Model: SoLU Model: 6 Layers, 3072 Neurons per Layer

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

Neuron 1400 in Layer 4

Load this data into an Interactive Neuroscope

See Documentation here

Transformer Lens Loading: HookedTransformer.from_pretrained('solu-6l')

Text #0

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1522. Min Act: -0.0001

Data Index: 345201 (C4 (Web Text))

Max Activating Token Index: 143

Text #1

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1529. Min Act: -0.0001

Data Index: 998460 (C4 (Web Text))

Max Activating Token Index: 89

Text #2

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1520. Min Act: -0.0001

Data Index: 932179 (C4 (Web Text))

Max Activating Token Index: 43

Text #3

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1541. Min Act: -0.0001

Data Index: 1376889 (Python Code)

Max Activating Token Index: 771

Text #4

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1437. Min Act: -0.0001

Data Index: 199768 (C4 (Web Text))

Max Activating Token Index: 577

Text #5

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1524. Min Act: -0.0001

Data Index: 742337 (C4 (Web Text))

Max Activating Token Index: 917

Text #6

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1419. Min Act: -0.0001

Data Index: 448989 (C4 (Web Text))

Max Activating Token Index: 60

Text #7

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1354. Min Act: -0.0001

Data Index: 1253597 (C4 (Web Text))

Max Activating Token Index: 490

Text #8

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1459. Min Act: -0.0001

Data Index: 892917 (C4 (Web Text))

Max Activating Token Index: 164

Text #9

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1343. Min Act: -0.0001

Data Index: 548860 (C4 (Web Text))

Max Activating Token Index: 754

Text #10

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1439. Min Act: -0.0001

Data Index: 668107 (C4 (Web Text))

Max Activating Token Index: 966

Text #11

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1244. Min Act: -0.0001

Data Index: 317424 (C4 (Web Text))

Max Activating Token Index: 615

Text #12

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1343. Min Act: -0.0001

Data Index: 726170 (C4 (Web Text))

Max Activating Token Index: 593

Text #13

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1433. Min Act: -0.0001

Data Index: 414247 (C4 (Web Text))

Max Activating Token Index: 173

Text #14

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1277. Min Act: -0.0001

Data Index: 457522 (C4 (Web Text))

Max Activating Token Index: 304

Text #15

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1373. Min Act: -0.0001

Data Index: 1450788 (Python Code)

Max Activating Token Index: 849

Text #16

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1323. Min Act: -0.0001

Data Index: 151411 (C4 (Web Text))

Max Activating Token Index: 646

Text #17

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1355. Min Act: -0.0001

Data Index: 1304742 (C4 (Web Text))

Max Activating Token Index: 771

Text #18

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1308. Min Act: -0.0001

Data Index: 324370 (C4 (Web Text))

Max Activating Token Index: 667

Text #19

Max Range: 0.1541. Min Range: -0.1541

Max Act: 0.1462. Min Act: -0.0001

Data Index: 1188008 (C4 (Web Text))

Max Activating Token Index: 259