
  279b7f5

Commits on Oct 26, 2019

• PointMaze: add velocity to observations (HumanCompatibleAI#110)

  Show description for a6faef8
  AdamGleave

  authored
  a6faef8

Commits on Oct 18, 2019

• Add PEP561 support: mark our project as using type checking

  AdamGleave

  committed
  659cde1

Commits on Oct 15, 2019

• Move env fixture to common code (HumanCompatibleAI#108)

  AdamGleave

  authored
  4fd3b82

• Keep Sacred at 0.7.4

  AdamGleave

  committed
  ab993b1

Commits on Sep 30, 2019

• ReplayBuffer: fix type annotation (Env->VecEnv)

  AdamGleave

  committed
  aaddd5e

Commits on Sep 24, 2019

• scripts/train_adversarial.py: Report expert demo quality in result. (HumanCompatibleAI#104)

  Show description for 33a2d63
  shwang

  authored
  33a2d63

• Parallel tensorboard for parallel.py (HumanCompatibleAI#101)

  Show description for 21c8c56
  shwang

  authored
  21c8c56

• Upload dir (HumanCompatibleAI#100)

  Show description for aa1d53a
  shwang

  authored
  aa1d53a

Commits on Sep 23, 2019

• Unnormalized rewards & pluralize rollout NamedTuple attributes (HumanCompatibleAI#97)

  Show description for 9526298
  shwang

  authored
  9526298

• Get experts from public S3 bucket folder (HumanCompatibleAI#103)

  Show description for ac958f4
  shwang

  authored
  ac958f4

Commits on Sep 19, 2019

• Test suite improvements: use tmpdir fixture and publicly expose environment test helper methods (HumanCompatibleAI#99)

  Show description for 1b3fec0
  AdamGleave

  authored
  1b3fec0

• Ray Tune parallelized meta-experiment and various experiment refactoring (HumanCompatibleAI#88)

  Show description for a4d42f8
  

  shwang

  authored and

  AdamGleave

  committed
  a4d42f8

Commits on Sep 11, 2019

• pipeline scripts: Generate rollouts from expert policies (HumanCompatibleAI#87)

  Show description for ca14db2
  shwang

  authored
  ca14db2

• Allow user to specify termination condition for rollouts (HumanCompatibleAI#96)

  Show description for 1fa100d
  

  AdamGleave

  authored and

  shwang

  committed
  1fa100d

• Bugfix in serialization: locals is mutated after assignment

  AdamGleave

  committed
  eade580

Commits on Sep 9, 2019

• Explicit types in util.rollout (HumanCompatibleAI#93)

  Show description for f24891a
  AdamGleave

  authored
  f24891a

Commits on Sep 7, 2019

• Single point-of-entry for deserialization (HumanCompatibleAI#92)

  Show description for 19e8944
  AdamGleave

  authored
  19e8944

Commits on Sep 4, 2019

• Minor docstring and type annotation fixes (HumanCompatibleAI#91)

  

  AdamGleave

  authored and

  shwang

  committed
  a345623

Commits on Aug 23, 2019

• whitespace minor fixes

  shwang

  committed
  dc20572

Commits on Aug 22, 2019

• discrim_net: Minor doc change

  shwang

  committed
  a60ef1d

Commits on Aug 21, 2019

• Time limit (HumanCompatibleAI#86)

  Show description for d37ad49
  shwang

  authored
  d37ad49

• Hotfix: check for null model

  AdamGleave

  committed
  5bc4c28

• Use context managers for deserialized policies and rewards (HumanCompatibleAI#85)

  Show description for ed4779f
  AdamGleave

  authored
  ed4779f

Commits on Aug 20, 2019

• Report wrapped rewards during training (HumanCompatibleAI#83)

  Show description for 04e1ed6
  AdamGleave

  authored
  04e1ed6

• Transfer learning: RL training using loaded reward model (HumanCompatibleAI#81)

  Show description for 0a79409
  AdamGleave

  authored
  0a79409

Commits on Aug 15, 2019

• HardCodedPolicy: return NumPy arrays (HumanCompatibleAI#82)

  Show description for 4058764
  

  AdamGleave

  authored and

  shwang

  committed
  4058764

• train_experts: Save results.txt and build zip file (HumanCompatibleAI#77)

  Show description for acdae78
  shwang

  authored
  acdae78

• Train experts for AIRL environments (HumanCompatibleAI#68)

  Show description for 98a7929
  shwang

  authored
  98a7929

• Restructure codebase (HumanCompatibleAI#80)

  Show description for 35857d4
  AdamGleave

  authored
  35857d4

• Create a registry for policies (HumanCompatibleAI#75)

  Show description for 18a4c96
  AdamGleave

  authored
  18a4c96

• Ensure log directories are unique (HumanCompatibleAI#78)

  

  AdamGleave

  authored and

  shwang

  committed
  376fcd0

Commits on Aug 14, 2019

• Train expert using AIRL transfer reward (HumanCompatibleAI#69)

  Show description for 589a828
  shwang

  authored
  589a828

Pagination

PreviousNext