diff --git a/docs/_static/img/split_graph.png b/docs/_static/img/split_graph.png new file mode 100644 index 0000000..c966c55 Binary files /dev/null and b/docs/_static/img/split_graph.png differ diff --git a/docs/guide/tensorboard.rst b/docs/guide/tensorboard.rst index dc39d8b..625c1be 100644 --- a/docs/guide/tensorboard.rst +++ b/docs/guide/tensorboard.rst @@ -26,10 +26,20 @@ You can also define custom logging name when training (by default it is the algo model.learn(total_timesteps=10_000, tb_log_name="first_run") # Pass reset_num_timesteps=False to continue the training curve in tensorboard # By default, it will create a new curve + # Keep tb_log_name constant to have continuous curve (see note below) model.learn(total_timesteps=10_000, tb_log_name="second_run", reset_num_timesteps=False) model.learn(total_timesteps=10_000, tb_log_name="third_run", reset_num_timesteps=False) +.. note:: + If you specify different ``tb_log_name`` in subsequent runs, you will have split graphs, like in the figure below. + If you want them to be continuous, you must keep the same ``tb_log_name`` (see `issue #975 `_). + And, if you still managed to get your graphs split by other means, just put tensorboard log files into the same folder. + + .. image:: ../_static/img/split_graph.png + :width: 330 + :alt: split_graph + Once the learn function is called, you can monitor the RL agent during or after the training, with the following bash command: .. code-block:: bash diff --git a/docs/misc/changelog.rst b/docs/misc/changelog.rst index b1ed3b1..6a42744 100644 --- a/docs/misc/changelog.rst +++ b/docs/misc/changelog.rst @@ -29,6 +29,7 @@ Others: Documentation: ^^^^^^^^^^^^^^ - Fix typo in docstring "nature" -> "Nature" (@Melanol) +- Add info on split tensorboard logs into (@Melanol) Release 1.6.0 (2022-07-11)