Model: SoLU Model: 3 Layers, 2048 Neurons per Layer

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

Neuron 283 in Layer 2

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-3l')

Text #0

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0356. Min Act: -0.0003

Data Index: 483989 (C4 (Web Text))

Max Activating Token Index: 643

Text #1

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0350. Min Act: -0.0003

Data Index: 183020 (C4 (Web Text))

Max Activating Token Index: 726

Text #2

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0363. Min Act: -0.0002

Data Index: 1346413 (C4 (Web Text))

Max Activating Token Index: 114

Text #3

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0357. Min Act: -0.0003

Data Index: 7816 (C4 (Web Text))

Max Activating Token Index: 264

Text #4

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0354. Min Act: -0.0003

Data Index: 377282 (C4 (Web Text))

Max Activating Token Index: 345

Text #5

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0351. Min Act: -0.0003

Data Index: 1304658 (C4 (Web Text))

Max Activating Token Index: 70

Text #6

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0356. Min Act: -0.0004

Data Index: 593742 (C4 (Web Text))

Max Activating Token Index: 710

Text #7

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0346. Min Act: -0.0003

Data Index: 1237681 (C4 (Web Text))

Max Activating Token Index: 63

Text #8

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0351. Min Act: -0.0003

Data Index: 686671 (C4 (Web Text))

Max Activating Token Index: 397

Text #9

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0344. Min Act: -0.0003

Data Index: 1220511 (C4 (Web Text))

Max Activating Token Index: 531

Text #10

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0334. Min Act: -0.0003

Data Index: 1566254 (Python Code)

Max Activating Token Index: 952

Text #11

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0316. Min Act: -0.0003

Data Index: 493702 (C4 (Web Text))

Max Activating Token Index: 569

Text #12

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0303. Min Act: -0.0004

Data Index: 1438699 (Python Code)

Max Activating Token Index: 997

Text #13

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0326. Min Act: -0.0003

Data Index: 1655278 (Python Code)

Max Activating Token Index: 925

Text #14

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0321. Min Act: -0.0003

Data Index: 463100 (C4 (Web Text))

Max Activating Token Index: 690

Text #15

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0320. Min Act: -0.0002

Data Index: 728665 (C4 (Web Text))

Max Activating Token Index: 505

Text #16

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0331. Min Act: -0.0003

Data Index: 543427 (C4 (Web Text))

Max Activating Token Index: 454

Text #17

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0326. Min Act: -0.0003

Data Index: 404941 (C4 (Web Text))

Max Activating Token Index: 41

Text #18

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0326. Min Act: -0.0003

Data Index: 199572 (C4 (Web Text))

Max Activating Token Index: 133

Text #19

Max Range: 0.0363. Min Range: -0.0363

Max Act: 0.0316. Min Act: -0.0002

Data Index: 250861 (C4 (Web Text))

Max Activating Token Index: 601