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

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

Neuron 1535 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.6471. Min Act: -0.1700

Data Index: 588299 (C4 (Web Text))

Max Activating Token Index: 958

Text #1

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.5219. Min Act: -0.1700

Data Index: 452986 (C4 (Web Text))

Max Activating Token Index: 863

Text #2

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4749. Min Act: -0.1700

Data Index: 1095661 (C4 (Web Text))

Max Activating Token Index: 288

Text #3

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4905. Min Act: -0.1700

Data Index: 386235 (C4 (Web Text))

Max Activating Token Index: 962

Text #4

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4749. Min Act: -0.1700

Data Index: 1444040 (Python Code)

Max Activating Token Index: 524

Text #5

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4749. Min Act: -0.1700

Data Index: 840341 (C4 (Web Text))

Max Activating Token Index: 946

Text #6

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.5219. Min Act: -0.1700

Data Index: 781549 (C4 (Web Text))

Max Activating Token Index: 751

Text #7

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4435. Min Act: -0.1700

Data Index: 113470 (C4 (Web Text))

Max Activating Token Index: 873

Text #8

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4592. Min Act: -0.1700

Data Index: 1041242 (C4 (Web Text))

Max Activating Token Index: 624

Text #9

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4278. Min Act: -0.1700

Data Index: 1312395 (C4 (Web Text))

Max Activating Token Index: 286

Text #10

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4278. Min Act: -0.1700

Data Index: 969165 (C4 (Web Text))

Max Activating Token Index: 329

Text #11

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4435. Min Act: -0.1700

Data Index: 1278123 (C4 (Web Text))

Max Activating Token Index: 889

Text #12

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4278. Min Act: -0.1700

Data Index: 894383 (C4 (Web Text))

Max Activating Token Index: 797

Text #13

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4278. Min Act: -0.1700

Data Index: 944403 (C4 (Web Text))

Max Activating Token Index: 185

Text #14

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.3965. Min Act: -0.1700

Data Index: 85712 (C4 (Web Text))

Max Activating Token Index: 968

Text #15

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.3808. Min Act: -0.1700

Data Index: 1307438 (C4 (Web Text))

Max Activating Token Index: 297

Text #16

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.4122. Min Act: -0.1700

Data Index: 248065 (C4 (Web Text))

Max Activating Token Index: 596

Text #17

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.3965. Min Act: -0.1700

Data Index: 1418365 (Python Code)

Max Activating Token Index: 228

Text #18

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.3965. Min Act: -0.1700

Data Index: 831060 (C4 (Web Text))

Max Activating Token Index: 721

Text #19

Max Range: 3.6471. Min Range: -3.6471

Max Act: 3.3651. Min Act: -0.1700

Data Index: 531384 (C4 (Web Text))

Max Activating Token Index: 885