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

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

Neuron 384 in Layer 1

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.5866. Min Act: -0.1700

Data Index: 1452109 (Python Code)

Max Activating Token Index: 659

Text #1

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.5866. Min Act: -0.1700

Data Index: 19514 (C4 (Web Text))

Max Activating Token Index: 510

Text #2

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.5239. Min Act: -0.1700

Data Index: 1652500 (Python Code)

Max Activating Token Index: 62

Text #3

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4769. Min Act: -0.1700

Data Index: 518463 (C4 (Web Text))

Max Activating Token Index: 955

Text #4

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4769. Min Act: -0.1700

Data Index: 127673 (C4 (Web Text))

Max Activating Token Index: 665

Text #5

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4456. Min Act: -0.1700

Data Index: 458989 (C4 (Web Text))

Max Activating Token Index: 835

Text #6

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4299. Min Act: -0.1700

Data Index: 1228528 (C4 (Web Text))

Max Activating Token Index: 791

Text #7

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4142. Min Act: -0.1700

Data Index: 1402979 (Python Code)

Max Activating Token Index: 246

Text #8

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3985. Min Act: -0.1700

Data Index: 1300624 (C4 (Web Text))

Max Activating Token Index: 264

Text #9

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4613. Min Act: -0.1700

Data Index: 1653669 (Python Code)

Max Activating Token Index: 807

Text #10

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4299. Min Act: -0.1700

Data Index: 143909 (C4 (Web Text))

Max Activating Token Index: 65

Text #11

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3672. Min Act: -0.1700

Data Index: 664484 (C4 (Web Text))

Max Activating Token Index: 876

Text #12

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3515. Min Act: -0.1700

Data Index: 1349147 (C4 (Web Text))

Max Activating Token Index: 609

Text #13

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3672. Min Act: -0.1700

Data Index: 534408 (C4 (Web Text))

Max Activating Token Index: 706

Text #14

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3828. Min Act: -0.1700

Data Index: 78021 (C4 (Web Text))

Max Activating Token Index: 912

Text #15

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3515. Min Act: -0.1700

Data Index: 269138 (C4 (Web Text))

Max Activating Token Index: 680

Text #16

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.4142. Min Act: -0.1700

Data Index: 1050275 (C4 (Web Text))

Max Activating Token Index: 63

Text #17

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3515. Min Act: -0.1700

Data Index: 396929 (C4 (Web Text))

Max Activating Token Index: 356

Text #18

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3672. Min Act: -0.1700

Data Index: 1493717 (Python Code)

Max Activating Token Index: 810

Text #19

Max Range: 3.5866. Min Range: -3.5866

Max Act: 3.3985. Min Act: -0.1700

Data Index: 103873 (C4 (Web Text))

Max Activating Token Index: 743