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

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

Neuron 1632 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.5161. Min Act: -0.1700

Data Index: 1131585 (C4 (Web Text))

Max Activating Token Index: 580

Text #1

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.4063. Min Act: -0.1700

Data Index: 627436 (C4 (Web Text))

Max Activating Token Index: 109

Text #2

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.3436. Min Act: -0.1700

Data Index: 540157 (C4 (Web Text))

Max Activating Token Index: 247

Text #3

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2965. Min Act: -0.1700

Data Index: 621094 (C4 (Web Text))

Max Activating Token Index: 21

Text #4

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2493. Min Act: -0.1700

Data Index: 121232 (C4 (Web Text))

Max Activating Token Index: 530

Text #5

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2650. Min Act: -0.1700

Data Index: 566252 (C4 (Web Text))

Max Activating Token Index: 268

Text #6

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1864. Min Act: -0.1700

Data Index: 961551 (C4 (Web Text))

Max Activating Token Index: 758

Text #7

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2493. Min Act: -0.1700

Data Index: 1450747 (Python Code)

Max Activating Token Index: 726

Text #8

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2493. Min Act: -0.1700

Data Index: 1321721 (C4 (Web Text))

Max Activating Token Index: 522

Text #9

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2022. Min Act: -0.1700

Data Index: 555978 (C4 (Web Text))

Max Activating Token Index: 419

Text #10

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2179. Min Act: -0.1700

Data Index: 304985 (C4 (Web Text))

Max Activating Token Index: 630

Text #11

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2336. Min Act: -0.1700

Data Index: 183138 (C4 (Web Text))

Max Activating Token Index: 318

Text #12

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1549. Min Act: -0.1700

Data Index: 1308438 (C4 (Web Text))

Max Activating Token Index: 25

Text #13

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1864. Min Act: -0.1700

Data Index: 955365 (C4 (Web Text))

Max Activating Token Index: 629

Text #14

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1707. Min Act: -0.1700

Data Index: 1270043 (C4 (Web Text))

Max Activating Token Index: 660

Text #15

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.2179. Min Act: -0.1700

Data Index: 1617008 (Python Code)

Max Activating Token Index: 614

Text #16

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1549. Min Act: -0.1700

Data Index: 1145360 (C4 (Web Text))

Max Activating Token Index: 559

Text #17

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1707. Min Act: -0.1700

Data Index: 736535 (C4 (Web Text))

Max Activating Token Index: 154

Text #18

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1707. Min Act: -0.1700

Data Index: 41179 (C4 (Web Text))

Max Activating Token Index: 156

Text #19

Max Range: 3.5161. Min Range: -3.5161

Max Act: 3.1549. Min Act: -0.1700

Data Index: 1421099 (Python Code)

Max Activating Token Index: 678