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

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

Neuron 1271 in Layer 5

Load this data into an Interactive Neuroscope

See Documentation here

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

Text #0

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1641. Min Act: -0.0001

Data Index: 424634 (C4 (Web Text))

Max Activating Token Index: 289

Text #1

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1658. Min Act: -0.0001

Data Index: 39619 (C4 (Web Text))

Max Activating Token Index: 978

Text #2

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1532. Min Act: -0.0001

Data Index: 359190 (C4 (Web Text))

Max Activating Token Index: 114

Text #3

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1545. Min Act: -0.0001

Data Index: 95059 (C4 (Web Text))

Max Activating Token Index: 920

Text #4

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1421. Min Act: -0.0001

Data Index: 624112 (C4 (Web Text))

Max Activating Token Index: 476

Text #5

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1600. Min Act: -0.0001

Data Index: 416832 (C4 (Web Text))

Max Activating Token Index: 499

Text #6

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1422. Min Act: -0.0001

Data Index: 1323336 (C4 (Web Text))

Max Activating Token Index: 448

Text #7

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1371. Min Act: -0.0001

Data Index: 260508 (C4 (Web Text))

Max Activating Token Index: 171

Text #8

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1389. Min Act: -0.0001

Data Index: 1365880 (Python Code)

Max Activating Token Index: 448

Text #9

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1244. Min Act: -0.0001

Data Index: 1022674 (C4 (Web Text))

Max Activating Token Index: 569

Text #10

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1425. Min Act: -0.0001

Data Index: 715828 (C4 (Web Text))

Max Activating Token Index: 277

Text #11

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1309. Min Act: -0.0001

Data Index: 1372236 (Python Code)

Max Activating Token Index: 137

Text #12

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1320. Min Act: -0.0001

Data Index: 1573954 (Python Code)

Max Activating Token Index: 575

Text #13

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1383. Min Act: -0.0001

Data Index: 1646124 (Python Code)

Max Activating Token Index: 125

Text #14

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1383. Min Act: -0.0001

Data Index: 1033240 (C4 (Web Text))

Max Activating Token Index: 957

Text #15

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1276. Min Act: -0.0001

Data Index: 1283847 (C4 (Web Text))

Max Activating Token Index: 826

Text #16

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1301. Min Act: -0.0001

Data Index: 1596342 (Python Code)

Max Activating Token Index: 533

Text #17

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1325. Min Act: -0.0001

Data Index: 342419 (C4 (Web Text))

Max Activating Token Index: 445

Text #18

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1317. Min Act: -0.0001

Data Index: 1606355 (Python Code)

Max Activating Token Index: 132

Text #19

Max Range: 0.1658. Min Range: -0.1658

Max Act: 0.1261. Min Act: -0.0001

Data Index: 1224267 (C4 (Web Text))

Max Activating Token Index: 449