[rllib] Reorganize trainer config, add warnings about high VF loss magnitude for PPO (#6181)

2026-08-20 12:40:44 +08:00 · 2019-11-18 10:39:07 -08:00
parent 2965dc1b72
commit 8fc2272f43
3 changed files with 162 additions and 117 deletions
@@ -173,7 +173,7 @@ You can configure experience output for an agent using the following options:
 .. literalinclude:: ../../rllib/agents/trainer.py
   :language: python
   :start-after: shuffle_buffer_size
-   :end-before: === Multiagent ===
+   :end-before: Settings for Multi-Agent Environments

 The interface for a custom output writer is as follows: