All pull requests

Pull requests

• Fix EpisodeWrapper metrics accumulation with action_repeat

  #676·dafahaha opened Sep 16, 2026·

  ·
  1

  1

• Fix PPO input sharding to match pmap axis

  #674·lmj0209 opened Jul 21, 2026·

  ·

• Fix EpisodeWrapper dropping sub-step metrics during action_repeat (#610)

  #670·SAY-5 opened Apr 17, 2026·

  ·
  1

• Fix SAC checkpoint restore device axis mismatch (#659)

  #669·SAY-5 opened Apr 17, 2026·

  ·
  1

• Warn when init_noise_std and related params are ignored with tanh_normal (#661)

  #668·SAY-5 opened Apr 16, 2026·

  ·
  1

• Security: Remediation of RCE via Insecure Pickle Deserialization in Model I/O

  #667·JoshuaProvoste opened Apr 5, 2026·

  ·

• Upgrade GitHub Actions for Node 24 compatibility

  #666·salmanmkc opened Mar 11, 2026·

  ·

• Fix checkpoint.load_config

  #665·fengzileee opened Mar 9, 2026·

  ·
  1

• Support callable entropy schedules in PPO

  #660·PhysicistJohn opened Feb 26, 2026·

  ·
  4

• math: scan solve_pgs iterations

  #658·chai1any opened Jan 24, 2026·

  ·

• simple history encoder for brax

  #651·samindaa opened Dec 2, 2025·

  ·
  1

• Fix #647 gradient_update_fn compatible with AdamW

  #648·ShanningZhuang opened Nov 25, 2025·

  ·
  1

• Add _get_obs to fast environment

  #644·YannBerthelot opened Oct 15, 2025·

  ·

• Store terminal observations in AutoResetWrapper infos for value bootstrap

  #641·YannBerthelot opened Sep 25, 2025·

  ·
  2

• policy_params_fn

  #636·jamesheald opened Sep 9, 2025·

  ·

• Add real-time rendering callback

  #634·bsyh opened Aug 30, 2025·

  ·
  15

• Add support for initial alpha value for SAC

  #598·varunagrawal opened Apr 18, 2025·

  ·
  6

• TD3 baseline

  #499·anasbekheit opened Jul 5, 2024·

  ·
  1

  

• Switch gym dependency to gymnasium

  #486·giotherobot opened May 14, 2024·

  ·
  9

• io/image: Replace PyTinyrenderer with a Pure JAX renderer

  #367·JoeyTeng opened Jun 2, 2023·

  ·
  7

• Agents: Short-Horizon Actor Critic

  #262·peabody124 opened Nov 19, 2022Contributor·

  ·
  16