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

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

Neuron 2463 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2432. Min Act: -0.0001

Data Index: 623866 (C4 (Web Text))

Max Activating Token Index: 627

Text #1

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2498. Min Act: -0.0001

Data Index: 50669 (C4 (Web Text))

Max Activating Token Index: 897

Text #2

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1890. Min Act: -0.0001

Data Index: 192063 (C4 (Web Text))

Max Activating Token Index: 303

Text #3

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2259. Min Act: -0.0001

Data Index: 565760 (C4 (Web Text))

Max Activating Token Index: 638

Text #4

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1958. Min Act: -0.0001

Data Index: 771255 (C4 (Web Text))

Max Activating Token Index: 541

Text #5

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2038. Min Act: -0.0001

Data Index: 1645918 (Python Code)

Max Activating Token Index: 854

Text #6

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2001. Min Act: -0.0001

Data Index: 260948 (C4 (Web Text))

Max Activating Token Index: 283

Text #7

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1936. Min Act: -0.0001

Data Index: 127747 (C4 (Web Text))

Max Activating Token Index: 47

Text #8

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.2064. Min Act: -0.0001

Data Index: 1334904 (C4 (Web Text))

Max Activating Token Index: 357

Text #9

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1968. Min Act: -0.0001

Data Index: 1366917 (Python Code)

Max Activating Token Index: 935

Text #10

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1877. Min Act: -0.0001

Data Index: 1470702 (Python Code)

Max Activating Token Index: 595

Text #11

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1903. Min Act: -0.0001

Data Index: 1401141 (Python Code)

Max Activating Token Index: 569

Text #12

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1786. Min Act: -0.0001

Data Index: 1205925 (C4 (Web Text))

Max Activating Token Index: 393

Text #13

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1861. Min Act: -0.0001

Data Index: 1568521 (Python Code)

Max Activating Token Index: 413

Text #14

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1882. Min Act: -0.0001

Data Index: 1572997 (Python Code)

Max Activating Token Index: 328

Text #15

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1911. Min Act: -0.0001

Data Index: 1001930 (C4 (Web Text))

Max Activating Token Index: 725

Text #16

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1823. Min Act: -0.0001

Data Index: 1525254 (Python Code)

Max Activating Token Index: 144

Text #17

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1799. Min Act: -0.0001

Data Index: 1368339 (Python Code)

Max Activating Token Index: 119

Text #18

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1745. Min Act: -0.0001

Data Index: 1589435 (Python Code)

Max Activating Token Index: 781

Text #19

Max Range: 0.2498. Min Range: -0.2498

Max Act: 0.1645. Min Act: -0.0001

Data Index: 156752 (C4 (Web Text))

Max Activating Token Index: 995