Model: SoLU Model: 8 Layers, 4096 Neurons per Layer

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

Neuron 3361 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-8l')

Text #0

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.7634. Min Act: -0.0001

Data Index: 842521 (C4 (Web Text))

Max Activating Token Index: 562

Text #1

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.8029. Min Act: -0.0001

Data Index: 407181 (C4 (Web Text))

Max Activating Token Index: 50

Text #2

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.8087. Min Act: -0.0001

Data Index: 1563504 (Python Code)

Max Activating Token Index: 1006

Text #3

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.6608. Min Act: -0.0001

Data Index: 264971 (C4 (Web Text))

Max Activating Token Index: 764

Text #4

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.4887. Min Act: -0.0001

Data Index: 1072573 (C4 (Web Text))

Max Activating Token Index: 521

Text #5

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.6230. Min Act: -0.0001

Data Index: 1629100 (Python Code)

Max Activating Token Index: 581

Text #6

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.4379. Min Act: -0.0001

Data Index: 13157 (C4 (Web Text))

Max Activating Token Index: 133

Text #7

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.4838. Min Act: -0.0001

Data Index: 1353718 (C4 (Web Text))

Max Activating Token Index: 278

Text #8

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.3543. Min Act: -0.0001

Data Index: 25291 (C4 (Web Text))

Max Activating Token Index: 120

Text #9

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.4634. Min Act: -0.0001

Data Index: 798434 (C4 (Web Text))

Max Activating Token Index: 600

Text #10

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.3214. Min Act: -0.0001

Data Index: 734661 (C4 (Web Text))

Max Activating Token Index: 965

Text #11

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.3611. Min Act: -0.0001

Data Index: 1614982 (Python Code)

Max Activating Token Index: 455

Text #12

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.3451. Min Act: -0.0001

Data Index: 645960 (C4 (Web Text))

Max Activating Token Index: 461

Text #13

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2984. Min Act: -0.0001

Data Index: 1440831 (Python Code)

Max Activating Token Index: 272

Text #14

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2579. Min Act: -0.0001

Data Index: 1641998 (Python Code)

Max Activating Token Index: 144

Text #15

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2474. Min Act: -0.0001

Data Index: 671967 (C4 (Web Text))

Max Activating Token Index: 764

Text #16

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2382. Min Act: -0.0001

Data Index: 922185 (C4 (Web Text))

Max Activating Token Index: 844

Text #17

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2226. Min Act: -0.0001

Data Index: 1315317 (C4 (Web Text))

Max Activating Token Index: 272

Text #18

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2459. Min Act: -0.0001

Data Index: 475049 (C4 (Web Text))

Max Activating Token Index: 667

Text #19

Max Range: 1.8087. Min Range: -1.8087

Max Act: 1.2644. Min Act: -0.0001

Data Index: 49546 (C4 (Web Text))

Max Activating Token Index: 677