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

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

Neuron 1768 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3741. Min Act: -0.1700

Data Index: 1108299 (C4 (Web Text))

Max Activating Token Index: 440

Text #1

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3898. Min Act: -0.1700

Data Index: 1086890 (C4 (Web Text))

Max Activating Token Index: 343

Text #2

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3584. Min Act: -0.1700

Data Index: 247429 (C4 (Web Text))

Max Activating Token Index: 912

Text #3

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3741. Min Act: -0.1700

Data Index: 1649569 (Python Code)

Max Activating Token Index: 357

Text #4

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3584. Min Act: -0.1700

Data Index: 846609 (C4 (Web Text))

Max Activating Token Index: 438

Text #5

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3270. Min Act: -0.1700

Data Index: 927110 (C4 (Web Text))

Max Activating Token Index: 908

Text #6

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3741. Min Act: -0.1700

Data Index: 696052 (C4 (Web Text))

Max Activating Token Index: 714

Text #7

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3113. Min Act: -0.1700

Data Index: 659116 (C4 (Web Text))

Max Activating Token Index: 889

Text #8

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 985812 (C4 (Web Text))

Max Activating Token Index: 968

Text #9

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3270. Min Act: -0.1700

Data Index: 641923 (C4 (Web Text))

Max Activating Token Index: 1015

Text #10

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 259749 (C4 (Web Text))

Max Activating Token Index: 916

Text #11

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 1396842 (Python Code)

Max Activating Token Index: 607

Text #12

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3113. Min Act: -0.1700

Data Index: 1491671 (Python Code)

Max Activating Token Index: 696

Text #13

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3113. Min Act: -0.1700

Data Index: 274602 (C4 (Web Text))

Max Activating Token Index: 809

Text #14

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 171772 (C4 (Web Text))

Max Activating Token Index: 962

Text #15

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 1169582 (C4 (Web Text))

Max Activating Token Index: 1019

Text #16

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 245449 (C4 (Web Text))

Max Activating Token Index: 397

Text #17

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3270. Min Act: -0.1700

Data Index: 1164962 (C4 (Web Text))

Max Activating Token Index: 614

Text #18

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 1168309 (C4 (Web Text))

Max Activating Token Index: 837

Text #19

Max Range: 3.3898. Min Range: -3.3898

Max Act: 3.3427. Min Act: -0.1700

Data Index: 971314 (C4 (Web Text))

Max Activating Token Index: 958