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

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

Neuron 894 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.9923. Min Act: -0.1700

Data Index: 1267977 (C4 (Web Text))

Max Activating Token Index: 165

Text #1

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.9298. Min Act: -0.1700

Data Index: 933510 (C4 (Web Text))

Max Activating Token Index: 285

Text #2

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.9611. Min Act: -0.1700

Data Index: 1584118 (Python Code)

Max Activating Token Index: 21

Text #3

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8985. Min Act: -0.1700

Data Index: 1618209 (Python Code)

Max Activating Token Index: 97

Text #4

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8360. Min Act: -0.1700

Data Index: 1465802 (Python Code)

Max Activating Token Index: 269

Text #5

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8360. Min Act: -0.1700

Data Index: 1598933 (Python Code)

Max Activating Token Index: 85

Text #6

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8673. Min Act: -0.1700

Data Index: 301913 (C4 (Web Text))

Max Activating Token Index: 443

Text #7

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8047. Min Act: -0.1700

Data Index: 625621 (C4 (Web Text))

Max Activating Token Index: 16

Text #8

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8360. Min Act: -0.1700

Data Index: 1465214 (Python Code)

Max Activating Token Index: 551

Text #9

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8360. Min Act: -0.1700

Data Index: 1644783 (Python Code)

Max Activating Token Index: 311

Text #10

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7734. Min Act: -0.1700

Data Index: 1022807 (C4 (Web Text))

Max Activating Token Index: 850

Text #11

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7108. Min Act: -0.1700

Data Index: 1527976 (Python Code)

Max Activating Token Index: 987

Text #12

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.8047. Min Act: -0.1700

Data Index: 1403331 (Python Code)

Max Activating Token Index: 150

Text #13

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7421. Min Act: -0.1700

Data Index: 615808 (C4 (Web Text))

Max Activating Token Index: 62

Text #14

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.6795. Min Act: -0.1700

Data Index: 708887 (C4 (Web Text))

Max Activating Token Index: 279

Text #15

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.6482. Min Act: -0.1700

Data Index: 1184958 (C4 (Web Text))

Max Activating Token Index: 636

Text #16

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7734. Min Act: -0.1700

Data Index: 1440702 (Python Code)

Max Activating Token Index: 239

Text #17

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7734. Min Act: -0.1700

Data Index: 1624045 (Python Code)

Max Activating Token Index: 540

Text #18

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7421. Min Act: -0.1700

Data Index: 1474062 (Python Code)

Max Activating Token Index: 71

Text #19

Max Range: 3.9923. Min Range: -3.9923

Max Act: 3.7108. Min Act: -0.1700

Data Index: 1395346 (Python Code)

Max Activating Token Index: 391