Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1434 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0732. Min Act: -0.0003

Data Index: 129834 (C4 (Web Text))

Max Activating Token Index: 926

Text #1

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0770. Min Act: -0.0002

Data Index: 1494986 (Python Code)

Max Activating Token Index: 719

Text #2

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0678. Min Act: -0.0003

Data Index: 141063 (C4 (Web Text))

Max Activating Token Index: 920

Text #3

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0639. Min Act: -0.0002

Data Index: 355127 (C4 (Web Text))

Max Activating Token Index: 615

Text #4

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0584. Min Act: -0.0002

Data Index: 272365 (C4 (Web Text))

Max Activating Token Index: 890

Text #5

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0578. Min Act: -0.0002

Data Index: 775794 (C4 (Web Text))

Max Activating Token Index: 174

Text #6

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0507. Min Act: -0.0003

Data Index: 428983 (C4 (Web Text))

Max Activating Token Index: 733

Text #7

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0553. Min Act: -0.0002

Data Index: 1016556 (C4 (Web Text))

Max Activating Token Index: 876

Text #8

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0512. Min Act: -0.0003

Data Index: 155720 (C4 (Web Text))

Max Activating Token Index: 442

Text #9

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0478. Min Act: -0.0003

Data Index: 812960 (C4 (Web Text))

Max Activating Token Index: 124

Text #10

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0434. Min Act: -0.0003

Data Index: 1196559 (C4 (Web Text))

Max Activating Token Index: 1019

Text #11

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0452. Min Act: -0.0003

Data Index: 113532 (C4 (Web Text))

Max Activating Token Index: 963

Text #12

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0405. Min Act: -0.0002

Data Index: 702363 (C4 (Web Text))

Max Activating Token Index: 1019

Text #13

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0432. Min Act: -0.0003

Data Index: 164601 (C4 (Web Text))

Max Activating Token Index: 810

Text #14

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0496. Min Act: -0.0002

Data Index: 334116 (C4 (Web Text))

Max Activating Token Index: 901

Text #15

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0446. Min Act: -0.0003

Data Index: 863863 (C4 (Web Text))

Max Activating Token Index: 282

Text #16

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0455. Min Act: -0.0003

Data Index: 1437390 (Python Code)

Max Activating Token Index: 583

Text #17

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0400. Min Act: -0.0003

Data Index: 1324748 (C4 (Web Text))

Max Activating Token Index: 746

Text #18

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0444. Min Act: -0.0002

Data Index: 1296059 (C4 (Web Text))

Max Activating Token Index: 357

Text #19

Max Range: 0.0770. Min Range: -0.0770

Max Act: 0.0444. Min Act: -0.0003

Data Index: 250386 (C4 (Web Text))

Max Activating Token Index: 280