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

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

Neuron 912 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.6373. Min Act: -0.0001

Data Index: 57310 (C4 (Web Text))

Max Activating Token Index: 785

Text #1

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.6149. Min Act: -0.0001

Data Index: 200208 (C4 (Web Text))

Max Activating Token Index: 308

Text #2

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.5224. Min Act: -0.0001

Data Index: 1341025 (C4 (Web Text))

Max Activating Token Index: 874

Text #3

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.5172. Min Act: -0.0001

Data Index: 146586 (C4 (Web Text))

Max Activating Token Index: 246

Text #4

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.5366. Min Act: -0.0001

Data Index: 1151888 (C4 (Web Text))

Max Activating Token Index: 914

Text #5

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.5103. Min Act: -0.0001

Data Index: 889779 (C4 (Web Text))

Max Activating Token Index: 936

Text #6

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.5039. Min Act: -0.0001

Data Index: 720356 (C4 (Web Text))

Max Activating Token Index: 249

Text #7

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4432. Min Act: -0.0001

Data Index: 1411801 (Python Code)

Max Activating Token Index: 215

Text #8

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4305. Min Act: -0.0001

Data Index: 738987 (C4 (Web Text))

Max Activating Token Index: 117

Text #9

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4356. Min Act: -0.0001

Data Index: 1284993 (C4 (Web Text))

Max Activating Token Index: 133

Text #10

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4462. Min Act: -0.0001

Data Index: 1363954 (Python Code)

Max Activating Token Index: 8

Text #11

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4060. Min Act: -0.0001

Data Index: 561812 (C4 (Web Text))

Max Activating Token Index: 261

Text #12

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4366. Min Act: -0.0001

Data Index: 491008 (C4 (Web Text))

Max Activating Token Index: 107

Text #13

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.3973. Min Act: -0.0001

Data Index: 1159849 (C4 (Web Text))

Max Activating Token Index: 312

Text #14

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4073. Min Act: -0.0001

Data Index: 955428 (C4 (Web Text))

Max Activating Token Index: 983

Text #15

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4052. Min Act: -0.0000

Data Index: 1508153 (Python Code)

Max Activating Token Index: 189

Text #16

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4404. Min Act: -0.0001

Data Index: 886707 (C4 (Web Text))

Max Activating Token Index: 90

Text #17

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4081. Min Act: -0.0001

Data Index: 1014957 (C4 (Web Text))

Max Activating Token Index: 669

Text #18

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4001. Min Act: -0.0001

Data Index: 1445041 (Python Code)

Max Activating Token Index: 821

Text #19

Max Range: 0.6373. Min Range: -0.6373

Max Act: 0.4116. Min Act: -0.0001

Data Index: 13012 (C4 (Web Text))

Max Activating Token Index: 28