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

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

Neuron 783 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: 4.3620. Min Range: -4.3620

Max Act: 4.3620. Min Act: -0.1700

Data Index: 472661 (C4 (Web Text))

Max Activating Token Index: 455

Text #1

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2917. Min Act: -0.1700

Data Index: 931582 (C4 (Web Text))

Max Activating Token Index: 877

Text #2

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.3542. Min Act: -0.1700

Data Index: 1424164 (Python Code)

Max Activating Token Index: 1010

Text #3

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2761. Min Act: -0.1700

Data Index: 568433 (C4 (Web Text))

Max Activating Token Index: 836

Text #4

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2604. Min Act: -0.1700

Data Index: 415772 (C4 (Web Text))

Max Activating Token Index: 690

Text #5

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2526. Min Act: -0.1700

Data Index: 981882 (C4 (Web Text))

Max Activating Token Index: 630

Text #6

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2839. Min Act: -0.1700

Data Index: 367728 (C4 (Web Text))

Max Activating Token Index: 811

Text #7

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2448. Min Act: -0.1700

Data Index: 1369324 (Python Code)

Max Activating Token Index: 315

Text #8

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2136. Min Act: -0.1700

Data Index: 1566691 (Python Code)

Max Activating Token Index: 439

Text #9

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2292. Min Act: -0.1700

Data Index: 1234204 (C4 (Web Text))

Max Activating Token Index: 540

Text #10

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1979. Min Act: -0.1700

Data Index: 952738 (C4 (Web Text))

Max Activating Token Index: 980

Text #11

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.2057. Min Act: -0.1700

Data Index: 857214 (C4 (Web Text))

Max Activating Token Index: 494

Text #12

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1901. Min Act: -0.1700

Data Index: 259409 (C4 (Web Text))

Max Activating Token Index: 690

Text #13

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1745. Min Act: -0.1700

Data Index: 333380 (C4 (Web Text))

Max Activating Token Index: 993

Text #14

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1589. Min Act: -0.1700

Data Index: 733120 (C4 (Web Text))

Max Activating Token Index: 933

Text #15

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1354. Min Act: -0.1700

Data Index: 322829 (C4 (Web Text))

Max Activating Token Index: 344

Text #16

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1198. Min Act: -0.1700

Data Index: 1384718 (Python Code)

Max Activating Token Index: 524

Text #17

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1041. Min Act: -0.1700

Data Index: 422850 (C4 (Web Text))

Max Activating Token Index: 179

Text #18

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1041. Min Act: -0.1700

Data Index: 1345773 (C4 (Web Text))

Max Activating Token Index: 1011

Text #19

Max Range: 4.3620. Min Range: -4.3620

Max Act: 4.1510. Min Act: -0.1700

Data Index: 521207 (C4 (Web Text))

Max Activating Token Index: 311