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

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

Neuron 1030 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: 3.3068. Min Range: -3.3068

Max Act: 3.3068. Min Act: -0.1700

Data Index: 489188 (C4 (Web Text))

Max Activating Token Index: 964

Text #1

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.2125. Min Act: -0.1700

Data Index: 157170 (C4 (Web Text))

Max Activating Token Index: 972

Text #2

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.1810. Min Act: -0.1700

Data Index: 369566 (C4 (Web Text))

Max Activating Token Index: 710

Text #3

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.1338. Min Act: -0.1700

Data Index: 808775 (C4 (Web Text))

Max Activating Token Index: 591

Text #4

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.1023. Min Act: -0.1700

Data Index: 659795 (C4 (Web Text))

Max Activating Token Index: 14

Text #5

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0865. Min Act: -0.1700

Data Index: 1159443 (C4 (Web Text))

Max Activating Token Index: 941

Text #6

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.1023. Min Act: -0.1700

Data Index: 1164133 (C4 (Web Text))

Max Activating Token Index: 864

Text #7

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0707. Min Act: -0.1700

Data Index: 736200 (C4 (Web Text))

Max Activating Token Index: 34

Text #8

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0707. Min Act: -0.1700

Data Index: 878325 (C4 (Web Text))

Max Activating Token Index: 577

Text #9

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0707. Min Act: -0.1700

Data Index: 1210734 (C4 (Web Text))

Max Activating Token Index: 296

Text #10

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0392. Min Act: -0.1700

Data Index: 1380962 (Python Code)

Max Activating Token Index: 250

Text #11

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0550. Min Act: -0.1700

Data Index: 1297778 (C4 (Web Text))

Max Activating Token Index: 768

Text #12

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0392. Min Act: -0.1700

Data Index: 1084164 (C4 (Web Text))

Max Activating Token Index: 687

Text #13

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0392. Min Act: -0.1700

Data Index: 582732 (C4 (Web Text))

Max Activating Token Index: 988

Text #14

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0076. Min Act: -0.1700

Data Index: 1280091 (C4 (Web Text))

Max Activating Token Index: 96

Text #15

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0234. Min Act: -0.1700

Data Index: 1338753 (C4 (Web Text))

Max Activating Token Index: 700

Text #16

Max Range: 3.3068. Min Range: -3.3068

Max Act: 3.0234. Min Act: -0.1700

Data Index: 1622563 (Python Code)

Max Activating Token Index: 338

Text #17

Max Range: 3.3068. Min Range: -3.3068

Max Act: 2.9759. Min Act: -0.1700

Data Index: 1045389 (C4 (Web Text))

Max Activating Token Index: 425

Text #18

Max Range: 3.3068. Min Range: -3.3068

Max Act: 2.9759. Min Act: -0.1700

Data Index: 43018 (C4 (Web Text))

Max Activating Token Index: 64

Text #19

Max Range: 3.3068. Min Range: -3.3068

Max Act: 2.9443. Min Act: -0.1700

Data Index: 142266 (C4 (Web Text))

Max Activating Token Index: 782