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

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

Neuron 1266 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: 5.1682. Min Range: -5.1682

Max Act: 5.1213. Min Act: -0.1700

Data Index: 356626 (C4 (Web Text))

Max Activating Token Index: 574

Text #1

Max Range: 5.1682. Min Range: -5.1682

Max Act: 5.1682. Min Act: -0.1700

Data Index: 634817 (C4 (Web Text))

Max Activating Token Index: 561

Text #2

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.9338. Min Act: -0.1700

Data Index: 1548422 (Python Code)

Max Activating Token Index: 979

Text #3

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.9025. Min Act: -0.1700

Data Index: 1576252 (Python Code)

Max Activating Token Index: 994

Text #4

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.9650. Min Act: -0.1700

Data Index: 582220 (C4 (Web Text))

Max Activating Token Index: 480

Text #5

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.9650. Min Act: -0.1700

Data Index: 1364169 (Python Code)

Max Activating Token Index: 1018

Text #6

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.8713. Min Act: -0.1700

Data Index: 1638755 (Python Code)

Max Activating Token Index: 1016

Text #7

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7775. Min Act: -0.1700

Data Index: 1425310 (Python Code)

Max Activating Token Index: 376

Text #8

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.9025. Min Act: -0.1700

Data Index: 1290861 (C4 (Web Text))

Max Activating Token Index: 578

Text #9

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.8244. Min Act: -0.1699

Data Index: 1533177 (Python Code)

Max Activating Token Index: 416

Text #10

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.8244. Min Act: -0.1700

Data Index: 701359 (C4 (Web Text))

Max Activating Token Index: 259

Text #11

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7619. Min Act: -0.1700

Data Index: 458113 (C4 (Web Text))

Max Activating Token Index: 677

Text #12

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.6838. Min Act: -0.1700

Data Index: 1394576 (Python Code)

Max Activating Token Index: 323

Text #13

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7307. Min Act: -0.1700

Data Index: 1582036 (Python Code)

Max Activating Token Index: 138

Text #14

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7150. Min Act: -0.1700

Data Index: 1545606 (Python Code)

Max Activating Token Index: 95

Text #15

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7307. Min Act: -0.1700

Data Index: 553839 (C4 (Web Text))

Max Activating Token Index: 255

Text #16

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.7150. Min Act: -0.1700

Data Index: 1578418 (Python Code)

Max Activating Token Index: 400

Text #17

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.6994. Min Act: -0.1700

Data Index: 1511477 (Python Code)

Max Activating Token Index: 403

Text #18

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.6057. Min Act: -0.1700

Data Index: 1370610 (Python Code)

Max Activating Token Index: 579

Text #19

Max Range: 5.1682. Min Range: -5.1682

Max Act: 4.6213. Min Act: -0.1700

Data Index: 1471703 (Python Code)

Max Activating Token Index: 525