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

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

Neuron 1232 in Layer 6

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0404. Min Range: -1.0404

Max Act: 1.0404. Min Act: -0.0001

Data Index: 925880 (C4 (Web Text))

Max Activating Token Index: 644

Text #1

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.9268. Min Act: -0.0001

Data Index: 523079 (C4 (Web Text))

Max Activating Token Index: 814

Text #2

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8399. Min Act: -0.0001

Data Index: 304586 (C4 (Web Text))

Max Activating Token Index: 695

Text #3

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8952. Min Act: -0.0000

Data Index: 605256 (C4 (Web Text))

Max Activating Token Index: 633

Text #4

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8608. Min Act: -0.0001

Data Index: 245214 (C4 (Web Text))

Max Activating Token Index: 872

Text #5

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8543. Min Act: -0.0001

Data Index: 1427697 (Python Code)

Max Activating Token Index: 86

Text #6

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8832. Min Act: -0.0001

Data Index: 914393 (C4 (Web Text))

Max Activating Token Index: 955

Text #7

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8114. Min Act: -0.0000

Data Index: 611661 (C4 (Web Text))

Max Activating Token Index: 853

Text #8

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8346. Min Act: -0.0001

Data Index: 58390 (C4 (Web Text))

Max Activating Token Index: 266

Text #9

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8299. Min Act: -0.0001

Data Index: 103713 (C4 (Web Text))

Max Activating Token Index: 769

Text #10

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8187. Min Act: -0.0000

Data Index: 115503 (C4 (Web Text))

Max Activating Token Index: 1013

Text #11

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8231. Min Act: -0.0000

Data Index: 591623 (C4 (Web Text))

Max Activating Token Index: 284

Text #12

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8411. Min Act: -0.0000

Data Index: 75867 (C4 (Web Text))

Max Activating Token Index: 816

Text #13

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8224. Min Act: -0.0001

Data Index: 1119509 (C4 (Web Text))

Max Activating Token Index: 534

Text #14

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8176. Min Act: -0.0000

Data Index: 876140 (C4 (Web Text))

Max Activating Token Index: 335

Text #15

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.7876. Min Act: -0.0000

Data Index: 783600 (C4 (Web Text))

Max Activating Token Index: 581

Text #16

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8366. Min Act: -0.0001

Data Index: 1651305 (Python Code)

Max Activating Token Index: 983

Text #17

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.8048. Min Act: -0.0001

Data Index: 815084 (C4 (Web Text))

Max Activating Token Index: 310

Text #18

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.7960. Min Act: -0.0001

Data Index: 1649364 (Python Code)

Max Activating Token Index: 157

Text #19

Max Range: 1.0404. Min Range: -1.0404

Max Act: 0.7861. Min Act: -0.0000

Data Index: 407862 (C4 (Web Text))

Max Activating Token Index: 470