Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 2568 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2813. Min Act: -0.0001

Data Index: 459050 (C4 (Web Text))

Max Activating Token Index: 637

Text #1

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2422. Min Act: -0.0001

Data Index: 562947 (C4 (Web Text))

Max Activating Token Index: 759

Text #2

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2473. Min Act: -0.0001

Data Index: 1075054 (C4 (Web Text))

Max Activating Token Index: 248

Text #3

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2100. Min Act: -0.0001

Data Index: 62105 (C4 (Web Text))

Max Activating Token Index: 703

Text #4

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2193. Min Act: -0.0001

Data Index: 1451224 (Python Code)

Max Activating Token Index: 389

Text #5

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2241. Min Act: -0.0001

Data Index: 1085837 (C4 (Web Text))

Max Activating Token Index: 163

Text #6

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.2106. Min Act: -0.0001

Data Index: 581616 (C4 (Web Text))

Max Activating Token Index: 335

Text #7

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1780. Min Act: -0.0001

Data Index: 1585179 (Python Code)

Max Activating Token Index: 229

Text #8

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1851. Min Act: -0.0001

Data Index: 467214 (C4 (Web Text))

Max Activating Token Index: 989

Text #9

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1910. Min Act: -0.0001

Data Index: 167596 (C4 (Web Text))

Max Activating Token Index: 29

Text #10

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1784. Min Act: -0.0001

Data Index: 1462842 (Python Code)

Max Activating Token Index: 521

Text #11

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1799. Min Act: -0.0001

Data Index: 1425084 (Python Code)

Max Activating Token Index: 83

Text #12

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1783. Min Act: -0.0001

Data Index: 1514612 (Python Code)

Max Activating Token Index: 143

Text #13

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1723. Min Act: -0.0001

Data Index: 1532676 (Python Code)

Max Activating Token Index: 203

Text #14

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1649. Min Act: -0.0001

Data Index: 495788 (C4 (Web Text))

Max Activating Token Index: 747

Text #15

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1719. Min Act: -0.0001

Data Index: 469671 (C4 (Web Text))

Max Activating Token Index: 127

Text #16

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1768. Min Act: -0.0001

Data Index: 261716 (C4 (Web Text))

Max Activating Token Index: 843

Text #17

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1694. Min Act: -0.0001

Data Index: 950630 (C4 (Web Text))

Max Activating Token Index: 973

Text #18

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1745. Min Act: -0.0001

Data Index: 371082 (C4 (Web Text))

Max Activating Token Index: 370

Text #19

Max Range: 0.2813. Min Range: -0.2813

Max Act: 0.1646. Min Act: -0.0001

Data Index: 1298930 (C4 (Web Text))

Max Activating Token Index: 622