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

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

Neuron 1368 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 5.2093. Min Range: -5.2093

Max Act: 5.2093. Min Act: -0.1700

Data Index: 885781 (C4 (Web Text))

Max Activating Token Index: 1007

Text #1

Max Range: 5.2093. Min Range: -5.2093

Max Act: 5.1780. Min Act: -0.1700

Data Index: 1570235 (Python Code)

Max Activating Token Index: 172

Text #2

Max Range: 5.2093. Min Range: -5.2093

Max Act: 5.1780. Min Act: -0.1700

Data Index: 607606 (C4 (Web Text))

Max Activating Token Index: 490

Text #3

Max Range: 5.2093. Min Range: -5.2093

Max Act: 5.0999. Min Act: -0.1700

Data Index: 439727 (C4 (Web Text))

Max Activating Token Index: 715

Text #4

Max Range: 5.2093. Min Range: -5.2093

Max Act: 5.0061. Min Act: -0.1700

Data Index: 605999 (C4 (Web Text))

Max Activating Token Index: 518

Text #5

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9749. Min Act: -0.1700

Data Index: 934786 (C4 (Web Text))

Max Activating Token Index: 420

Text #6

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9280. Min Act: -0.1700

Data Index: 369902 (C4 (Web Text))

Max Activating Token Index: 174

Text #7

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8968. Min Act: -0.1700

Data Index: 693035 (C4 (Web Text))

Max Activating Token Index: 181

Text #8

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9124. Min Act: -0.1700

Data Index: 940422 (C4 (Web Text))

Max Activating Token Index: 1023

Text #9

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9436. Min Act: -0.1700

Data Index: 976685 (C4 (Web Text))

Max Activating Token Index: 884

Text #10

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9436. Min Act: -0.1700

Data Index: 598198 (C4 (Web Text))

Max Activating Token Index: 1023

Text #11

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8811. Min Act: -0.1700

Data Index: 1586980 (Python Code)

Max Activating Token Index: 910

Text #12

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.9124. Min Act: -0.1700

Data Index: 1650994 (Python Code)

Max Activating Token Index: 121

Text #13

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8499. Min Act: -0.1700

Data Index: 1185049 (C4 (Web Text))

Max Activating Token Index: 903

Text #14

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8499. Min Act: -0.1700

Data Index: 506353 (C4 (Web Text))

Max Activating Token Index: 280

Text #15

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8811. Min Act: -0.1700

Data Index: 282636 (C4 (Web Text))

Max Activating Token Index: 975

Text #16

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8811. Min Act: -0.1700

Data Index: 91233 (C4 (Web Text))

Max Activating Token Index: 440

Text #17

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8655. Min Act: -0.1700

Data Index: 842351 (C4 (Web Text))

Max Activating Token Index: 61

Text #18

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8186. Min Act: -0.1700

Data Index: 1125417 (C4 (Web Text))

Max Activating Token Index: 32

Text #19

Max Range: 5.2093. Min Range: -5.2093

Max Act: 4.8343. Min Act: -0.1700

Data Index: 179245 (C4 (Web Text))

Max Activating Token Index: 281