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

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

Neuron 3917 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1717. Min Act: -0.0000

Data Index: 886514 (C4 (Web Text))

Max Activating Token Index: 313

Text #1

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1491. Min Act: -0.0000

Data Index: 318772 (C4 (Web Text))

Max Activating Token Index: 269

Text #2

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1455. Min Act: -0.0000

Data Index: 816152 (C4 (Web Text))

Max Activating Token Index: 237

Text #3

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1364. Min Act: -0.0000

Data Index: 652701 (C4 (Web Text))

Max Activating Token Index: 762

Text #4

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1337. Min Act: -0.0000

Data Index: 321711 (C4 (Web Text))

Max Activating Token Index: 62

Text #5

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1321. Min Act: -0.0000

Data Index: 370645 (C4 (Web Text))

Max Activating Token Index: 934

Text #6

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1298. Min Act: -0.0000

Data Index: 1196596 (C4 (Web Text))

Max Activating Token Index: 822

Text #7

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1264. Min Act: -0.0000

Data Index: 1619675 (Python Code)

Max Activating Token Index: 738

Text #8

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1254. Min Act: -0.0000

Data Index: 788178 (C4 (Web Text))

Max Activating Token Index: 877

Text #9

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1303. Min Act: -0.0000

Data Index: 91660 (C4 (Web Text))

Max Activating Token Index: 90

Text #10

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1235. Min Act: -0.0000

Data Index: 972348 (C4 (Web Text))

Max Activating Token Index: 342

Text #11

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1235. Min Act: -0.0000

Data Index: 1647570 (Python Code)

Max Activating Token Index: 517

Text #12

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1229. Min Act: -0.0000

Data Index: 1068089 (C4 (Web Text))

Max Activating Token Index: 191

Text #13

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1096. Min Act: -0.0000

Data Index: 692632 (C4 (Web Text))

Max Activating Token Index: 799

Text #14

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1174. Min Act: -0.0000

Data Index: 655444 (C4 (Web Text))

Max Activating Token Index: 546

Text #15

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1119. Min Act: -0.0000

Data Index: 44592 (C4 (Web Text))

Max Activating Token Index: 650

Text #16

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1047. Min Act: -0.0000

Data Index: 185739 (C4 (Web Text))

Max Activating Token Index: 755

Text #17

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1164. Min Act: -0.0000

Data Index: 128153 (C4 (Web Text))

Max Activating Token Index: 656

Text #18

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1188. Min Act: -0.0000

Data Index: 1385236 (Python Code)

Max Activating Token Index: 298

Text #19

Max Range: 0.1717. Min Range: -0.1717

Max Act: 0.1178. Min Act: -0.0000

Data Index: 156855 (C4 (Web Text))

Max Activating Token Index: 311