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

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

Neuron 1678 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: 3.2327. Min Range: -3.2327

Max Act: 3.2327. Min Act: -0.1700

Data Index: 1349123 (C4 (Web Text))

Max Activating Token Index: 104

Text #1

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.2170. Min Act: -0.1700

Data Index: 960616 (C4 (Web Text))

Max Activating Token Index: 667

Text #2

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.2170. Min Act: -0.1700

Data Index: 1243126 (C4 (Web Text))

Max Activating Token Index: 446

Text #3

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1855. Min Act: -0.1700

Data Index: 178788 (C4 (Web Text))

Max Activating Token Index: 967

Text #4

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1855. Min Act: -0.1700

Data Index: 60983 (C4 (Web Text))

Max Activating Token Index: 612

Text #5

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 1543611 (Python Code)

Max Activating Token Index: 3

Text #6

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 1198195 (C4 (Web Text))

Max Activating Token Index: 395

Text #7

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 1353285 (C4 (Web Text))

Max Activating Token Index: 606

Text #8

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 874393 (C4 (Web Text))

Max Activating Token Index: 85

Text #9

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 378243 (C4 (Web Text))

Max Activating Token Index: 526

Text #10

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 676414 (C4 (Web Text))

Max Activating Token Index: 688

Text #11

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 49308 (C4 (Web Text))

Max Activating Token Index: 267

Text #12

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1540. Min Act: -0.1700

Data Index: 143708 (C4 (Web Text))

Max Activating Token Index: 817

Text #13

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 623563 (C4 (Web Text))

Max Activating Token Index: 710

Text #14

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 1246827 (C4 (Web Text))

Max Activating Token Index: 109

Text #15

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 249500 (C4 (Web Text))

Max Activating Token Index: 884

Text #16

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 470408 (C4 (Web Text))

Max Activating Token Index: 574

Text #17

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1383. Min Act: -0.1700

Data Index: 1627532 (Python Code)

Max Activating Token Index: 2

Text #18

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1067. Min Act: -0.1700

Data Index: 65024 (C4 (Web Text))

Max Activating Token Index: 222

Text #19

Max Range: 3.2327. Min Range: -3.2327

Max Act: 3.1067. Min Act: -0.1700

Data Index: 1282126 (C4 (Web Text))

Max Activating Token Index: 421