Log Template

1: What specific output am I working on right now?

2: Thinking out loud - e.g. hypotheses about the current problem, what to work on next, how can I verify

3: A record of currently ongoing runs along with a short reminder of what question each run is supposed to answer

• run1: title

4: Results of runs (TensorBoard graphs, any other significant observations), separated by type of run (e.g. by the environment the agent is being trained in)

run1

• hparams >
• mega.nz directory: 20180430rl_test_medium7

seq2seq

RL