Model: SoLU Model: 12 Layers, 6144 Neurons per Layer

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

Neuron 3841 in Layer 9

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-12l')

Text #0

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.4810. Min Act: -0.0000

Data Index: 1109758 (C4 (Web Text))

Max Activating Token Index: 84

Text #1

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.4311. Min Act: -0.0000

Data Index: 903146 (C4 (Web Text))

Max Activating Token Index: 521

Text #2

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.4086. Min Act: -0.0000

Data Index: 1390243 (Python Code)

Max Activating Token Index: 20

Text #3

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3842. Min Act: -0.0000

Data Index: 763518 (C4 (Web Text))

Max Activating Token Index: 420

Text #4

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3768. Min Act: -0.0000

Data Index: 181079 (C4 (Web Text))

Max Activating Token Index: 618

Text #5

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3743. Min Act: -0.0000

Data Index: 382562 (C4 (Web Text))

Max Activating Token Index: 275

Text #6

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3599. Min Act: -0.0000

Data Index: 281695 (C4 (Web Text))

Max Activating Token Index: 725

Text #7

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3758. Min Act: -0.0000

Data Index: 1029457 (C4 (Web Text))

Max Activating Token Index: 354

Text #8

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3586. Min Act: -0.0000

Data Index: 1652461 (Python Code)

Max Activating Token Index: 90

Text #9

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3465. Min Act: -0.0000

Data Index: 1193123 (C4 (Web Text))

Max Activating Token Index: 285

Text #10

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3650. Min Act: -0.0000

Data Index: 1285892 (C4 (Web Text))

Max Activating Token Index: 190

Text #11

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3379. Min Act: -0.0000

Data Index: 1159270 (C4 (Web Text))

Max Activating Token Index: 239

Text #12

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3371. Min Act: -0.0000

Data Index: 401349 (C4 (Web Text))

Max Activating Token Index: 585

Text #13

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3411. Min Act: -0.0000

Data Index: 789061 (C4 (Web Text))

Max Activating Token Index: 519

Text #14

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3390. Min Act: -0.0000

Data Index: 433128 (C4 (Web Text))

Max Activating Token Index: 721

Text #15

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3593. Min Act: -0.0000

Data Index: 1621488 (Python Code)

Max Activating Token Index: 207

Text #16

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3405. Min Act: -0.0000

Data Index: 1429156 (Python Code)

Max Activating Token Index: 151

Text #17

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.3124. Min Act: -0.0000

Data Index: 1300147 (C4 (Web Text))

Max Activating Token Index: 916

Text #18

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.2898. Min Act: -0.0000

Data Index: 157761 (C4 (Web Text))

Max Activating Token Index: 217

Text #19

Max Range: 0.4810. Min Range: -0.4810

Max Act: 0.2975. Min Act: -0.0000

Data Index: 1551718 (Python Code)

Max Activating Token Index: 569