Skip to content

Commit d09085d

Browse files
shihzyawjuliani
authored andcommitted
Added optional instructions (#192)
Setting these parameters will produce a very well trained model within 5-10 minutes for this exercise. The thought is to unblock a first time user.
1 parent d57cc29 commit d09085d

File tree

1 file changed

+3
-2
lines changed

1 file changed

+3
-2
lines changed

docs/Getting-Started-with-Balance-Ball.md

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -51,8 +51,9 @@ In order to train the agents within the Ball Balance environment:
5151

5252
1. Open `python/PPO.ipynb` notebook from Jupyter.
5353
2. Set `env_name` to the name of your environment file earlier.
54-
3. (optional) Set `run_path` directory to your choice.
55-
4. Run all cells of notebook with the exception of the last one under "Export the trained Tensorflow graph."
54+
3. (optional) In order to get the best results quickly, set `max_steps` to 50000, set `buffer_size` to 5000, and set `batch_size` to 512. For this exercise, this will train the model in approximately ~5-10 minutes.
55+
4. (optional) Set `run_path` directory to your choice.
56+
5. Run all cells of notebook with the exception of the last one under "Export the trained Tensorflow graph."
5657

5758
### Observing Training Progress
5859
In order to observe the training process in more detail, you can use Tensorboard.

0 commit comments

Comments
 (0)