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

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

Neuron 3574 in Layer 10

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.9536. Min Act: 0.0000

Data Index: 667760 (C4 (Web Text))

Max Activating Token Index: 180

Text #1

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8863. Min Act: -0.0000

Data Index: 1552001 (Python Code)

Max Activating Token Index: 410

Text #2

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8286. Min Act: -0.0000

Data Index: 262470 (C4 (Web Text))

Max Activating Token Index: 62

Text #3

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8480. Min Act: -0.0000

Data Index: 1460725 (Python Code)

Max Activating Token Index: 278

Text #4

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8641. Min Act: -0.0000

Data Index: 1042685 (C4 (Web Text))

Max Activating Token Index: 994

Text #5

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7406. Min Act: -0.0000

Data Index: 1516174 (Python Code)

Max Activating Token Index: 310

Text #6

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8626. Min Act: -0.0000

Data Index: 1257909 (C4 (Web Text))

Max Activating Token Index: 970

Text #7

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.8112. Min Act: -0.0000

Data Index: 875564 (C4 (Web Text))

Max Activating Token Index: 903

Text #8

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7529. Min Act: -0.0000

Data Index: 373194 (C4 (Web Text))

Max Activating Token Index: 307

Text #9

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7619. Min Act: -0.0000

Data Index: 1382974 (Python Code)

Max Activating Token Index: 180

Text #10

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7666. Min Act: -0.0000

Data Index: 854327 (C4 (Web Text))

Max Activating Token Index: 838

Text #11

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7767. Min Act: -0.0000

Data Index: 1496386 (Python Code)

Max Activating Token Index: 414

Text #12

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7990. Min Act: 0.0000

Data Index: 1194018 (C4 (Web Text))

Max Activating Token Index: 414

Text #13

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7421. Min Act: -0.0000

Data Index: 1404724 (Python Code)

Max Activating Token Index: 657

Text #14

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7684. Min Act: -0.0000

Data Index: 228856 (C4 (Web Text))

Max Activating Token Index: 997

Text #15

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7792. Min Act: 0.0000

Data Index: 202467 (C4 (Web Text))

Max Activating Token Index: 916

Text #16

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7285. Min Act: -0.0000

Data Index: 518546 (C4 (Web Text))

Max Activating Token Index: 363

Text #17

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7472. Min Act: -0.0000

Data Index: 950437 (C4 (Web Text))

Max Activating Token Index: 170

Text #18

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7469. Min Act: -0.0000

Data Index: 1207433 (C4 (Web Text))

Max Activating Token Index: 668

Text #19

Max Range: 0.9536. Min Range: -0.9536

Max Act: 0.7480. Min Act: -0.0000

Data Index: 320254 (C4 (Web Text))

Max Activating Token Index: 776