Apple tracking #11
Apple tracking #11
Conversation
… apple, and includes a script for parsing STDOUT and collecting the results into a csv file. Added the Not the best solution for apple tracking, but it does work. Unfortunately it might slow down evaluations due to the print spam.
|
Oh. But this is just the reward. I thought that we were going to track if the fields are depleted |
|
No. Peter was doing apple tracking from the start . We discuss it a few weeks ago and the goal was that , but Im not sure if you had another agreement. |
|
As I had told you and written in the document, apple tracking is /exactly/ the same as tracking reward. The only case where apple tracking makes sense is if you are tracking the amount of apples alive (not the amount of apples consumed). If feels weird that after having pointed this out multiple times the plan did not change, specially given the troubles implementing it. |
|
Many things are said and I guess people implement and do at the best of their abilities. When this task was made some metrics were defined under a meeting you where not part of that based on apple tracking would define some things . |
|
@rockretep don't erase or close. We will definetly decouple apple gathering from reward in future iterations for safety study concerns so this is useful even if this slows down evals a bit. |
Enables apple tracking by simply printing out when an avatar eats an apple, and includes a script for parsing STDOUT and collecting the results into a csv file. Updated the configuration for two substrates to support it, but substrate configurations can easily be updated to include it. Not the best solution for apple tracking, but it does work. Unfortunately it might slow down evaluations due to the print spam.