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

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

Neuron 5560 in Layer 3

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.9909. Min Act: -0.0001

Data Index: 31054 (C4 (Web Text))

Max Activating Token Index: 290

Text #1

Max Range: 1.0535. Min Range: -1.0535

Max Act: 1.0535. Min Act: -0.0001

Data Index: 1552382 (Python Code)

Max Activating Token Index: 964

Text #2

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7160. Min Act: -0.0001

Data Index: 236213 (C4 (Web Text))

Max Activating Token Index: 739

Text #3

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7216. Min Act: -0.0001

Data Index: 634507 (C4 (Web Text))

Max Activating Token Index: 408

Text #4

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.8311. Min Act: -0.0001

Data Index: 93484 (C4 (Web Text))

Max Activating Token Index: 433

Text #5

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.8076. Min Act: -0.0001

Data Index: 1252065 (C4 (Web Text))

Max Activating Token Index: 701

Text #6

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.9048. Min Act: -0.0001

Data Index: 1374526 (Python Code)

Max Activating Token Index: 178

Text #7

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7607. Min Act: -0.0001

Data Index: 125474 (C4 (Web Text))

Max Activating Token Index: 885

Text #8

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7443. Min Act: -0.0001

Data Index: 1428238 (Python Code)

Max Activating Token Index: 709

Text #9

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7973. Min Act: -0.0001

Data Index: 657537 (C4 (Web Text))

Max Activating Token Index: 658

Text #10

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7680. Min Act: -0.0001

Data Index: 973566 (C4 (Web Text))

Max Activating Token Index: 756

Text #11

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7189. Min Act: -0.0001

Data Index: 1461601 (Python Code)

Max Activating Token Index: 7

Text #12

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7097. Min Act: -0.0001

Data Index: 109478 (C4 (Web Text))

Max Activating Token Index: 913

Text #13

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7301. Min Act: -0.0001

Data Index: 1160518 (C4 (Web Text))

Max Activating Token Index: 735

Text #14

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.8346. Min Act: -0.0001

Data Index: 1316521 (C4 (Web Text))

Max Activating Token Index: 25

Text #15

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7078. Min Act: -0.0001

Data Index: 1436847 (Python Code)

Max Activating Token Index: 140

Text #16

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7468. Min Act: -0.0001

Data Index: 305348 (C4 (Web Text))

Max Activating Token Index: 656

Text #17

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.7220. Min Act: -0.0001

Data Index: 1476942 (Python Code)

Max Activating Token Index: 128

Text #18

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.6319. Min Act: -0.0001

Data Index: 964048 (C4 (Web Text))

Max Activating Token Index: 996

Text #19

Max Range: 1.0535. Min Range: -1.0535

Max Act: 0.6061. Min Act: -0.0001

Data Index: 588794 (C4 (Web Text))

Max Activating Token Index: 80