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

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

Neuron 1502 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: 3.8078. Min Range: -3.8078

Max Act: 3.8078. Min Act: -0.1700

Data Index: 1016639 (C4 (Web Text))

Max Activating Token Index: 965

Text #1

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.7609. Min Act: -0.1700

Data Index: 283702 (C4 (Web Text))

Max Activating Token Index: 873

Text #2

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.6826. Min Act: -0.1700

Data Index: 403204 (C4 (Web Text))

Max Activating Token Index: 712

Text #3

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.6200. Min Act: -0.1700

Data Index: 1269123 (C4 (Web Text))

Max Activating Token Index: 1013

Text #4

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.5731. Min Act: -0.1700

Data Index: 158467 (C4 (Web Text))

Max Activating Token Index: 448

Text #5

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4947. Min Act: -0.1700

Data Index: 141383 (C4 (Web Text))

Max Activating Token Index: 617

Text #6

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.5417. Min Act: -0.1700

Data Index: 77090 (C4 (Web Text))

Max Activating Token Index: 304

Text #7

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4947. Min Act: -0.1700

Data Index: 140982 (C4 (Web Text))

Max Activating Token Index: 453

Text #8

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4791. Min Act: -0.1700

Data Index: 165658 (C4 (Web Text))

Max Activating Token Index: 18

Text #9

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4477. Min Act: -0.1700

Data Index: 1167334 (C4 (Web Text))

Max Activating Token Index: 901

Text #10

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4634. Min Act: -0.1700

Data Index: 844779 (C4 (Web Text))

Max Activating Token Index: 393

Text #11

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4320. Min Act: -0.1700

Data Index: 262091 (C4 (Web Text))

Max Activating Token Index: 470

Text #12

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4320. Min Act: -0.1700

Data Index: 98338 (C4 (Web Text))

Max Activating Token Index: 570

Text #13

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4477. Min Act: -0.1700

Data Index: 1368020 (Python Code)

Max Activating Token Index: 172

Text #14

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4163. Min Act: -0.1700

Data Index: 333055 (C4 (Web Text))

Max Activating Token Index: 662

Text #15

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4163. Min Act: -0.1700

Data Index: 1368539 (Python Code)

Max Activating Token Index: 912

Text #16

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4163. Min Act: -0.1700

Data Index: 461920 (C4 (Web Text))

Max Activating Token Index: 298

Text #17

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4007. Min Act: -0.1700

Data Index: 888815 (C4 (Web Text))

Max Activating Token Index: 340

Text #18

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4007. Min Act: -0.1700

Data Index: 420413 (C4 (Web Text))

Max Activating Token Index: 140

Text #19

Max Range: 3.8078. Min Range: -3.8078

Max Act: 3.4007. Min Act: -0.1700

Data Index: 1069662 (C4 (Web Text))

Max Activating Token Index: 917