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

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

Neuron 1200 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: 8.4464. Min Range: -8.4464

Max Act: 8.4464. Min Act: -0.1699

Data Index: 55485 (C4 (Web Text))

Max Activating Token Index: 76

Text #1

Max Range: 8.4464. Min Range: -8.4464

Max Act: 8.0089. Min Act: -0.1699

Data Index: 1115428 (C4 (Web Text))

Max Activating Token Index: 60

Text #2

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9777. Min Act: -0.1698

Data Index: 531162 (C4 (Web Text))

Max Activating Token Index: 73

Text #3

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9464. Min Act: -0.1699

Data Index: 1066449 (C4 (Web Text))

Max Activating Token Index: 614

Text #4

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8839. Min Act: -0.1699

Data Index: 656134 (C4 (Web Text))

Max Activating Token Index: 184

Text #5

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8839. Min Act: -0.1700

Data Index: 346391 (C4 (Web Text))

Max Activating Token Index: 243

Text #6

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9464. Min Act: -0.1700

Data Index: 301863 (C4 (Web Text))

Max Activating Token Index: 268

Text #7

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9152. Min Act: -0.1700

Data Index: 1001707 (C4 (Web Text))

Max Activating Token Index: 244

Text #8

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9152. Min Act: -0.1699

Data Index: 574966 (C4 (Web Text))

Max Activating Token Index: 190

Text #9

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8839. Min Act: -0.1700

Data Index: 320604 (C4 (Web Text))

Max Activating Token Index: 39

Text #10

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.9152. Min Act: -0.1700

Data Index: 866380 (C4 (Web Text))

Max Activating Token Index: 645

Text #11

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8839. Min Act: -0.1700

Data Index: 102985 (C4 (Web Text))

Max Activating Token Index: 126

Text #12

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8839. Min Act: -0.1696

Data Index: 106483 (C4 (Web Text))

Max Activating Token Index: 802

Text #13

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.7902. Min Act: -0.1700

Data Index: 591583 (C4 (Web Text))

Max Activating Token Index: 62

Text #14

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8214. Min Act: -0.1700

Data Index: 1290201 (C4 (Web Text))

Max Activating Token Index: 540

Text #15

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8214. Min Act: -0.1697

Data Index: 725234 (C4 (Web Text))

Max Activating Token Index: 60

Text #16

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.8214. Min Act: -0.1700

Data Index: 162396 (C4 (Web Text))

Max Activating Token Index: 346

Text #17

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.7902. Min Act: -0.1671

Data Index: 1091046 (C4 (Web Text))

Max Activating Token Index: 10

Text #18

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.7902. Min Act: -0.1680

Data Index: 1280468 (C4 (Web Text))

Max Activating Token Index: 594

Text #19

Max Range: 8.4464. Min Range: -8.4464

Max Act: 7.7589. Min Act: -0.1697

Data Index: 1203223 (C4 (Web Text))

Max Activating Token Index: 135