All pull requests

Pull requests

• refactor: clean up CombinedExtractor features_dim initialization

  check the checklist
  #2295·RuchithaManne opened Oct 26, 2026·7 of 14 tasks·

  ·

• Score squashed gSDE actions on the pre-tanh sample

  LLM generated
  #2290·shaneraphel opened Oct 1, 2026·5 of 16 tasks·

  ·
  2

• Stop VecNormalize rewriting the observation space of the env it wraps

  LLM generated
  #2286·DenisDrobyshev opened Sep 11, 2026·

  ·
  1

• Fix missing header when Monitor appends to a new file

  LLM generated
  #2283·yurekami opened Sep 4, 2026·8 of 16 tasks·

  ·
  1

• Secure deserialization (.load()) by default

  #2264·araffin opened Jun 19, 2026Member·8 of 16 tasks·

  ·
  1

  20

• docs: warn against using Gymnasium FrameStackObservation for image observations

  #2258·midhunxavier opened May 27, 2026·7 of 16 tasks·

  ·

• Render initial state in evaluate_policy (Fixes #1692)

  #2257·midhunxavier opened May 26, 2026·11 of 16 tasks·

  ·
  1

• Add opt-in tanh squashing for DiagGaussianDistribution mean actions

  LLM generated
  #2249·cgliner opened May 5, 2026·8 of 16 tasks·

  ·

• Fix load path handling for zip files

  PR template not filled
  #2239·Xenios91 opened Apr 13, 2026·

  ·

• WIP: Beta distribution support for PPO

  #2228·araffin opened Mar 13, 2026Member·16 tasks·

  ·
  1

  1

• Add PoolVecEnv support and update imports across the codebase

  #2204·drl-test opened Dec 13, 2025·8 of 15 tasks·

  ·

• Add MPS Backend

  #2203·oursland opened Dec 13, 2025·3 of 15 tasks·

  ·
  1

• Add PAL to approximate Prioritized Experience Replay

  #2199·bilelsgh opened Dec 5, 2025·6 of 16 tasks·

  ·
  3

• Fix: Add close() method to BaseAlgorithm to prevent memory leaks

  Maintainers on vacation
  #2193·RPalmr opened Nov 19, 2025·9 of 16 tasks·

  ·

• Add check_vecenv function for VecEnv validation similar to env_checker

  #2178·araffin with CopilotAI opened Sep 26, 2025·

  ·
  1

  4

  

• Add Custom FrozenLakePlus Environment with Dynamic Slipperiness and Training/Evaluation Scripts

  #2148·llDanieLll opened Jun 16, 2025·

  ·

• use lerp for polyak updates

  #2136·ClashLuke opened May 12, 2025·

  ·

• Optional BatchNorm integration in NatureCNN

  #2132·Mahsarnzh opened May 6, 2025·12 of 16 tasks·

  ·

• Speedup handling of terminal observations

  #2086·araffin opened Feb 17, 2025Member·16 tasks·

  ·

• Fix tests for mps support

  mac os
  #2005·deathcoder opened Sep 14, 2024·50 of 61 tasks·

  ·
  4

• Recalculate Returns and Advantages After Callback to Ensure Reward Consistency (common/on_policy_algorithm.py)

  PR template not filled
  #2000·mhyrzt opened Sep 7, 2024·4 of 16 tasks·

  ·
  4

• proposed fix for RunningMeanStd overflow

  #1954·spiglerg opened Jun 29, 2024·7 of 8 tasks·

  ·
  1

• Weights only param

  #1902·markscsmith opened Apr 19, 2024Contributor·11 of 16 tasks·

  ·
  2

• Fix missing seed / option args in dummy and subproc vec_env resets during stepping

  #1805·npit opened Jan 11, 2024Contributor·12 of 16 tasks·

  ·
  1

  9

• Prioritized experience replay

  #1622·AlexPasqua opened Jul 23, 2023Contributor·7 of 16 tasks·

  ·
  1

  42