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

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

Neuron 168 in Layer 8

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 2.5837. Min Range: -2.5837

Max Act: 2.5837. Min Act: -0.0000

Data Index: 695804 (C4 (Web Text))

Max Activating Token Index: 761

Text #1

Max Range: 2.5837. Min Range: -2.5837

Max Act: 2.4741. Min Act: -0.0000

Data Index: 302189 (C4 (Web Text))

Max Activating Token Index: 645

Text #2

Max Range: 2.5837. Min Range: -2.5837

Max Act: 2.0103. Min Act: -0.0000

Data Index: 434991 (C4 (Web Text))

Max Activating Token Index: 285

Text #3

Max Range: 2.5837. Min Range: -2.5837

Max Act: 2.0734. Min Act: -0.0000

Data Index: 267296 (C4 (Web Text))

Max Activating Token Index: 789

Text #4

Max Range: 2.5837. Min Range: -2.5837

Max Act: 1.7910. Min Act: -0.0000

Data Index: 666828 (C4 (Web Text))

Max Activating Token Index: 697

Text #5

Max Range: 2.5837. Min Range: -2.5837

Max Act: 1.5715. Min Act: -0.0000

Data Index: 1218743 (C4 (Web Text))

Max Activating Token Index: 931

Text #6

Max Range: 2.5837. Min Range: -2.5837

Max Act: 1.4126. Min Act: -0.0000

Data Index: 1175753 (C4 (Web Text))

Max Activating Token Index: 597

Text #7

Max Range: 2.5837. Min Range: -2.5837

Max Act: 1.1473. Min Act: -0.0000

Data Index: 1040486 (C4 (Web Text))

Max Activating Token Index: 618

Text #8

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.9694. Min Act: -0.0000

Data Index: 162581 (C4 (Web Text))

Max Activating Token Index: 255

Text #9

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.9232. Min Act: -0.0000

Data Index: 48154 (C4 (Web Text))

Max Activating Token Index: 1001

Text #10

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.7580. Min Act: -0.0000

Data Index: 235722 (C4 (Web Text))

Max Activating Token Index: 420

Text #11

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.7463. Min Act: -0.0000

Data Index: 271108 (C4 (Web Text))

Max Activating Token Index: 931

Text #12

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6857. Min Act: -0.0000

Data Index: 653380 (C4 (Web Text))

Max Activating Token Index: 508

Text #13

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6872. Min Act: -0.0000

Data Index: 1014863 (C4 (Web Text))

Max Activating Token Index: 91

Text #14

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6716. Min Act: -0.0000

Data Index: 389795 (C4 (Web Text))

Max Activating Token Index: 855

Text #15

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6743. Min Act: -0.0000

Data Index: 1028553 (C4 (Web Text))

Max Activating Token Index: 830

Text #16

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6683. Min Act: -0.0000

Data Index: 626020 (C4 (Web Text))

Max Activating Token Index: 720

Text #17

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6382. Min Act: -0.0000

Data Index: 1393679 (Python Code)

Max Activating Token Index: 103

Text #18

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6718. Min Act: -0.0000

Data Index: 1213866 (C4 (Web Text))

Max Activating Token Index: 561

Text #19

Max Range: 2.5837. Min Range: -2.5837

Max Act: 0.6480. Min Act: -0.0000

Data Index: 60729 (C4 (Web Text))

Max Activating Token Index: 504