Model: GELU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 1833 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-3l')

Text #0

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.7180. Min Act: -0.1700

Data Index: 9413 (C4 (Web Text))

Max Activating Token Index: 8

Text #1

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.6241. Min Act: -0.1700

Data Index: 184197 (C4 (Web Text))

Max Activating Token Index: 135

Text #2

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.5459. Min Act: -0.1700

Data Index: 12856 (C4 (Web Text))

Max Activating Token Index: 173

Text #3

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.4989. Min Act: -0.1700

Data Index: 73384 (C4 (Web Text))

Max Activating Token Index: 344

Text #4

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.4362. Min Act: -0.1700

Data Index: 1056529 (C4 (Web Text))

Max Activating Token Index: 280

Text #5

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.3420. Min Act: -0.1700

Data Index: 422965 (C4 (Web Text))

Max Activating Token Index: 758

Text #6

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.2478. Min Act: -0.1700

Data Index: 1287831 (C4 (Web Text))

Max Activating Token Index: 87

Text #7

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.1849. Min Act: -0.1700

Data Index: 1302500 (C4 (Web Text))

Max Activating Token Index: 306

Text #8

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.1219. Min Act: -0.1700

Data Index: 359457 (C4 (Web Text))

Max Activating Token Index: 189

Text #9

Max Range: 3.7180. Min Range: -3.7180

Max Act: 3.0588. Min Act: -0.1700

Data Index: 1217898 (C4 (Web Text))

Max Activating Token Index: 682

Text #10

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9956. Min Act: -0.1700

Data Index: 588602 (C4 (Web Text))

Max Activating Token Index: 123

Text #11

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9956. Min Act: -0.1700

Data Index: 583449 (C4 (Web Text))

Max Activating Token Index: 985

Text #12

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9798. Min Act: -0.1700

Data Index: 1170714 (C4 (Web Text))

Max Activating Token Index: 69

Text #13

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9481. Min Act: -0.1700

Data Index: 1628025 (Python Code)

Max Activating Token Index: 285

Text #14

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9323. Min Act: -0.1700

Data Index: 1148745 (C4 (Web Text))

Max Activating Token Index: 662

Text #15

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9006. Min Act: -0.1700

Data Index: 469825 (C4 (Web Text))

Max Activating Token Index: 887

Text #16

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9165. Min Act: -0.1700

Data Index: 915992 (C4 (Web Text))

Max Activating Token Index: 20

Text #17

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.9006. Min Act: -0.1700

Data Index: 829437 (C4 (Web Text))

Max Activating Token Index: 700

Text #18

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.8530. Min Act: -0.1700

Data Index: 789804 (C4 (Web Text))

Max Activating Token Index: 604

Text #19

Max Range: 3.7180. Min Range: -3.7180

Max Act: 2.8847. Min Act: -0.1700

Data Index: 914984 (C4 (Web Text))

Max Activating Token Index: 758