-
Notifications
You must be signed in to change notification settings - Fork 19
Logs28: Mutual Information Observe
Higepon Taro Minowa edited this page Jun 28, 2018
·
6 revisions
| Log Type | Detail |
|---|---|
| 1: What specific output am I working on right now? | Ran RL for real data, see if things are as expected. |
| 2: Thinking out loud - hypotheses about the current problem - what to work on next - how can I verify |
There may be a bug which I don't know yet. If we observe some key data, we should know them. |
| 3: A record of currently ongoing runs along with a short reminder of what question each run is supposed to answer | - Do key values of RL look good? - Does actual reply look good? - Are beam results diverse enough? |
| 4: Results of runs and conclusion | - Reward went up and it hit upper limit looks okay. |
| 5: Next steps | Run with big data. |
| 6: mega.nz |