Model: GELU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 409 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 5.2243. Min Range: -5.2243

Max Act: 5.1774. Min Act: -0.1700

Data Index: 569486 (C4 (Web Text))

Max Activating Token Index: 272

Text #1

Max Range: 5.2243. Min Range: -5.2243

Max Act: 5.2243. Min Act: -0.1700

Data Index: 1190058 (C4 (Web Text))

Max Activating Token Index: 825

Text #2

Max Range: 5.2243. Min Range: -5.2243

Max Act: 5.0211. Min Act: -0.1700

Data Index: 220640 (C4 (Web Text))

Max Activating Token Index: 129

Text #3

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.9899. Min Act: -0.1700

Data Index: 1636550 (Python Code)

Max Activating Token Index: 942

Text #4

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.9586. Min Act: -0.1700

Data Index: 1594548 (Python Code)

Max Activating Token Index: 1010

Text #5

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.9274. Min Act: -0.1700

Data Index: 1160276 (C4 (Web Text))

Max Activating Token Index: 281

Text #6

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8805. Min Act: -0.1696

Data Index: 1576445 (Python Code)

Max Activating Token Index: 964

Text #7

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.9118. Min Act: -0.1700

Data Index: 1485420 (Python Code)

Max Activating Token Index: 926

Text #8

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8493. Min Act: -0.1700

Data Index: 1468730 (Python Code)

Max Activating Token Index: 996

Text #9

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8571. Min Act: -0.1700

Data Index: 1420548 (Python Code)

Max Activating Token Index: 556

Text #10

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8258. Min Act: -0.1700

Data Index: 1409984 (Python Code)

Max Activating Token Index: 914

Text #11

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8024. Min Act: -0.1700

Data Index: 1431305 (Python Code)

Max Activating Token Index: 737

Text #12

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7946. Min Act: -0.1689

Data Index: 1485658 (Python Code)

Max Activating Token Index: 956

Text #13

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.8258. Min Act: -0.1700

Data Index: 676611 (C4 (Web Text))

Max Activating Token Index: 834

Text #14

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7946. Min Act: -0.1700

Data Index: 1032616 (C4 (Web Text))

Max Activating Token Index: 612

Text #15

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7946. Min Act: -0.1696

Data Index: 1471297 (Python Code)

Max Activating Token Index: 989

Text #16

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7477. Min Act: -0.1700

Data Index: 349349 (C4 (Web Text))

Max Activating Token Index: 988

Text #17

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7868. Min Act: -0.1693

Data Index: 1422645 (Python Code)

Max Activating Token Index: 946

Text #18

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7946. Min Act: -0.1684

Data Index: 1443929 (Python Code)

Max Activating Token Index: 885

Text #19

Max Range: 5.2243. Min Range: -5.2243

Max Act: 4.7555. Min Act: -0.1684

Data Index: 1478596 (Python Code)

Max Activating Token Index: 817