Code

CrossQ#204

Code

Draft

beardyFace wants to merge 33 commits intomainUoA-CARES/cares_reinforcement_learning:mainfrom alg/cross-qUoA-CARES/cares_reinforcement_learning:alg/cross-qCopy head branch name to clipboard

+513-3

Commits

Commits on Oct 11, 2024

• CrossQ Implementation

  beardyFace

  committed
  bf635fd

Commits on Oct 21, 2024

• Fixes to CrossQ implementation and default parameters

  beardyFace

  committed
  dfb33a5

Commits on Oct 23, 2024

• Eval vs Train setting of actor/critic

  beardyFace

  committed
  304f430

Commits on Oct 25, 2024

• Still not working

  beardyFace

  committed
  14f6b1c

Commits on Oct 28, 2024

• best crossq so far

  beardyFace

  committed
  aeabcfa

Commits on Nov 20, 2024

• MLP Generic norm and activation functions

  beardyFace

  committed
  deadff2

• Shifting Configs into Actor/Critics to enable MLP parameters to be passed down

  beardyFace

  committed
  643425f

Commits on Nov 21, 2024

• Overkill MLP

  beardyFace

  committed
  16ff53e

Commits on Nov 22, 2024

• Updated all networks to use MLP

  beardyFace

  committed
  463b881

Commits on Nov 25, 2024

• Updated all networks but AE based for base + default + MLP networks

  beardyFace

  committed
  3dcd047

• REDQ fix using Ensemble method from CTD4

  beardyFace

  committed
  388ec91

• SACAE and TD3AE updated to reflect default + base + actor/critic layout - lays groundwork for future generalisation of AE for all methods

  beardyFace

  committed
  d2679fc

• Type hint for types in networks

  beardyFace

  committed
  56be414

Commits on Nov 26, 2024

• update for BatchNorm eval vs test

  beardyFace

  committed
  2588626

Commits on Dec 2, 2024

• NaSATD3 Default + Base + Custom + algorithm tests

  beardyFace

  committed
  b3c27e7

• added asserts

  beardyFace

  committed
  5651e72

• merged main

  beardyFace

  committed
  909fa8a

• Updated with MLP update

  beardyFace

  committed
  91c3532

• CrossQ Base, Default, Custom - not MLP yet as different layout of batchnorm

  beardyFace

  committed
  2d54abb

• Type hint for CrossQ

  beardyFace

  committed
  4bee841

• merged main

  beardyFace

  committed
  aee0a08

Commits on Dec 3, 2024

• update main

  beardyFace

  committed
  95763d3

• BatchReNorm with warm up moved into code base

  beardyFace

  committed
  64ed5ba

Commits on Dec 4, 2024

• batchrenorm

  beardyFace

  committed
  b4a3c3d

• BatchReNorm from pytorch rl

  beardyFace

  committed
  38a4219

• batchrenorm from stablebaselines best so far - but not perfect

  beardyFace

  committed
  5944c1e

Commits on Dec 11, 2024

• Merge branch 'main' into alg/cross-q

  beardyFace

  committed
  07aad0a

• Merged with main - configurable CrossQ - shifted BatchReNorm to networks folder

  beardyFace

  committed
  297b6cf

• beta for optimizers for actor and critic

  beardyFace

  committed
  1ae79c0

Commits on Dec 12, 2024

• merge with main

  beardyFace

  committed
  3542ab2

• README

  beardyFace

  committed
  991cd46

• README

  beardyFace

  committed
  ca96b21

• Merge branch 'main' into alg/cross-q

  beardyFace

  authored
  119e698