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

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

Neuron 633 in Layer 0

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.3857. Min Act: -0.1700

Data Index: 689267 (C4 (Web Text))

Max Activating Token Index: 298

Text #1

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.2915. Min Act: -0.1700

Data Index: 334673 (C4 (Web Text))

Max Activating Token Index: 582

Text #2

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.1185. Min Act: -0.1700

Data Index: 1458464 (Python Code)

Max Activating Token Index: 825

Text #3

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.1342. Min Act: -0.1700

Data Index: 1422716 (Python Code)

Max Activating Token Index: 462

Text #4

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.1185. Min Act: -0.1700

Data Index: 1321418 (C4 (Web Text))

Max Activating Token Index: 812

Text #5

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0711. Min Act: -0.1700

Data Index: 1522100 (Python Code)

Max Activating Token Index: 25

Text #6

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0554. Min Act: -0.1700

Data Index: 475996 (C4 (Web Text))

Max Activating Token Index: 692

Text #7

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0554. Min Act: -0.1700

Data Index: 100691 (C4 (Web Text))

Max Activating Token Index: 243

Text #8

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0554. Min Act: -0.1700

Data Index: 1045156 (C4 (Web Text))

Max Activating Token Index: 988

Text #9

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0554. Min Act: -0.1700

Data Index: 1615115 (Python Code)

Max Activating Token Index: 766

Text #10

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0396. Min Act: -0.1700

Data Index: 561722 (C4 (Web Text))

Max Activating Token Index: 2

Text #11

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0554. Min Act: -0.1700

Data Index: 467037 (C4 (Web Text))

Max Activating Token Index: 817

Text #12

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0080. Min Act: -0.1700

Data Index: 168655 (C4 (Web Text))

Max Activating Token Index: 735

Text #13

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0080. Min Act: -0.1700

Data Index: 1499831 (Python Code)

Max Activating Token Index: 2

Text #14

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0080. Min Act: -0.1700

Data Index: 441410 (C4 (Web Text))

Max Activating Token Index: 85

Text #15

Max Range: 3.3857. Min Range: -3.3857

Max Act: 2.9922. Min Act: -0.1700

Data Index: 1000767 (C4 (Web Text))

Max Activating Token Index: 654

Text #16

Max Range: 3.3857. Min Range: -3.3857

Max Act: 2.9764. Min Act: -0.1700

Data Index: 587739 (C4 (Web Text))

Max Activating Token Index: 399

Text #17

Max Range: 3.3857. Min Range: -3.3857

Max Act: 2.9922. Min Act: -0.1700

Data Index: 176777 (C4 (Web Text))

Max Activating Token Index: 150

Text #18

Max Range: 3.3857. Min Range: -3.3857

Max Act: 3.0238. Min Act: -0.1700

Data Index: 449306 (C4 (Web Text))

Max Activating Token Index: 512

Text #19

Max Range: 3.3857. Min Range: -3.3857

Max Act: 2.9764. Min Act: -0.1700

Data Index: 318769 (C4 (Web Text))

Max Activating Token Index: 13