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

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

Neuron 2777 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: 1.3160. Min Range: -1.3160

Max Act: 1.3160. Min Act: -0.0001

Data Index: 784250 (C4 (Web Text))

Max Activating Token Index: 33

Text #1

Max Range: 1.3160. Min Range: -1.3160

Max Act: 1.1492. Min Act: -0.0001

Data Index: 1107962 (C4 (Web Text))

Max Activating Token Index: 28

Text #2

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.9988. Min Act: -0.0001

Data Index: 353130 (C4 (Web Text))

Max Activating Token Index: 21

Text #3

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.8667. Min Act: -0.0001

Data Index: 338912 (C4 (Web Text))

Max Activating Token Index: 25

Text #4

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7541. Min Act: -0.0001

Data Index: 568016 (C4 (Web Text))

Max Activating Token Index: 37

Text #5

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.6687. Min Act: -0.0001

Data Index: 216924 (C4 (Web Text))

Max Activating Token Index: 181

Text #6

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7814. Min Act: -0.0001

Data Index: 170356 (C4 (Web Text))

Max Activating Token Index: 593

Text #7

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7084. Min Act: -0.0001

Data Index: 1499197 (Python Code)

Max Activating Token Index: 745

Text #8

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7245. Min Act: -0.0001

Data Index: 1271881 (C4 (Web Text))

Max Activating Token Index: 109

Text #9

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7298. Min Act: -0.0001

Data Index: 193456 (C4 (Web Text))

Max Activating Token Index: 27

Text #10

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7508. Min Act: -0.0001

Data Index: 1063472 (C4 (Web Text))

Max Activating Token Index: 45

Text #11

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7065. Min Act: -0.0001

Data Index: 320741 (C4 (Web Text))

Max Activating Token Index: 670

Text #12

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7182. Min Act: -0.0001

Data Index: 40747 (C4 (Web Text))

Max Activating Token Index: 892

Text #13

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7330. Min Act: -0.0001

Data Index: 861595 (C4 (Web Text))

Max Activating Token Index: 973

Text #14

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.6772. Min Act: -0.0001

Data Index: 615123 (C4 (Web Text))

Max Activating Token Index: 124

Text #15

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.6912. Min Act: -0.0001

Data Index: 1642970 (Python Code)

Max Activating Token Index: 30

Text #16

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7052. Min Act: -0.0001

Data Index: 91984 (C4 (Web Text))

Max Activating Token Index: 165

Text #17

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7693. Min Act: -0.0001

Data Index: 916392 (C4 (Web Text))

Max Activating Token Index: 83

Text #18

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.7070. Min Act: -0.0001

Data Index: 584299 (C4 (Web Text))

Max Activating Token Index: 24

Text #19

Max Range: 1.3160. Min Range: -1.3160

Max Act: 0.6745. Min Act: -0.0001

Data Index: 486870 (C4 (Web Text))

Max Activating Token Index: 371