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

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

Neuron 559 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: 7.0343. Min Range: -7.0343

Max Act: 7.0343. Min Act: -0.1700

Data Index: 1609283 (Python Code)

Max Activating Token Index: 756

Text #1

Max Range: 7.0343. Min Range: -7.0343

Max Act: 6.1905. Min Act: -0.1700

Data Index: 1480405 (Python Code)

Max Activating Token Index: 654

Text #2

Max Range: 7.0343. Min Range: -7.0343

Max Act: 6.0343. Min Act: -0.1700

Data Index: 1327438 (C4 (Web Text))

Max Activating Token Index: 388

Text #3

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.7843. Min Act: -0.1700

Data Index: 1585059 (Python Code)

Max Activating Token Index: 1016

Text #4

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.8155. Min Act: -0.1700

Data Index: 250692 (C4 (Web Text))

Max Activating Token Index: 337

Text #5

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.8155. Min Act: -0.1700

Data Index: 508455 (C4 (Web Text))

Max Activating Token Index: 778

Text #6

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.6280. Min Act: -0.1700

Data Index: 1541883 (Python Code)

Max Activating Token Index: 356

Text #7

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.5655. Min Act: -0.1700

Data Index: 1045877 (C4 (Web Text))

Max Activating Token Index: 637

Text #8

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.5343. Min Act: -0.1700

Data Index: 103979 (C4 (Web Text))

Max Activating Token Index: 796

Text #9

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.4405. Min Act: -0.1700

Data Index: 1180450 (C4 (Web Text))

Max Activating Token Index: 833

Text #10

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.5343. Min Act: -0.1700

Data Index: 675917 (C4 (Web Text))

Max Activating Token Index: 540

Text #11

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.3780. Min Act: -0.1700

Data Index: 1451090 (Python Code)

Max Activating Token Index: 634

Text #12

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.3780. Min Act: -0.1700

Data Index: 1528806 (Python Code)

Max Activating Token Index: 736

Text #13

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.5030. Min Act: -0.1700

Data Index: 1473960 (Python Code)

Max Activating Token Index: 429

Text #14

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.4093. Min Act: -0.1700

Data Index: 1599990 (Python Code)

Max Activating Token Index: 651

Text #15

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.5030. Min Act: -0.1700

Data Index: 806658 (C4 (Web Text))

Max Activating Token Index: 721

Text #16

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.4093. Min Act: -0.1700

Data Index: 196132 (C4 (Web Text))

Max Activating Token Index: 1003

Text #17

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.4405. Min Act: -0.1700

Data Index: 283103 (C4 (Web Text))

Max Activating Token Index: 799

Text #18

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.4093. Min Act: -0.1700

Data Index: 1459983 (Python Code)

Max Activating Token Index: 898

Text #19

Max Range: 7.0343. Min Range: -7.0343

Max Act: 5.3780. Min Act: -0.1700

Data Index: 1400554 (Python Code)

Max Activating Token Index: 764