Model: SoLU Model: 10 Layers, 5120 Neurons per Layer

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

Neuron 1936 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-10l')

Text #0

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1424. Min Act: -0.0000

Data Index: 171656 (C4 (Web Text))

Max Activating Token Index: 482

Text #1

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1353. Min Act: -0.0000

Data Index: 331986 (C4 (Web Text))

Max Activating Token Index: 131

Text #2

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1281. Min Act: -0.0000

Data Index: 880007 (C4 (Web Text))

Max Activating Token Index: 253

Text #3

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1249. Min Act: -0.0000

Data Index: 87078 (C4 (Web Text))

Max Activating Token Index: 805

Text #4

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1195. Min Act: -0.0000

Data Index: 574824 (C4 (Web Text))

Max Activating Token Index: 150

Text #5

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1170. Min Act: -0.0000

Data Index: 877889 (C4 (Web Text))

Max Activating Token Index: 780

Text #6

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1175. Min Act: -0.0000

Data Index: 94005 (C4 (Web Text))

Max Activating Token Index: 364

Text #7

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1132. Min Act: -0.0000

Data Index: 835986 (C4 (Web Text))

Max Activating Token Index: 613

Text #8

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1188. Min Act: -0.0001

Data Index: 1411445 (Python Code)

Max Activating Token Index: 252

Text #9

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1113. Min Act: -0.0000

Data Index: 754557 (C4 (Web Text))

Max Activating Token Index: 430

Text #10

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1096. Min Act: -0.0000

Data Index: 881252 (C4 (Web Text))

Max Activating Token Index: 103

Text #11

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1049. Min Act: -0.0000

Data Index: 348547 (C4 (Web Text))

Max Activating Token Index: 356

Text #12

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1122. Min Act: -0.0000

Data Index: 737408 (C4 (Web Text))

Max Activating Token Index: 481

Text #13

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1105. Min Act: -0.0000

Data Index: 687882 (C4 (Web Text))

Max Activating Token Index: 496

Text #14

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1140. Min Act: -0.0000

Data Index: 731345 (C4 (Web Text))

Max Activating Token Index: 351

Text #15

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1129. Min Act: -0.0000

Data Index: 466328 (C4 (Web Text))

Max Activating Token Index: 397

Text #16

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1098. Min Act: -0.0001

Data Index: 1577724 (Python Code)

Max Activating Token Index: 527

Text #17

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1078. Min Act: -0.0000

Data Index: 203951 (C4 (Web Text))

Max Activating Token Index: 879

Text #18

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1031. Min Act: -0.0000

Data Index: 571584 (C4 (Web Text))

Max Activating Token Index: 221

Text #19

Max Range: 0.1424. Min Range: -0.1424

Max Act: 0.1052. Min Act: -0.0000

Data Index: 310968 (C4 (Web Text))

Max Activating Token Index: 257