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

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

Neuron 1560 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: 3.4514. Min Range: -3.4514

Max Act: 3.4514. Min Act: -0.1700

Data Index: 231377 (C4 (Web Text))

Max Activating Token Index: 116

Text #1

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3730. Min Act: -0.1700

Data Index: 1084370 (C4 (Web Text))

Max Activating Token Index: 647

Text #2

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3259. Min Act: -0.1700

Data Index: 661100 (C4 (Web Text))

Max Activating Token Index: 769

Text #3

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2473. Min Act: -0.1700

Data Index: 1442479 (Python Code)

Max Activating Token Index: 537

Text #4

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3416. Min Act: -0.1700

Data Index: 1146341 (C4 (Web Text))

Max Activating Token Index: 695

Text #5

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3573. Min Act: -0.1700

Data Index: 224937 (C4 (Web Text))

Max Activating Token Index: 668

Text #6

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3102. Min Act: -0.1700

Data Index: 766732 (C4 (Web Text))

Max Activating Token Index: 1002

Text #7

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3259. Min Act: -0.1700

Data Index: 450874 (C4 (Web Text))

Max Activating Token Index: 851

Text #8

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2316. Min Act: -0.1700

Data Index: 349474 (C4 (Web Text))

Max Activating Token Index: 530

Text #9

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3102. Min Act: -0.1700

Data Index: 635862 (C4 (Web Text))

Max Activating Token Index: 546

Text #10

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.3102. Min Act: -0.1700

Data Index: 208934 (C4 (Web Text))

Max Activating Token Index: 645

Text #11

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2473. Min Act: -0.1700

Data Index: 1262403 (C4 (Web Text))

Max Activating Token Index: 1001

Text #12

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2945. Min Act: -0.1700

Data Index: 248865 (C4 (Web Text))

Max Activating Token Index: 764

Text #13

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2473. Min Act: -0.1700

Data Index: 1320544 (C4 (Web Text))

Max Activating Token Index: 152

Text #14

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2002. Min Act: -0.1700

Data Index: 1614380 (Python Code)

Max Activating Token Index: 647

Text #15

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2159. Min Act: -0.1700

Data Index: 733852 (C4 (Web Text))

Max Activating Token Index: 161

Text #16

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2002. Min Act: -0.1700

Data Index: 1227200 (C4 (Web Text))

Max Activating Token Index: 379

Text #17

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.1844. Min Act: -0.1700

Data Index: 987699 (C4 (Web Text))

Max Activating Token Index: 796

Text #18

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.2002. Min Act: -0.1700

Data Index: 1101321 (C4 (Web Text))

Max Activating Token Index: 398

Text #19

Max Range: 3.4514. Min Range: -3.4514

Max Act: 3.1687. Min Act: -0.1700

Data Index: 648837 (C4 (Web Text))

Max Activating Token Index: 335