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

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

Neuron 836 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.4167. Min Act: -0.1700

Data Index: 1104378 (C4 (Web Text))

Max Activating Token Index: 204

Text #1

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.1667. Min Act: -0.1700

Data Index: 465596 (C4 (Web Text))

Max Activating Token Index: 501

Text #2

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.1355. Min Act: -0.1700

Data Index: 596403 (C4 (Web Text))

Max Activating Token Index: 900

Text #3

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.0730. Min Act: -0.1700

Data Index: 1212022 (C4 (Web Text))

Max Activating Token Index: 306

Text #4

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.0417. Min Act: -0.1700

Data Index: 1354891 (C4 (Web Text))

Max Activating Token Index: 422

Text #5

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.1042. Min Act: -0.1700

Data Index: 909041 (C4 (Web Text))

Max Activating Token Index: 711

Text #6

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9480. Min Act: -0.1700

Data Index: 128863 (C4 (Web Text))

Max Activating Token Index: 230

Text #7

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9480. Min Act: -0.1700

Data Index: 1220350 (C4 (Web Text))

Max Activating Token Index: 243

Text #8

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.0417. Min Act: -0.1700

Data Index: 288917 (C4 (Web Text))

Max Activating Token Index: 456

Text #9

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.8855. Min Act: -0.1700

Data Index: 1653032 (Python Code)

Max Activating Token Index: 260

Text #10

Max Range: 7.4167. Min Range: -7.4167

Max Act: 7.0417. Min Act: -0.1700

Data Index: 1104161 (C4 (Web Text))

Max Activating Token Index: 654

Text #11

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9480. Min Act: -0.1700

Data Index: 1063920 (C4 (Web Text))

Max Activating Token Index: 453

Text #12

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9167. Min Act: -0.1700

Data Index: 316849 (C4 (Web Text))

Max Activating Token Index: 547

Text #13

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9167. Min Act: -0.1700

Data Index: 647495 (C4 (Web Text))

Max Activating Token Index: 102

Text #14

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.8855. Min Act: -0.1700

Data Index: 322350 (C4 (Web Text))

Max Activating Token Index: 155

Text #15

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9480. Min Act: -0.1700

Data Index: 357848 (C4 (Web Text))

Max Activating Token Index: 383

Text #16

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9167. Min Act: -0.1700

Data Index: 163927 (C4 (Web Text))

Max Activating Token Index: 465

Text #17

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9167. Min Act: -0.1700

Data Index: 1481762 (Python Code)

Max Activating Token Index: 929

Text #18

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.9167. Min Act: -0.1700

Data Index: 1557619 (Python Code)

Max Activating Token Index: 867

Text #19

Max Range: 7.4167. Min Range: -7.4167

Max Act: 6.8855. Min Act: -0.1700

Data Index: 762972 (C4 (Web Text))

Max Activating Token Index: 679