Issues

• Learning rate does not decay in examples/vizdoom_ppo.py when using ICM wrapper

  thu-ml/tianshou#1304 · Filter by author TheDansmod opened on Aug 12, 2026

• New feature request: reward-hacking onset monitoring hooks

  thu-ml/tianshou#1302 · Filter by author Aarav500 opened on Jul 10, 2026
  1

• TD3Params policy_noise docstring

  thu-ml/tianshou#1301 · Filter by author RolandStolz opened on Jun 8, 2026

• Reconsider closed issues for Tianshou v3

  thu-ml/tianshou#1299 · Filter by author opcode81 opened on May 9, 2026

• Loading policy from persitence dir

  thu-ml/tianshou#1297 · Filter by author AnsgarKasselmannTUM opened on Apr 14, 2026

• I have idea for predict the future events

  thu-ml/tianshou#1286 · Filter by author Zevene-dot opened on Jan 19, 2026
  1

• The examples under examples/atari in version 2.0.0 run very slowly, and the test reward does not change with training.

  thu-ml/tianshou#1285 · Filter by author zhangrui-wolf opened on Jan 7, 2026

• Buffer observation equals to next observation using ShmemVectorEnv

  thu-ml/tianshou#1281 · Filter by author heitor57 opened on Oct 28, 2025

• Implement GRPO

  thu-ml/tianshou#1271 · Filter by author MischaPanch opened on Aug 1, 2025
  2

  

• 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?

  thu-ml/tianshou#1267 · Filter by author wisestruct opened on Jul 11, 2025

• How to train Hierachical RL in tianshou

  thu-ml/tianshou#1264 · Filter by author fawu opened on Jun 15, 2025

• In MARL (Multi Agent Reinforcement Learning), is it possible for agents to move simultaneously and take different actions?

  thu-ml/tianshou#1251 · Filter by author G-JOEY opened on Mar 4, 2025

• Using graph models with structured env observations

  thu-ml/tianshou#1249 · Filter by author mjmartis opened on Feb 28, 2025
  2

• PettingZoo AECEnv wrapper causing issues with Batch class

  thu-ml/tianshou#1244 · Filter by author encratite opened on Feb 11, 2025
  6

• batch info None in target_q of ddpg

  thu-ml/tianshou#1241 · Filter by author JoshuaSchenk opened on Feb 3, 2025

• record training data for each vectorized environment in SubprocVectorEnv [Need help]

  thu-ml/tianshou#1233 · Filter by author Xiong5Heng opened on Jan 6, 2025

• next_dist of function _target_dist in c51.py maybe not correctly calculated.

  thu-ml/tianshou#1223 · Filter by author musicaudience opened on Oct 22, 2024

• Wandb logger

  thu-ml/tianshou#1222 · Filter by author dev0Guy opened on Oct 14, 2024
  4

• Question on benchmark hyperparameters

  thu-ml/tianshou#1220 · Filter by author RolandStolz opened on Oct 9, 2024
  1

• AttributeError: 'dict' object has no attribute 'terminated' with multi agent env having gymnasium.spaces.Dict observation space

  thu-ml/tianshou#1219 · Filter by author fortminors opened on Oct 7, 2024

• No response after setting render

  thu-ml/tianshou#1217 · Filter by author zdx3578 opened on Sep 21, 2024

• Question about Actor's outputs and selected actions

  thu-ml/tianshou#1216 · Filter by author flammmes opened on Sep 19, 2024

• Support custom scores for Trainer in high-level interfaces

  high-level-api
  thu-ml/tianshou#1205 · Filter by author MischaPanch opened on Aug 14, 2024 · Release 2.0.0

• Suggestion - Redesign RayEnvWorker for Improved Performance

  performance issues
  thu-ml/tianshou#1172 · Filter by author destin-v opened on Jul 10, 2024
  14

  

• MPO Implementation

  new algorithm
  thu-ml/tianshou#1165 · Filter by author ziqiao30 opened on Jul 4, 2024 · Release 2.0.0
  11