Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 3904 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1351. Min Act: -0.0001

Data Index: 275084 (C4 (Web Text))

Max Activating Token Index: 806

Text #1

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1137. Min Act: -0.0001

Data Index: 1576886 (Python Code)

Max Activating Token Index: 328

Text #2

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1093. Min Act: -0.0001

Data Index: 1650731 (Python Code)

Max Activating Token Index: 675

Text #3

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1038. Min Act: -0.0001

Data Index: 1576110 (Python Code)

Max Activating Token Index: 236

Text #4

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1033. Min Act: -0.0001

Data Index: 479549 (C4 (Web Text))

Max Activating Token Index: 806

Text #5

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1161. Min Act: -0.0001

Data Index: 129967 (C4 (Web Text))

Max Activating Token Index: 233

Text #6

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1014. Min Act: -0.0001

Data Index: 1305742 (C4 (Web Text))

Max Activating Token Index: 941

Text #7

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1001. Min Act: -0.0001

Data Index: 737634 (C4 (Web Text))

Max Activating Token Index: 857

Text #8

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0992. Min Act: -0.0001

Data Index: 1557838 (Python Code)

Max Activating Token Index: 3

Text #9

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1102. Min Act: -0.0001

Data Index: 179757 (C4 (Web Text))

Max Activating Token Index: 942

Text #10

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1058. Min Act: -0.0001

Data Index: 256125 (C4 (Web Text))

Max Activating Token Index: 668

Text #11

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.1054. Min Act: -0.0001

Data Index: 249359 (C4 (Web Text))

Max Activating Token Index: 726

Text #12

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0953. Min Act: -0.0001

Data Index: 38776 (C4 (Web Text))

Max Activating Token Index: 900

Text #13

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0982. Min Act: -0.0001

Data Index: 1646097 (Python Code)

Max Activating Token Index: 960

Text #14

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0977. Min Act: -0.0001

Data Index: 333059 (C4 (Web Text))

Max Activating Token Index: 329

Text #15

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0921. Min Act: -0.0001

Data Index: 601110 (C4 (Web Text))

Max Activating Token Index: 358

Text #16

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0881. Min Act: -0.0001

Data Index: 1122490 (C4 (Web Text))

Max Activating Token Index: 767

Text #17

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0900. Min Act: -0.0001

Data Index: 1157537 (C4 (Web Text))

Max Activating Token Index: 1022

Text #18

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0867. Min Act: -0.0001

Data Index: 547068 (C4 (Web Text))

Max Activating Token Index: 182

Text #19

Max Range: 0.1351. Min Range: -0.1351

Max Act: 0.0870. Min Act: -0.0001

Data Index: 780391 (C4 (Web Text))

Max Activating Token Index: 687