Model: SoLU Model: 4 Layers, 2048 Neurons per Layer

Dataset: 80% C4 (Web Text) and 20% Python Code

Neuron 1169 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-4l')

Text #0

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1501. Min Act: -0.0002

Data Index: 1176120 (C4 (Web Text))

Max Activating Token Index: 309

Text #1

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1495. Min Act: -0.0002

Data Index: 41178 (C4 (Web Text))

Max Activating Token Index: 926

Text #2

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1478. Min Act: -0.0002

Data Index: 1154285 (C4 (Web Text))

Max Activating Token Index: 693

Text #3

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1420. Min Act: -0.0002

Data Index: 111389 (C4 (Web Text))

Max Activating Token Index: 648

Text #4

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1376. Min Act: -0.0002

Data Index: 1113645 (C4 (Web Text))

Max Activating Token Index: 284

Text #5

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1361. Min Act: -0.0002

Data Index: 203617 (C4 (Web Text))

Max Activating Token Index: 811

Text #6

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1389. Min Act: -0.0002

Data Index: 39598 (C4 (Web Text))

Max Activating Token Index: 260

Text #7

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1233. Min Act: -0.0001

Data Index: 1642910 (Python Code)

Max Activating Token Index: 22

Text #8

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1308. Min Act: -0.0002

Data Index: 1201899 (C4 (Web Text))

Max Activating Token Index: 713

Text #9

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1230. Min Act: -0.0001

Data Index: 1622985 (Python Code)

Max Activating Token Index: 71

Text #10

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1236. Min Act: -0.0002

Data Index: 854929 (C4 (Web Text))

Max Activating Token Index: 191

Text #11

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1118. Min Act: -0.0001

Data Index: 1478605 (Python Code)

Max Activating Token Index: 56

Text #12

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1149. Min Act: -0.0001

Data Index: 1625988 (Python Code)

Max Activating Token Index: 24

Text #13

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1240. Min Act: -0.0002

Data Index: 538679 (C4 (Web Text))

Max Activating Token Index: 661

Text #14

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1190. Min Act: -0.0002

Data Index: 15012 (C4 (Web Text))

Max Activating Token Index: 241

Text #15

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1126. Min Act: -0.0002

Data Index: 509163 (C4 (Web Text))

Max Activating Token Index: 411

Text #16

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1116. Min Act: -0.0002

Data Index: 1305879 (C4 (Web Text))

Max Activating Token Index: 436

Text #17

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1149. Min Act: -0.0002

Data Index: 857250 (C4 (Web Text))

Max Activating Token Index: 725

Text #18

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1146. Min Act: -0.0002

Data Index: 2670 (C4 (Web Text))

Max Activating Token Index: 583

Text #19

Max Range: 0.1501. Min Range: -0.1501

Max Act: 0.1197. Min Act: -0.0002

Data Index: 20120 (C4 (Web Text))

Max Activating Token Index: 206