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

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

Neuron 68 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0157. Min Act: -0.0001

Data Index: 366526 (C4 (Web Text))

Max Activating Token Index: 53

Text #1

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0122. Min Act: -0.0001

Data Index: 556563 (C4 (Web Text))

Max Activating Token Index: 987

Text #2

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0115. Min Act: -0.0001

Data Index: 1307244 (C4 (Web Text))

Max Activating Token Index: 359

Text #3

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0130. Min Act: -0.0000

Data Index: 952657 (C4 (Web Text))

Max Activating Token Index: 730

Text #4

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0105. Min Act: -0.0001

Data Index: 385652 (C4 (Web Text))

Max Activating Token Index: 158

Text #5

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0113. Min Act: -0.0000

Data Index: 370681 (C4 (Web Text))

Max Activating Token Index: 636

Text #6

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0115. Min Act: -0.0000

Data Index: 594648 (C4 (Web Text))

Max Activating Token Index: 816

Text #7

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0112. Min Act: -0.0001

Data Index: 1236544 (C4 (Web Text))

Max Activating Token Index: 234

Text #8

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0107. Min Act: -0.0001

Data Index: 1021374 (C4 (Web Text))

Max Activating Token Index: 633

Text #9

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0110. Min Act: -0.0001

Data Index: 721135 (C4 (Web Text))

Max Activating Token Index: 444

Text #10

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0110. Min Act: -0.0001

Data Index: 362146 (C4 (Web Text))

Max Activating Token Index: 874

Text #11

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0109. Min Act: -0.0000

Data Index: 1378416 (Python Code)

Max Activating Token Index: 75

Text #12

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0097. Min Act: -0.0001

Data Index: 337550 (C4 (Web Text))

Max Activating Token Index: 941

Text #13

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0103. Min Act: -0.0001

Data Index: 992088 (C4 (Web Text))

Max Activating Token Index: 24

Text #14

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0101. Min Act: -0.0000

Data Index: 1049053 (C4 (Web Text))

Max Activating Token Index: 709

Text #15

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0109. Min Act: -0.0000

Data Index: 271868 (C4 (Web Text))

Max Activating Token Index: 687

Text #16

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0098. Min Act: -0.0001

Data Index: 237968 (C4 (Web Text))

Max Activating Token Index: 226

Text #17

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0106. Min Act: -0.0001

Data Index: 1190634 (C4 (Web Text))

Max Activating Token Index: 970

Text #18

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0098. Min Act: -0.0000

Data Index: 518695 (C4 (Web Text))

Max Activating Token Index: 142

Text #19

Max Range: 0.0157. Min Range: -0.0157

Max Act: 0.0101. Min Act: -0.0001

Data Index: 136850 (C4 (Web Text))

Max Activating Token Index: 184