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

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

Neuron 3824 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1581. Min Act: -0.0001

Data Index: 1623213 (Python Code)

Max Activating Token Index: 846

Text #1

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1678. Min Act: -0.0001

Data Index: 1277998 (C4 (Web Text))

Max Activating Token Index: 674

Text #2

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1591. Min Act: -0.0000

Data Index: 1335137 (C4 (Web Text))

Max Activating Token Index: 917

Text #3

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1321. Min Act: -0.0001

Data Index: 428492 (C4 (Web Text))

Max Activating Token Index: 788

Text #4

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1350. Min Act: -0.0001

Data Index: 1500125 (Python Code)

Max Activating Token Index: 446

Text #5

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1414. Min Act: -0.0001

Data Index: 530459 (C4 (Web Text))

Max Activating Token Index: 837

Text #6

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1316. Min Act: -0.0000

Data Index: 21174 (C4 (Web Text))

Max Activating Token Index: 262

Text #7

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1259. Min Act: -0.0001

Data Index: 456350 (C4 (Web Text))

Max Activating Token Index: 982

Text #8

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1311. Min Act: -0.0000

Data Index: 286739 (C4 (Web Text))

Max Activating Token Index: 60

Text #9

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1187. Min Act: -0.0000

Data Index: 193306 (C4 (Web Text))

Max Activating Token Index: 906

Text #10

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1221. Min Act: -0.0001

Data Index: 1186531 (C4 (Web Text))

Max Activating Token Index: 84

Text #11

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1302. Min Act: -0.0000

Data Index: 1555029 (Python Code)

Max Activating Token Index: 713

Text #12

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1287. Min Act: -0.0000

Data Index: 1487460 (Python Code)

Max Activating Token Index: 979

Text #13

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1290. Min Act: -0.0000

Data Index: 120148 (C4 (Web Text))

Max Activating Token Index: 526

Text #14

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1104. Min Act: -0.0001

Data Index: 792584 (C4 (Web Text))

Max Activating Token Index: 257

Text #15

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1144. Min Act: -0.0001

Data Index: 116523 (C4 (Web Text))

Max Activating Token Index: 417

Text #16

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1204. Min Act: -0.0001

Data Index: 1631531 (Python Code)

Max Activating Token Index: 79

Text #17

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1163. Min Act: -0.0001

Data Index: 1125848 (C4 (Web Text))

Max Activating Token Index: 163

Text #18

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1198. Min Act: -0.0001

Data Index: 1100258 (C4 (Web Text))

Max Activating Token Index: 115

Text #19

Max Range: 0.1678. Min Range: -0.1678

Max Act: 0.1158. Min Act: -0.0001

Data Index: 1267487 (C4 (Web Text))

Max Activating Token Index: 204