v.0.4.0

New Acme release.

PiperOrigin-RevId: 427654046
Change-Id: Idac621a962ebcdbf9954b14cfea86991d765a13d

v0.3.0

Allow to specify the logger for PPO agents.

PiperOrigin-RevId: 420289254
Change-Id: Iec00166e2e86308ca06bfd916549bed7c3c6c2e7

0.2.4

Release new PyPi version.

PiperOrigin-RevId: 415230385
Change-Id: Iade47e6c6762c2dfbd485f67fbef128df623cbf3

0.2.3

Release new PyPi version.

PiperOrigin-RevId: 413885824
Change-Id: I744416c02347a1f61b32dc1750260e35b39467af

0.2.2

Highlights

• This is a minor release to update the dependencies to new stable versions of Tensorflow and Reverb (14cb62f).
• New SVG-0 agent (1ecabdb).

Minor changes and fixes

• Fixes to docs and testing.
• Minor updates to tf-DQN.
• Updates to LP tear down.
• Update to how signals are handled.
• Update to setup/requirements.
• Reorganize Builder and integrate jax-specific functionality.
• Improvements to typing.

0.2.1

Highlights

• Released distributed versions of some agents 3bc0426.
• Adders use Reverb's TrajectoryWriter (instead of Writer).
• Introduce the Builder API 329aae5.
• Add JAX behavior cloning agent eb7073f.
• Add Python 3.8 support.

Minor changes and fixes

• Use nightly versions of TensorFlow, Reverb, and TensorFlow Probability.
• Remove pandas dependency.
• Skip tests that depend on Atari ROMs (atari_py no longer provides them as part of its pypi package).
• Fix links to ipynb pages.

0.2.0

Highlights

• Using stable releases for TensorFlow (>=2.3.0), Reverb, and TensorFlow Probability.
• Added Critic Regularized Regression (code, paper)
• Added Discrete Batch-Constrained Deep Q-learning (code, paper)
• Added EnvironmentLoop.run_episode() for running a single episode.
• Update EnvironmentLoop.run() to take num_steps, allowing the control of step count rather than just episode count.
• Add more distribution types (e.g. GaussianMixture) which can be used by policies.
• Added a environment wrapper for action repeats.
• Improvements/tuning to datasets exposed by make_dataset.
• Add support for nested / multidimensional rewards and discounts.

Minor changes and fixes

• ConstantInfo logger for logging constant information.
• Added a should_update parameter to the EnvironmentLoop.
• Various modifications and optimizations to the make_reverb_dataset() function.
• Improvements to typing and pytype usage.
• Other minor bug and documentation fixes.

0.1.8

Minor version to fix a mismatch inversions between tf/tfp.

Other changes include:

• a jax R2D2 agent.
• movement of various utility/helper functions for additional clarity.
• doc fixes.
• bugfix for the mcts agent.

0.1.7

Bump version again (0.1.6 tag was associated with wrong commit).

PiperOrigin-RevId: 318255261
Change-Id: Ie5f644e7fed89c212d9015059bf6e30acd890763

0.1.6

Fix ZeroDiscountonLifeLoss wrapper and add test covereage.

PiperOrigin-RevId: 318249207
Change-Id: I01c2f0b76d1b15fc42bf83538fbb1bc8c0757776