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

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

Neuron 435 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: 5.4894. Min Range: -5.4894

Max Act: 5.4894. Min Act: -0.1700

Data Index: 1155791 (C4 (Web Text))

Max Activating Token Index: 312

Text #1

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.3331. Min Act: -0.1700

Data Index: 1059329 (C4 (Web Text))

Max Activating Token Index: 533

Text #2

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.3331. Min Act: -0.1700

Data Index: 1010639 (C4 (Web Text))

Max Activating Token Index: 533

Text #3

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2706. Min Act: -0.1700

Data Index: 610073 (C4 (Web Text))

Max Activating Token Index: 986

Text #4

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.3019. Min Act: -0.1700

Data Index: 165699 (C4 (Web Text))

Max Activating Token Index: 989

Text #5

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2706. Min Act: -0.1700

Data Index: 1329939 (C4 (Web Text))

Max Activating Token Index: 151

Text #6

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2394. Min Act: -0.1700

Data Index: 466553 (C4 (Web Text))

Max Activating Token Index: 539

Text #7

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2706. Min Act: -0.1700

Data Index: 1413276 (Python Code)

Max Activating Token Index: 874

Text #8

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2394. Min Act: -0.1700

Data Index: 20138 (C4 (Web Text))

Max Activating Token Index: 51

Text #9

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2081. Min Act: -0.1700

Data Index: 426900 (C4 (Web Text))

Max Activating Token Index: 57

Text #10

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.2081. Min Act: -0.1700

Data Index: 123483 (C4 (Web Text))

Max Activating Token Index: 557

Text #11

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1769. Min Act: -0.1700

Data Index: 1014175 (C4 (Web Text))

Max Activating Token Index: 92

Text #12

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1456. Min Act: -0.1700

Data Index: 464638 (C4 (Web Text))

Max Activating Token Index: 32

Text #13

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1144. Min Act: -0.1700

Data Index: 228336 (C4 (Web Text))

Max Activating Token Index: 885

Text #14

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1456. Min Act: -0.1700

Data Index: 20500 (C4 (Web Text))

Max Activating Token Index: 923

Text #15

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1769. Min Act: -0.1700

Data Index: 1639914 (Python Code)

Max Activating Token Index: 925

Text #16

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1456. Min Act: -0.1700

Data Index: 1613521 (Python Code)

Max Activating Token Index: 784

Text #17

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1144. Min Act: -0.1700

Data Index: 1412954 (Python Code)

Max Activating Token Index: 897

Text #18

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.0831. Min Act: -0.1700

Data Index: 1400840 (Python Code)

Max Activating Token Index: 774

Text #19

Max Range: 5.4894. Min Range: -5.4894

Max Act: 5.1144. Min Act: -0.1700

Data Index: 1644339 (Python Code)

Max Activating Token Index: 913