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

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

Neuron 2678 in Layer 7

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1882. Min Act: -0.0000

Data Index: 729314 (C4 (Web Text))

Max Activating Token Index: 793

Text #1

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1723. Min Act: -0.0000

Data Index: 603427 (C4 (Web Text))

Max Activating Token Index: 741

Text #2

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1625. Min Act: 0.0000

Data Index: 577137 (C4 (Web Text))

Max Activating Token Index: 417

Text #3

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1619. Min Act: -0.0001

Data Index: 744767 (C4 (Web Text))

Max Activating Token Index: 361

Text #4

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1581. Min Act: -0.0000

Data Index: 316927 (C4 (Web Text))

Max Activating Token Index: 978

Text #5

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1566. Min Act: -0.0000

Data Index: 80636 (C4 (Web Text))

Max Activating Token Index: 320

Text #6

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1597. Min Act: -0.0001

Data Index: 404964 (C4 (Web Text))

Max Activating Token Index: 719

Text #7

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1537. Min Act: -0.0001

Data Index: 1498044 (Python Code)

Max Activating Token Index: 69

Text #8

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1597. Min Act: -0.0000

Data Index: 729701 (C4 (Web Text))

Max Activating Token Index: 880

Text #9

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1491. Min Act: -0.0000

Data Index: 176750 (C4 (Web Text))

Max Activating Token Index: 110

Text #10

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1433. Min Act: -0.0001

Data Index: 390896 (C4 (Web Text))

Max Activating Token Index: 84

Text #11

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1484. Min Act: -0.0000

Data Index: 472718 (C4 (Web Text))

Max Activating Token Index: 65

Text #12

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1398. Min Act: -0.0001

Data Index: 994182 (C4 (Web Text))

Max Activating Token Index: 89

Text #13

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1386. Min Act: -0.0000

Data Index: 2026 (C4 (Web Text))

Max Activating Token Index: 876

Text #14

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1487. Min Act: -0.0000

Data Index: 1580621 (Python Code)

Max Activating Token Index: 56

Text #15

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1386. Min Act: -0.0000

Data Index: 265579 (C4 (Web Text))

Max Activating Token Index: 740

Text #16

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1442. Min Act: -0.0001

Data Index: 115097 (C4 (Web Text))

Max Activating Token Index: 863

Text #17

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1428. Min Act: -0.0000

Data Index: 974300 (C4 (Web Text))

Max Activating Token Index: 979

Text #18

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1355. Min Act: -0.0000

Data Index: 1030399 (C4 (Web Text))

Max Activating Token Index: 719

Text #19

Max Range: 0.1882. Min Range: -0.1882

Max Act: 0.1449. Min Act: -0.0000

Data Index: 308218 (C4 (Web Text))

Max Activating Token Index: 1015