interface

Trajectories

1
2
3
4
5
6
7
8

Bookmarks

Lowest advantage
episodes
(unexpected failures):

trajectory 8, frame 118
trajectory 6, frame 471
trajectory 5, frame 59
trajectory 5, frame 329
trajectory 7, frame 107
trajectory 2, frame 405
trajectory 7, frame 354
trajectory 2, frame 225
trajectory 6, frame 121
trajectory 3, frame 370
trajectory 4, frame 23
trajectory 7, frame 226
trajectory 3, frame 20
trajectory 5, frame 198
trajectory 8, frame 268
trajectory 1, frame 250

Highest advantage
episodes
(unexpected successes):

trajectory 5, frame 485
trajectory 6, frame 240
trajectory 3, frame 288
trajectory 5, frame 241
trajectory 8, frame 76
trajectory 1, frame 126
trajectory 2, frame 424
trajectory 1, frame 218
trajectory 6, frame 6
trajectory 6, frame 29
trajectory 6, frame 365
trajectory 4, frame 175
trajectory 1, frame 347
trajectory 5, frame 257
trajectory 8, frame 497
trajectory 4, frame 78

Layers

Timeline

		frame: 1	policy:	next action: →
no-op → ← ↑ ↗ ↖ ↓ A B

fps

advantage

0.0394

value function

9.35

Attribution

overlay on observations combine positive and negative attribution show residual feature

Observation

Positive attribution

Negative attribution

value function

policy logits:
no-op → ← ↑ ↗ ↖ ↓ A B

sums of policy logits:
→ | ↗ ← | ↖ ↑ | ↗ | ↖ no-op | A | B

Attribution legend

Click to expand feature
Hover to isolate

not
shown

residual
(everything
else)

Hotkeys

go backwards
go forwards
toggle play/pause

Select a feature

Feature visualization

zoom in		zoom out
fewer patches		more patches