Model: GELU Model: 4 Layers, 2048 Neurons per Layer

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

Neuron 1193 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('gelu-4l')

Text #0

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.8823. Min Act: -0.1700

Data Index: 1253814 (C4 (Web Text))

Max Activating Token Index: 129

Text #1

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.8198. Min Act: -0.1700

Data Index: 1088644 (C4 (Web Text))

Max Activating Token Index: 946

Text #2

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.4135. Min Act: -0.1700

Data Index: 1471582 (Python Code)

Max Activating Token Index: 879

Text #3

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.3823. Min Act: -0.1700

Data Index: 141003 (C4 (Web Text))

Max Activating Token Index: 568

Text #4

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.2573. Min Act: -0.1700

Data Index: 1256625 (C4 (Web Text))

Max Activating Token Index: 933

Text #5

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.1010. Min Act: -0.1700

Data Index: 641787 (C4 (Web Text))

Max Activating Token Index: 369

Text #6

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.1635. Min Act: -0.1700

Data Index: 1449488 (Python Code)

Max Activating Token Index: 897

Text #7

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.1635. Min Act: -0.1700

Data Index: 314715 (C4 (Web Text))

Max Activating Token Index: 414

Text #8

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.1635. Min Act: -0.1700

Data Index: 726170 (C4 (Web Text))

Max Activating Token Index: 635

Text #9

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.9760. Min Act: -0.1700

Data Index: 207838 (C4 (Web Text))

Max Activating Token Index: 887

Text #10

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.0385. Min Act: -0.1700

Data Index: 863316 (C4 (Web Text))

Max Activating Token Index: 371

Text #11

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.0073. Min Act: -0.1700

Data Index: 750793 (C4 (Web Text))

Max Activating Token Index: 370

Text #12

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.0073. Min Act: -0.1700

Data Index: 602483 (C4 (Web Text))

Max Activating Token Index: 712

Text #13

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.0385. Min Act: -0.1700

Data Index: 48987 (C4 (Web Text))

Max Activating Token Index: 728

Text #14

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.9760. Min Act: -0.1700

Data Index: 853984 (C4 (Web Text))

Max Activating Token Index: 747

Text #15

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.9135. Min Act: -0.1700

Data Index: 1410205 (Python Code)

Max Activating Token Index: 258

Text #16

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.8823. Min Act: -0.1700

Data Index: 1556734 (Python Code)

Max Activating Token Index: 536

Text #17

Max Range: 7.8823. Min Range: -7.8823

Max Act: 7.0073. Min Act: -0.1700

Data Index: 728137 (C4 (Web Text))

Max Activating Token Index: 363

Text #18

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.7260. Min Act: -0.1700

Data Index: 90459 (C4 (Web Text))

Max Activating Token Index: 942

Text #19

Max Range: 7.8823. Min Range: -7.8823

Max Act: 6.8510. Min Act: -0.1700

Data Index: 95068 (C4 (Web Text))

Max Activating Token Index: 924