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

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

Neuron 1026 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: 5.4915. Min Range: -5.4915

Max Act: 5.4915. Min Act: -0.1700

Data Index: 1444270 (Python Code)

Max Activating Token Index: 737

Text #1

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.4915. Min Act: -0.1700

Data Index: 1334434 (C4 (Web Text))

Max Activating Token Index: 259

Text #2

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.3978. Min Act: -0.1700

Data Index: 1165865 (C4 (Web Text))

Max Activating Token Index: 928

Text #3

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.3978. Min Act: -0.1700

Data Index: 1476368 (Python Code)

Max Activating Token Index: 621

Text #4

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.3665. Min Act: -0.1700

Data Index: 453079 (C4 (Web Text))

Max Activating Token Index: 699

Text #5

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2415. Min Act: -0.1700

Data Index: 1632814 (Python Code)

Max Activating Token Index: 998

Text #6

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.3353. Min Act: -0.1700

Data Index: 469895 (C4 (Web Text))

Max Activating Token Index: 461

Text #7

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.3353. Min Act: -0.1700

Data Index: 379962 (C4 (Web Text))

Max Activating Token Index: 863

Text #8

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2415. Min Act: -0.1700

Data Index: 24558 (C4 (Web Text))

Max Activating Token Index: 451

Text #9

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2103. Min Act: -0.1700

Data Index: 1512770 (Python Code)

Max Activating Token Index: 982

Text #10

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1478. Min Act: -0.1700

Data Index: 1486638 (Python Code)

Max Activating Token Index: 346

Text #11

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2415. Min Act: -0.1700

Data Index: 103800 (C4 (Web Text))

Max Activating Token Index: 63

Text #12

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2415. Min Act: -0.1700

Data Index: 354309 (C4 (Web Text))

Max Activating Token Index: 79

Text #13

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.2103. Min Act: -0.1700

Data Index: 387103 (C4 (Web Text))

Max Activating Token Index: 676

Text #14

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1790. Min Act: -0.1700

Data Index: 506135 (C4 (Web Text))

Max Activating Token Index: 909

Text #15

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1790. Min Act: -0.1700

Data Index: 179829 (C4 (Web Text))

Max Activating Token Index: 217

Text #16

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1165. Min Act: -0.1700

Data Index: 1644767 (Python Code)

Max Activating Token Index: 392

Text #17

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1790. Min Act: -0.1700

Data Index: 405352 (C4 (Web Text))

Max Activating Token Index: 431

Text #18

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.1165. Min Act: -0.1700

Data Index: 1385425 (Python Code)

Max Activating Token Index: 410

Text #19

Max Range: 5.4915. Min Range: -5.4915

Max Act: 5.0853. Min Act: -0.1700

Data Index: 1393449 (Python Code)

Max Activating Token Index: 898