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

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

Neuron 2558 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: 0.0383. Min Range: -0.0383

Max Act: 0.0345. Min Act: -0.0001

Data Index: 1585342 (Python Code)

Max Activating Token Index: 979

Text #1

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0383. Min Act: -0.0001

Data Index: 865924 (C4 (Web Text))

Max Activating Token Index: 947

Text #2

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0365. Min Act: -0.0001

Data Index: 574370 (C4 (Web Text))

Max Activating Token Index: 959

Text #3

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0305. Min Act: -0.0001

Data Index: 24678 (C4 (Web Text))

Max Activating Token Index: 722

Text #4

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0357. Min Act: -0.0001

Data Index: 1154561 (C4 (Web Text))

Max Activating Token Index: 778

Text #5

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0283. Min Act: -0.0001

Data Index: 595996 (C4 (Web Text))

Max Activating Token Index: 962

Text #6

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0316. Min Act: -0.0001

Data Index: 1569125 (Python Code)

Max Activating Token Index: 997

Text #7

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0299. Min Act: -0.0001

Data Index: 820597 (C4 (Web Text))

Max Activating Token Index: 906

Text #8

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0315. Min Act: -0.0001

Data Index: 663697 (C4 (Web Text))

Max Activating Token Index: 810

Text #9

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0313. Min Act: -0.0001

Data Index: 6941 (C4 (Web Text))

Max Activating Token Index: 578

Text #10

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0304. Min Act: -0.0001

Data Index: 921180 (C4 (Web Text))

Max Activating Token Index: 3

Text #11

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0332. Min Act: -0.0001

Data Index: 1291862 (C4 (Web Text))

Max Activating Token Index: 648

Text #12

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0383. Min Act: -0.0001

Data Index: 678720 (C4 (Web Text))

Max Activating Token Index: 100

Text #13

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0298. Min Act: -0.0001

Data Index: 41775 (C4 (Web Text))

Max Activating Token Index: 415

Text #14

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0288. Min Act: -0.0001

Data Index: 1541512 (Python Code)

Max Activating Token Index: 160

Text #15

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0261. Min Act: -0.0001

Data Index: 145284 (C4 (Web Text))

Max Activating Token Index: 879

Text #16

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0329. Min Act: -0.0001

Data Index: 1449358 (Python Code)

Max Activating Token Index: 285

Text #17

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0294. Min Act: -0.0001

Data Index: 970586 (C4 (Web Text))

Max Activating Token Index: 911

Text #18

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0294. Min Act: -0.0001

Data Index: 939417 (C4 (Web Text))

Max Activating Token Index: 870

Text #19

Max Range: 0.0383. Min Range: -0.0383

Max Act: 0.0280. Min Act: -0.0001

Data Index: 924550 (C4 (Web Text))

Max Activating Token Index: 885