1 Release published by 1 person

• v2.0.0b1 Beta release of 2.0.0

  published Jul 15, 2025

1 Pull request merged by 1 person

• Tianshou v2

  #1259 merged Jul 15, 2025

14 Issues closed by 3 people

• Remove kwargs in learn and update

  #949 closed Jul 15, 2025

• Refactoring: make Trainer more flexible

  #1204 closed Jul 15, 2025

• Basepolicy requires replay buffer restricting modularity.

  #898 closed Jul 15, 2025

• Change all signatures for policy.process_fn and policy.post_process_fn from BatchProtocol to RolloutBatchProtocol

  #919 closed Jul 15, 2025

• Remove kwargs in policies' forward

  #948 closed Jul 15, 2025

• "test_reward" KeyError in epoch_stat["test_reward"] when saving checkpint through trainer information

  #913 closed Jul 15, 2025

• Fix handling of torch "device" association

  #810 closed Jul 15, 2025

• Don't pass optimizers to policies

  #959 closed Jul 15, 2025

• Better interfaces and names for Actor, Critic, Net and other classes

  #1091 closed Jul 15, 2025

• DDPG Policy Actor Gradient Question

  #1268 closed Jul 14, 2025

• Wrong output of forward for custom policy

  #1029 closed Jul 10, 2025

• Adjust locations of setting the policy in train/eval mode

  #1122 closed Jul 10, 2025

• atari_dqn.py Training Speed Drops Significantly Over Time

  #1265 closed Jul 10, 2025

• SACPolicy Unsupported action space: Box(-1.0, 1.0, (1,), float32)")

  #1232 closed Jul 3, 2025

3 Issues opened by 3 people

• Enhancement: Support Batch Input For Net Class

  #1270 opened Jul 15, 2025

• link to contributing docs is broken

  #1269 opened Jul 12, 2025

• May I ask — most papers report algorithm baselines using average return, but yours uses best return. How can I change it to report average return instead?

  #1267 opened Jul 11, 2025

10 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• ValueError: Initial obs and info should not be None. Either reset the collector

  #1266 commented on Jul 10, 2025 • 0 new comments

• Support Dict observation spaces

  #1065 commented on Jul 10, 2025 • 0 new comments

• Use Altair inside a notebook to display benchmark results

  #1136 commented on Jul 12, 2025 • 0 new comments

• MPO Implementation

  #1165 commented on Jul 18, 2025 • 0 new comments

• Feat/collect equal episode num in all envs

  #1127 commented on Jun 27, 2025 • 0 new comments

• Bump setuptools from 68.2.2 to 70.0.0

  #1174 commented on Jun 27, 2025 • 0 new comments

• Bump jinja2 from 3.1.4 to 3.1.6

  #1252 commented on Jun 27, 2025 • 0 new comments

• Bump ray from 2.8.0 to 2.43.0

  #1253 commented on Jun 27, 2025 • 0 new comments

• Bump torch from 2.1.1 to 2.4.0

  #1254 commented on Jun 27, 2025 • 0 new comments

• Improve offline algo performance

  #1261 commented on Jul 15, 2025 • 0 new comments