All pull requests

Pull requests

• Synchronize replay tensor references after loading

  #482·yunyv opened Oct 20, 2026·

  ·

• Fix occupied size of multidimensional MultiDiscrete spaces

  #481·yunyv opened Oct 19, 2026·

  ·

• Retain remainder transitions when sampling all memory

  #476·antonsoo opened Oct 9, 2026Contributor·

  ·
  1

• Compare complete data shapes when reusing memory tensors

  #473·aalquwayfili opened Oct 8, 2026·

  ·
  1

• Use population batch variance when merging running moments

  #471·Afloat16 opened Oct 7, 2026Contributor·

  ·
  1

• Restore the learned SAC entropy coefficient on resume

  #465·Afloat16 opened Oct 7, 2026Contributor·

  ·
  1

• Fix finished-episode tracking across all agents

  #450·tandede opened Aug 25, 2026·

  ·
  1

• Fix missing Weights & Biases metric logging for tracking data

  #446·marina-aoyama opened Jun 10, 2026·

  ·

• fix(postprocessing): correct boolean parsing in MemoryFileIterator CSV reader

  #417·awanawana opened Mar 17, 2026·

  ·

• Add exporter utils for model exporting (currently only implemented for PyTorch)

  #410·Jadeiin opened Jan 24, 2026·

  ·
  2

• Update memory_index once per batch, not per tensor

  #328·AlexD15216 opened Jun 25, 2025Contributor·

  ·

• Fix WandB model config logging for JAX agents

  #312·andregonz opened Apr 16, 2025·

  ·

• Beta Policy

  #307·AntoineRichard opened Apr 3, 2025·

  ·

• Implementation of a CrossQ agent in Jax and PyTorch

  #283·Soralsei opened Feb 26, 2025·

  ·
  1

• Add support for per-agent state_space in MAPPO

  #274·bikcrum opened Feb 17, 2025·

  ·

• Do not save checkpoints for eval runs

  #154·lopatovsky opened Jun 10, 2024Contributor·

  ·