Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 106 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2555. Min Act: -0.0001

Data Index: 359780 (C4 (Web Text))

Max Activating Token Index: 509

Text #1

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2371. Min Act: -0.0001

Data Index: 308491 (C4 (Web Text))

Max Activating Token Index: 381

Text #2

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2290. Min Act: -0.0001

Data Index: 770249 (C4 (Web Text))

Max Activating Token Index: 390

Text #3

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2383. Min Act: -0.0001

Data Index: 344343 (C4 (Web Text))

Max Activating Token Index: 859

Text #4

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2325. Min Act: -0.0001

Data Index: 575986 (C4 (Web Text))

Max Activating Token Index: 494

Text #5

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2294. Min Act: -0.0000

Data Index: 1278461 (C4 (Web Text))

Max Activating Token Index: 612

Text #6

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2277. Min Act: -0.0001

Data Index: 154980 (C4 (Web Text))

Max Activating Token Index: 791

Text #7

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2415. Min Act: -0.0001

Data Index: 1637527 (Python Code)

Max Activating Token Index: 984

Text #8

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2248. Min Act: -0.0001

Data Index: 402766 (C4 (Web Text))

Max Activating Token Index: 247

Text #9

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2227. Min Act: -0.0001

Data Index: 823656 (C4 (Web Text))

Max Activating Token Index: 56

Text #10

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2145. Min Act: -0.0001

Data Index: 139337 (C4 (Web Text))

Max Activating Token Index: 493

Text #11

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2270. Min Act: -0.0001

Data Index: 963694 (C4 (Web Text))

Max Activating Token Index: 114

Text #12

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2351. Min Act: -0.0001

Data Index: 1494091 (Python Code)

Max Activating Token Index: 376

Text #13

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2186. Min Act: -0.0001

Data Index: 198398 (C4 (Web Text))

Max Activating Token Index: 46

Text #14

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2153. Min Act: -0.0001

Data Index: 1195094 (C4 (Web Text))

Max Activating Token Index: 37

Text #15

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2006. Min Act: -0.0001

Data Index: 1558016 (Python Code)

Max Activating Token Index: 756

Text #16

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2135. Min Act: -0.0001

Data Index: 1206984 (C4 (Web Text))

Max Activating Token Index: 324

Text #17

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2126. Min Act: -0.0001

Data Index: 865045 (C4 (Web Text))

Max Activating Token Index: 203

Text #18

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2056. Min Act: -0.0001

Data Index: 20902 (C4 (Web Text))

Max Activating Token Index: 303

Text #19

Max Range: 0.2555. Min Range: -0.2555

Max Act: 0.2112. Min Act: -0.0000

Data Index: 1239005 (C4 (Web Text))

Max Activating Token Index: 150